Select different targets for building in CI on Alpine

The Alpine repository for the PowerPC target got more and more unreliable recently. Additionally, the PowerPC target is covered with the cross-build rules on Ubuntu now. Drop some (expensive) targets to reduce the overlap in coverage with the workflow that cross-builds on Ubuntu. Add LoongArch64 target which is not (yet?) available for Ubuntu. Keep targets that can be build natively (which should be reasonably fast). Use an Ubuntu on ARM runner for the armv7 target. Rename workflow because out of the three remaining targets only one is actually emulated - the others can execute natively.
Add CI workflow that cross-builds for several targets on Ubuntu
2026-08-16 14:01:08 +02:00 · 2026-08-16 14:01:08 +02:00 · 2026-08-01 15:59:19 +02:00 · 2026-07-26 18:28:41 +02:00 · 2026-07-25 17:22:12 +02:00 · 2026-07-24 19:48:01 +02:00
105 changed files with 1755 additions and 703 deletions
@@ -0,0 +1,10 @@
+# Set update schedule for GitHub Actions
+
+version: 2
+updates:
+
+  - package-ecosystem: "github-actions"
+    directory: "/"
+    schedule:
+      # Check for updates to GitHub Actions every week
+      interval: "weekly"
@@ -0,0 +1,128 @@
+name: build-alpine
+on:
+  workflow_dispatch:
+  push:
+  pull_request:
+
+concurrency: ci-arch-emu-${{ github.ref }}
+
+jobs:
+
+  alpine:
+    # Run armv7 on aarch64 runners
+    runs-on: ${{ matrix.arch == 'armv7' && 'ubuntu-24.04-arm' || 'ubuntu-latest' }}
+
+    defaults:
+      run:
+        # Use emulated shell as default
+        shell: alpine.sh {0}
+
+    strategy:
+      # Allow other runners in the matrix to continue if some fail
+      fail-fast: false
+
+      matrix:
+        # For available CPU architectures, see:
+        # https://github.com/marketplace/actions/setup-alpine-linux-environment
+        arch: [x86, armv7, loongarch64]
+
+    name: alpine (${{ matrix.arch }})
+
+    steps:
+      - name: get CPU information (host)
+        shell: bash
+        run: lscpu
+
+      - name: checkout repository
+        uses: actions/checkout@v7
+
+      - name: install dependencies
+        uses: jirutka/setup-alpine@v1
+        with:
+          arch: ${{ matrix.arch }}
+          apk-tools-url: ${{ matrix.arch == 'armv7'
+              && 'https://gitlab.alpinelinux.org/api/v4/projects/5/packages/generic/v2.14.7/aarch64/apk.static#!sha256!27a975638ddc95a411c9f17c63383e335da9edf6bb7de2281d950c291a11f878'
+              || 'https://gitlab.alpinelinux.org/api/v4/projects/5/packages/generic/v2.14.7/x86_64/apk.static#!sha256!bdd044e0fd6cc388c5e571e1093efa5f35f7767cc5aa338b0a2576a429009a62' }}
+          packages: >
+            bash
+            build-base
+            ccache
+            cmake
+            gfortran
+            eigen-dev
+            lapack-dev
+            lscpu
+
+      - name: disable QEMU emulation
+        if: matrix.arch == 'armv7'
+        shell: bash
+        run: sudo update-binfmts --disable qemu-arm
+
+      - name: get CPU information (emulated)
+        run: lscpu
+
+      - name: prepare ccache
+        # create key with human readable timestamp
+        # used in action/cache/restore and action/cache/save steps
+        id: ccache-prepare
+        run: |
+          echo "key=ccache:alpine:${{ matrix.arch }}:${{ github.ref }}:$(date +"%Y-%m-%d_%H-%M-%S"):${{ github.sha }}" >> $GITHUB_OUTPUT
+
+      - name: restore ccache
+        # setup the GitHub cache used to maintain the ccache from one job to the next
+        uses: actions/cache/restore@v6
+        with:
+          # location of the ccache of the chroot in the root file system
+          path: /home/runner/rootfs/alpine-latest-${{ matrix.arch }}/home/runner/.ccache
+          key: ${{ steps.ccache-prepare.outputs.key }}
+          # Prefer caches from the same branch. Fall back to caches from the default branch.
+          restore-keys: |
+            ccache:alpine:${{ matrix.arch }}:${{ github.ref }}
+            ccache:alpine:${{ matrix.arch }}
+
+      - name: configure ccache
+        run: |
+          test -d ~/.ccache || mkdir ~/.ccache
+          echo "max_size = 20M" >> ~/.ccache/ccache.conf
+          echo "compression = true" >> ~/.ccache/ccache.conf
+          ccache -s
+          which ccache
+
+      - name: configure
+        run: |
+          echo "gcc --version"
+          gcc --version
+          echo "gcc -dumpmachine"
+          gcc -dumpmachine
+          echo " "
+          mkdir -p ${GITHUB_WORKSPACE}/build && cd ${GITHUB_WORKSPACE}/build
+          cmake \
+            -DEXAMPLES=ON \
+            -DMPI=OFF \
+            -DICB=ON \
+            -DEIGEN=ON \
+            -DCMAKE_C_COMPILER_LAUNCHER="ccache" \
+            -DCMAKE_CXX_COMPILER_LAUNCHER="ccache" \
+            ..
+
+      - name: build
+        run: |
+          cd ${GITHUB_WORKSPACE}/build
+          cmake --build .
+
+      - name: ccache status
+        continue-on-error: true
+        run: ccache -s
+
+      - name: save ccache
+        # Save the cache after we are done (successfully) building
+        # This helps to retain the ccache even if the subsequent steps are failing.
+        uses: actions/cache/save@v6
+        with:
+          path: /home/runner/rootfs/alpine-latest-${{ matrix.arch }}/home/runner/.ccache
+          key: ${{ steps.ccache-prepare.outputs.key }}
+
+      - name: test
+        run: |
+          cd ${GITHUB_WORKSPACE}/build
+          CTEST_OUTPUT_ON_FAILURE=1 ctest .
@@ -0,0 +1,219 @@
+name: cross-build-ubuntu
+on:
+  workflow_dispatch:
+  push:
+  pull_request:
+
+concurrency: ci-cross-build-ubuntu-${{ github.ref }}
+
+jobs:
+
+  cross-build:
+    runs-on: ${{ matrix.os }}
+
+    strategy:
+      # Allow other runners in the matrix to continue if some fail
+      fail-fast: false
+
+      matrix:
+        # List of architectures that are supported by Ubuntu:
+        # https://canonical-ubuntu-packaging-guide.readthedocs-hosted.com/en/latest/reference/architectures/
+        arch: [armhf, ppc64el, s390x, riscv64]
+        include:
+          - arch: armhf
+            target-triple: arm-linux-gnueabihf
+            os: ubuntu-24.04-arm
+            ccache-max: 1G
+            qemu-user:
+          - arch: ppc64el
+            target-triple: powerpc64le-linux-gnu
+            os: ubuntu-24.04-arm
+            ccache-max: 1G
+            qemu-user: qemu-ppc64le
+          - arch: s390x
+            target-triple: s390x-linux-gnu
+            os: ubuntu-24.04
+            ccache-max: 1G
+            qemu-user: qemu-s390x
+          - arch: riscv64
+            target-triple: riscv64-linux-gnu
+            os: ubuntu-24.04-arm
+            ccache-max: 1.4G
+            qemu-user: qemu-riscv64
+
+    name: cross ${{ matrix.arch }}
+
+    env:
+      CC: ${{ matrix.target-triple }}-gcc
+      CXX: ${{ matrix.target-triple }}-g++
+      FC: ${{ matrix.target-triple }}-gfortran
+
+    steps:
+      - name: get CPU information
+        run: lscpu
+
+      - name: checkout repository
+        uses: actions/checkout@v7
+
+      - name: install toolchain for ${{ matrix.arch }}
+        run: |
+          sudo apt -qq update
+          sudo apt install -y \
+            g++-${{ matrix.target-triple }} \
+            gfortran-${{ matrix.target-triple }} \
+            cmake \
+            ccache \
+            ${{ matrix.arch == 'armhf' && ' ' || 'qemu-user-binfmt' }}
+
+      - name: add repositories for ${{ matrix.arch }} packages
+        # deb822-style format:
+        # https://manpages.ubuntu.com/manpages/noble/man5/sources.list.5.html
+        run: |
+          sudo dpkg --add-architecture ${{ matrix.arch }}
+          case ${{ matrix.os }} in
+            "ubuntu-24.04-arm")
+              sudo bash -c 'cat - >/etc/apt/sources.list.d/ubuntu.sources' <<-EOF
+          Types: deb
+          URIs: http://ports.ubuntu.com/ubuntu-ports/
+          Suites: noble noble-updates noble-backports
+          Components: main restricted universe multiverse
+          Architectures: arm64 ${{ matrix.arch }}
+          Signed-By: /usr/share/keyrings/ubuntu-archive-keyring.gpg
+          
+          Types: deb
+          URIs: http://security.ports.ubuntu.com/ubuntu-ports/
+          Suites: noble-security
+          Components: main restricted universe multiverse
+          Architectures: arm64 ${{ matrix.arch }}
+          Signed-By: /usr/share/keyrings/ubuntu-archive-keyring.gpg
+          
+          EOF
+              ;;
+            "ubuntu-24.04")
+              sudo bash -c 'cat - >/etc/apt/sources.list.d/ubuntu.sources' <<-EOF
+          Types: deb
+          URIs: http://archive.ubuntu.com/ubuntu/
+          Suites: noble noble-updates noble-backports
+          Components: main restricted universe multiverse
+          Architectures: amd64
+          Signed-By: /usr/share/keyrings/ubuntu-archive-keyring.gpg
+          
+          Types: deb
+          URIs: http://security.ubuntu.com/ubuntu/
+          Suites: noble-security
+          Components: main restricted universe multiverse
+          Architectures: amd64
+          Signed-By: /usr/share/keyrings/ubuntu-archive-keyring.gpg
+          
+          Types: deb
+          URIs: http://ports.ubuntu.com/ubuntu-ports/
+          Suites: noble noble-updates noble-backports
+          Components: main restricted universe multiverse
+          Architectures: ${{ matrix.arch }}
+          Signed-By: /usr/share/keyrings/ubuntu-archive-keyring.gpg
+          
+          Types: deb
+          URIs: http://security.ports.ubuntu.com/ubuntu-ports/
+          Suites: noble-security
+          Components: main restricted universe multiverse
+          Architectures: ${{ matrix.arch }}
+          Signed-By: /usr/share/keyrings/ubuntu-archive-keyring.gpg
+          
+          EOF
+              ;;
+          esac
+          sudo apt update
+
+      - name: install dependencies for ${{ matrix.arch }}
+        run: |
+          sudo apt install -y \
+            openmpi-bin:${{ matrix.arch }} \
+            libblas-dev:${{ matrix.arch }} \
+            liblapack-dev:${{ matrix.arch }} \
+            libeigen3-dev:${{ matrix.arch }}
+          # Trying to install libopenmpi-dev for the cross-environment causes
+          # apt to complain about missing .mod file dependencies.  Those .mod
+          # files are already installed with the cross-compiler.  Ignore those
+          # bogus dependency errors and install the package manually.
+          apt-get download libopenmpi-dev:${{ matrix.arch }}
+          sudo dpkg -i --force-depends libopenmpi-dev_*_${{ matrix.arch }}.deb
+
+      - name: prepare ccache
+        # create key with human readable timestamp
+        # used in action/cache/restore and action/cache/save steps
+        id: ccache-prepare
+        run: |
+          echo "key=ccache:cross:${{ matrix.os }}:${{ matrix.arch }}:${{ github.ref }}:$(date +"%Y-%m-%d_%H-%M-%S"):${{ github.sha }}" >> $GITHUB_OUTPUT
+
+      - name: restore ccache
+        # setup the github cache used to maintain the ccache from one job to the next
+        uses: actions/cache/restore@v6
+        with:
+          path: ~/.ccache
+          key: ${{ steps.ccache-prepare.outputs.key }}
+          restore-keys: |
+            ccache:cross:${{ matrix.os }}:${{ matrix.arch }}:${{ github.ref }}
+            ccache:cross:${{ matrix.os }}:${{ matrix.arch }}:refs/heads/default
+
+      - name: configure ccache
+        env:
+          CCACHE_MAX: ${{ matrix.ccache-max }}
+        run: |
+          test -d ~/.ccache || mkdir ~/.ccache
+          echo "max_size = $CCACHE_MAX" >> ~/.ccache/ccache.conf
+          echo "compression = true" >> ~/.ccache/ccache.conf
+          ccache -s
+          echo "/usr/lib/ccache" >> $GITHUB_PATH
+
+      - name: configure
+        run: |
+          echo $PATH
+          echo which ccache
+          which ccache
+          echo which $CC
+          which $CC
+          echo $CC --version
+          $CC --version
+          echo which $CXX
+          which $CXX
+          echo $CXX --version
+          $CXX --version
+          echo which $FC
+          which $FC
+          echo $FC --version
+          $FC --version
+          mkdir -p ${GITHUB_WORKSPACE}/build && cd ${GITHUB_WORKSPACE}/build
+          echo "::group::Configure with CMake"
+          cmake \
+            -DEXAMPLES=ON \
+            -DMPI=OFF \
+            -DICB=ON \
+            -DEIGEN=ON \
+            -DCMAKE_C_COMPILER_LAUNCHER="ccache" \
+            -DCMAKE_CXX_COMPILER_LAUNCHER="ccache" \
+            -DBLA_VENDOR="Generic" \
+            ..
+          echo "::endgroup::"
+
+      - name: build
+        run: |
+          cd ${GITHUB_WORKSPACE}/build
+          cmake --build .
+
+      - name: ccache status
+        continue-on-error: true
+        run: ccache -s
+
+      - name: save ccache
+        # Save the cache after we are done (successfully) building
+        uses: actions/cache/save@v6
+        with:
+          path: ~/.ccache
+          key: ${{ steps.ccache-prepare.outputs.key }}
+
+      - name: test
+        env:
+          driver: ${{ matrix.qemu-user }}
+        run: |
+          cd ${GITHUB_WORKSPACE}/build
+          CTEST_OUTPUT_ON_FAILURE=1 ctest .
@@ -5,7 +5,7 @@ jobs:
    runs-on: ubuntu-latest
    steps:
      - name: Clone and check out repository code
-        uses: actions/checkout@v2
+        uses: actions/checkout@v7
        with:
          fetch-depth: 0
          ref: ${{github.event.pull_request.head.ref}} # Branch where changes are implemented.
@@ -21,7 +21,7 @@ jobs:
        run: |
          mkdir build
          cd build
-          cmake -DEXAMPLES=ON -DMPI=ON -DICB=ON -DICBEXMM=ON ..
+          cmake -DEXAMPLES=ON -DMPI=ON -DICB=ON -DEIGEN=ON ..
          make all
          CTEST_OUTPUT_ON_FAILURE=1 make test
          make package_source
@@ -29,7 +29,7 @@ jobs:
    runs-on: ubuntu-latest
    steps:
      - name: Clone and check out repository code
-        uses: actions/checkout@v2
+        uses: actions/checkout@v7
        with:
          fetch-depth: 0
          ref: ${{github.event.pull_request.head.ref}} # Branch where changes are implemented.
@@ -40,7 +40,7 @@ jobs:
      - name: Update OS
        run: sudo apt-get update
      - name: Install apt-get dependencies
-        run: sudo apt-get install -y gfortran gcc g++ openmpi-bin libopenmpi-dev libblas-dev liblapack-dev cmake
+        run: sudo apt-get install -y gfortran gcc g++ openmpi-bin libopenmpi-dev libblas-dev liblapack-dev libeigen3-dev cmake
      - name: Run job
        run: |
          mkdir build
@@ -54,7 +54,7 @@ jobs:
    runs-on: ubuntu-latest
    steps:
      - name: Clone and check out repository code
-        uses: actions/checkout@v2
+        uses: actions/checkout@v7
        with:
          fetch-depth: 0
          ref: ${{github.event.pull_request.head.ref}} # Branch where changes are implemented.
@@ -69,7 +69,7 @@ jobs:
      - name: Run job
        run: |
          ./bootstrap
-          ./configure --enable-mpi --enable-icb --enable-icbexmm
+          ./configure --enable-mpi --enable-icb --enable-eigen
          make all
          make check
          make distcheck
@@ -77,7 +77,7 @@ jobs:
    runs-on: ubuntu-latest
    steps:
      - name: Clone and check out repository code
-        uses: actions/checkout@v2
+        uses: actions/checkout@v7
        with:
          fetch-depth: 0
          ref: ${{github.event.pull_request.head.ref}} # Branch where changes are implemented.
@@ -88,7 +88,7 @@ jobs:
      - name: Update OS
        run: sudo apt-get update
      - name: Install apt-get dependencies
-        run: sudo apt-get install -y gfortran gcc g++ openmpi-bin libopenmpi-dev libblas-dev liblapack-dev automake autoconf pkg-config libtool
+        run: sudo apt-get install -y gfortran gcc g++ openmpi-bin libopenmpi-dev libblas-dev liblapack-dev libeigen3-dev automake autoconf pkg-config libtool
      - name: Run job
        run: |
          ./bootstrap
@@ -101,7 +101,7 @@ jobs:
    runs-on: ubuntu-latest
    steps:
      - name: Clone and check out repository code
-        uses: actions/checkout@v2
+        uses: actions/checkout@v7
        with:
          fetch-depth: 0
          ref: ${{github.event.pull_request.head.ref}} # Branch where changes are implemented.
@@ -129,14 +129,14 @@ jobs:
        run: |
          mkdir build
          cd build
-          cmake -DEXAMPLES=ON -DMPI=ON -DICB=ON -DICBEXMM=ON -DPYTHON3=ON ..
+          cmake -DEXAMPLES=ON -DMPI=ON -DICB=ON -DEIGEN=ON -DPYTHON3=ON ..
          make all
          CTEST_OUTPUT_ON_FAILURE=1 make test
  ubuntu_latest_autotools_ilp64:
    runs-on: ubuntu-latest
    steps:
      - name: Clone and check out repository code
-        uses: actions/checkout@v2
+        uses: actions/checkout@v7
        with:
          fetch-depth: 0
          ref: ${{github.event.pull_request.head.ref}} # Branch where changes are implemented.
@@ -165,7 +165,7 @@ jobs:
    runs-on: macos-latest
    steps:
      - name: Clone and check out repository code
-        uses: actions/checkout@v2
+        uses: actions/checkout@v7
        with:
          fetch-depth: 0
          ref: ${{github.event.pull_request.head.ref}} # Branch where changes are implemented.
@@ -176,25 +176,28 @@ jobs:
      - name: Install brew dependencies
        run: |
          brew reinstall gcc # brings gfortran on path
-          brew install cmake mpich
+          brew install cmake open-mpi eigen veclibfort
      - name: Run job
        run: |
          mkdir -p build
          cd build
          export FC=mpif90 # Uses gfortran.
-          export FFLAGS="-ff2c -fno-second-underscore"
          export CC=mpicc # Uses clang.
          export CFLAGS="-Qunused-arguments"
          export CXX=mpic++ # Uses clang++.
          export CXXFLAGS="-Qunused-arguments"
-          LIBS="-framework Accelerate" cmake -DBLA_VENDOR=Generic -DEXAMPLES=ON -DICB=ON -DMPI=ON ..
+          export VECLIBFORT_PREFIX=$(brew --prefix veclibfort)
+          cmake \
+            -DBLAS_LIBRARIES="${VECLIBFORT_PREFIX}/lib/libvecLibFort.dylib" \
+            -DLAPACK_LIBRARIES="${VECLIBFORT_PREFIX}/lib/libvecLibFort.dylib" \
+            -DEXAMPLES=ON -DICB=ON -DEIGEN=ON -DMPI=ON ..
          make all
          CTEST_OUTPUT_ON_FAILURE=1 make test
  macos_latest_cmake_python:
    runs-on: macos-latest
    steps:
      - name: Clone and check out repository code
-        uses: actions/checkout@v2
+        uses: actions/checkout@v7
        with:
          fetch-depth: 0
          ref: ${{github.event.pull_request.head.ref}} # Branch where changes are implemented.
@@ -213,26 +216,28 @@ jobs:
          # https://github.com/actions/runner-images/issues/2322
          brew list -1 | grep python | while read formula; do brew unlink $formula; brew link --overwrite $formula; done
          brew reinstall gcc # brings gfortran on path
-          brew install cmake eigen boost-python3 python3 
-          pip3 install numpy
+          brew install cmake eigen boost-python3 python3 numpy veclibfort
      - name: Run job
        run: |
          mkdir -p build
          cd build
          export FC=gfortran
-          export FFLAGS="-ff2c -fno-second-underscore"
          export CC=clang
          export CFLAGS="-Qunused-arguments"
          export CXX=clang++
          export CXXFLAGS="-Qunused-arguments"
-          LIBS="-framework Accelerate" cmake -DBLA_VENDOR=Generic -DEXAMPLES=ON -DICB=ON -DPYTHON3=ON ..
+          export VECLIBFORT_PREFIX=$(brew --prefix veclibfort)
+          cmake \
+            -DBLAS_LIBRARIES="${VECLIBFORT_PREFIX}/lib/libvecLibFort.dylib" \
+            -DLAPACK_LIBRARIES="${VECLIBFORT_PREFIX}/lib/libvecLibFort.dylib" \
+            -DEXAMPLES=ON -DICB=ON -DEIGEN=ON -DPYTHON3=ON ..
          make all
          CTEST_OUTPUT_ON_FAILURE=1 make test
  macos_latest_autotools:
    runs-on: macos-latest
    steps:
      - name: Clone and check out repository code
-        uses: actions/checkout@v2
+        uses: actions/checkout@v7
        with:
          fetch-depth: 0
          ref: ${{github.event.pull_request.head.ref}} # Branch where changes are implemented.
@@ -243,15 +248,28 @@ jobs:
      - name: Install brew dependencies
        run: |
          brew reinstall gcc # brings gfortran on path
-          brew install autoconf automake libtool pkg-config mpich
+          brew install autoconf automake libtool pkgconfig open-mpi eigen veclibfort
      - name: Run job
        run: |
          ./bootstrap
-          LIBS="-framework Accelerate" FFLAGS="-ff2c -fno-second-underscore" FCFLAGS="-ff2c -fno-second-underscore" ./configure --enable-icb --enable-mpi
+          export VECLIBFORT_PREFIX=$(brew --prefix veclibfort)
+          BLAS_LIBS="-L${VECLIBFORT_PREFIX}/lib -lvecLibFort" \
+          LAPACK_LIBS="-L${VECLIBFORT_PREFIX}/lib -lvecLibFort" \
+            ./configure --enable-icb --enable-eigen --enable-mpi
          make all
          make check
  windows_latest_cmake:
    runs-on: windows-latest
+    name: MinGW-w64 ${{ matrix.msystem }} INTERFACE64=${{ matrix.int64 }}/MPI=${{ matrix.mpi }}
+    strategy:
+      fail-fast: false
+      matrix:
+        msystem: [UCRT64, CLANG64]
+        int64: [ON, OFF]
+        mpi: [ON, OFF]
+        exclude:
+          - int64: ON
+            mpi: ON
    defaults:
      run:
        # Use MSYS2 as default shell
@@ -261,19 +279,32 @@ jobs:
        uses: msys2/setup-msys2@v2
        with:
          update: true
-          msystem: MINGW64
+          msystem: ${{ matrix.msystem }}
          install: >-
            base-devel
            git
-            mingw-w64-x86_64-cmake
-            mingw-w64-x86_64-ninja
-            mingw-w64-x86_64-gcc-fortran
-            mingw-w64-x86_64-openblas
-            mingw-w64-x86_64-msmpi
+          pacboy: >-
+            cmake:p
+            ninja:p
+            cc:p
+            fc:p
+            eigen3:p
+      - name: Install OpenBLAS and MS-MPI from MSYS2
+        run: |
+          if [[ ${{ matrix.int64 }} != ON ]]; then
+            pacboy -S openblas:p --noconfirm
+          else
+            pacboy -S openblas64:p --noconfirm
+          fi
+          if [[ ${{ matrix.mpi }} = ON ]]; then
+            # This installs only the link library.
+            # The actual library will be installed in the next step.
+            pacboy -S msmpi:p --noconfirm
+          fi
      - name: Install MS-MPI (for mpiexec)
        uses: mpi4py/setup-mpi@v1
      - name: Clone and check out repository code
-        uses: actions/checkout@v2
+        uses: actions/checkout@v7
        with:
          fetch-depth: 0
          ref: ${{github.event.pull_request.head.ref}} # Branch where changes are implemented.
@@ -284,10 +315,45 @@ jobs:
      - name: Run job
        run: |
          mkdir -p build && cd build
-          cmake -GNinja -DICB=ON -DEXAMPLES=ON -DMPI=ON ..
+          echo "::group::Configure"
+          if [[ ${{ matrix.int64 }} == ON ]]; then
+            _blas_lib_flag="-DBLAS_LIBRARIES=openblas_64"
+          fi
+          cmake \
+            -GNinja \
+            -DICB=ON \
+            -DEIGEN=ON \
+            -DEXAMPLES=ON \
+            -DMPI=${{ matrix.mpi }} \
+            -DINTERFACE64=${{ matrix.int64 }} \
+            ${_blas_lib_flag} \
+            ..
+          echo "::endgroup::"
+          echo "::group::Build"
          cmake --build . -v
+          echo "::endgroup::"
      - name: Run tests
+        id: run-ctest
        run: |
-          export PATH="/c/Program Files/Microsoft MPI/Bin":$PATH # add mpiexec to msys2 path
+          if [[ ${{ matrix.mpi }} == ON ]]; then
+            export PATH="/c/Program Files/Microsoft MPI/Bin":$PATH # add mpiexec to msys2 path
+          fi
          cd build
-          ctest --output-on-failure
+          ctest
+      - name: Re-run tests
+        if: always() && (steps.run-ctest.outcome == 'failure')
+        timeout-minutes: 60
+        run: |
+          if [[ ${{ matrix.mpi }} == ON ]]; then
+            export PATH="/c/Program Files/Microsoft MPI/Bin":$PATH # add mpiexec to msys2 path
+          fi
+          cd build
+          echo "::group::Re-run ctest"
+          ctest --rerun-failed --output-on-failure || true
+          echo "::endgroup::"
+          echo "::group::Log from these tests"
+          [ ! -f Testing/Temporary/LastTest.log ] || cat Testing/Temporary/LastTest.log
+          echo "::endgroup::"
+          echo "::group::Content of arpackmm.run.log"
+          [ ! -f EXAMPLES/MATRIX_MARKET/arpackmm.run.log ] || cat EXAMPLES/MATRIX_MARKET/arpackmm.run.log
+          echo "::endgroup::"
@@ -17,10 +17,11 @@ libtool
 .deps/
 arpack*.pc
 parpack*.pc
-arpackSolver*.pc
 arpackdef.h
 arpackicb.h
 tstAutotoolsInstall.sh
+cmake/arpackng-config-version.cmake
+cmake/arpackng-config.cmake

 # Generated by `make`
 .dirstamp
@@ -1,4 +1,63 @@
- -- Franck Houssen <fghoussen@users.noreply.github.com> Sat, 11 Feb 2023 13:52:57 +0100
+arpack-ng - next
+
+[ Evan Biederstedt ]
+* Fix warnings for "Warning: Obsolescent feature: Old-style character length at (1)", compiled with `gfortran-16  -fpic  -g -O2 -Wall -pedantic -mtune=native`
+* Changed GitHub CI workflows for Mac OS to use 'brew install open-mpi', as `mpich` fails. 
+
+[ Theodore Chang ]
+ * CMake: Improve dependency linking, bump up minimum C++ standard version.
+
+[ Franck Houssen ]
+ * [BUG FIX] Fix arpack solver to handle real and imag shifts.
+ * [BUG FIX] Make sure the restart file of the arpack solver i written.
+ * [BUG FIX] Change arpack solver API.
+
+[ Henri Menke ]
+ * [BUG FIX] Add missing stdexcept header
+
+[ Kyle Guinn ]
+ * Build PARPACK p[sd]lamch10.f with FFLAGS from ./configure instead of forcing -O0.  Build all of PARPACK with AM_FFLAGS.
+ * [BUG FIX] Don't install pkg-config files for libraries that were not built.
+
+[ Markus Mützel ]
+ * CMake: Fix running CTests on Windows.
+ * Use GNU Fortran ABI in CI tests on macOS.
+ * Initialize machine dependent constants only once in `PDNAITR`.
+
+[ Szabolcs Horvát ]
+ * fix obsolescent character length warnings
+
+ -- TODO <TODO@users.noreply.github.com> DATE
+
+arpack-ng - 3.9.1
+
+[ Fabien Péan ]
+ * pyarpack: Ensure that the matrix properties (symmetric/hermitian) fit the solver (CG/LDL) with which they are used in the tests
+ * [BUG FIX] Tests for PARPACK with C/C++ bindings icb_parpack_c and icb_parpack_cpp are now really parallel and split the problem across MPI processes.
+ * Update arpackmm test suite: enable solving non-symmetric tests with BiCG solver
+ * README: Add details on Windows installation.
+ 
+[ Szabolcs Horvát ]
+ * [BUG FIX] Ensure that LAPACK RNG state is propagated (regression in 3.9.0).
+ * [BUG FIX] Ensure that separate random seeds are used on different parallel thread in D and S versions of functions (issue from original ARPACK).
+
+[ Ruoyu Feng ]
+ * ICB(arpackdef.h): distinct intel llvm compiler (icx with clang-cl) from msvc on windows
+ * ICB(arpackdef.h): Undef macro I if complex.h from msvc version is loaded, which is an usual name and causes issues on arpackSolver.
+
+[ Franck Houssen ]
+ * [BUG FIX] Fix install: headers in /path/to/local/include/arpack for ICB samples
+ * [BUG FIX] Fix install: headers in /path/to/local/include/arpack
+ * arpackmm: allow for using LA/SA magnitudes.
+ * Rename icbexmm option into eigen option.
+ * README: document how to use ICB.
+ * [BUG FIX] arpackmm: fix restart.
+ * README: document where to find arpack user's guide.
+
+ -- Franck Houssen <fghoussen@users.noreply.github.com> Sat, 14 Oct 2023 13:37:37 +0200
+
+[ Franck Houssen ]
+ * Add documentation on ILP64.

 arpack-ng - 3.9.0

@@ -199,7 +258,7 @@ arpack-ng - 3.5.0
 * Improve cmake build system: disable C++ detection, set default build type.

 [ Marco Atzeri]
- * Use AC_PROG_FC instead of AC_PROG_F77 for proper inizialization
+ * Use AC_PROG_FC instead of AC_PROG_F77 for proper initialization
   for the usage of AC_FC_LINE_LENGTH. Noted on Cygwin.

 [ Denis Davydov ]
@@ -1,4 +1,4 @@
-cmake_minimum_required(VERSION 3.0)
+cmake_minimum_required(VERSION 3.5)

 if (NOT DEFINED CMAKE_BUILD_TYPE)
   set (CMAKE_BUILD_TYPE Release CACHE STRING "Build type")
@@ -8,7 +8,7 @@ project(arpack C Fortran)

 set(arpack_ng_MAJOR_VERSION 3)
 set(arpack_ng_MINOR_VERSION 9)
-set(arpack_ng_PATCH_VERSION 0)
+set(arpack_ng_PATCH_VERSION 1)
 set(arpack_ng_VERSION ${arpack_ng_MAJOR_VERSION}.${arpack_ng_MINOR_VERSION}.${arpack_ng_PATCH_VERSION})

 set(CMAKE_MODULE_PATH ${CMAKE_MODULE_PATH} ${PROJECT_SOURCE_DIR}/cmake)
@@ -27,7 +27,7 @@ endif ()
 option(BUILD_SHARED_LIBS "Build shared libraries instead of static libraries" ON)
 option(MPI "Enable parallel support" OFF)
 option(ICB "Enable support for *[ae]upd_c with ISO_C_BINDING" OFF)
-option(ICBEXMM "Enable support for matrix market example based on ICB" OFF)
+option(EIGEN "Enable support for eigenvalue-problems solver based on ICB and eigen" OFF)
 option(PYTHON3 "Enable python3 support" OFF)
 set(BOOST_PYTHON_LIBSUFFIX "" CACHE STRING "suffix to add to custom boost python libs")
 option(EXAMPLES "Compile ARPACK examples" OFF)
@@ -45,7 +45,7 @@ option(INTERFACE64 "use the 64-bit integer interface (ILP64) for ARPACK, BLAS an
 # 'make install' to the correct location, and also define
 # paths for target_include_directories and target_link_libraries
 include(GNUInstallDirs)
-set(ARPACK_INSTALL_INCLUDEDIR "${CMAKE_INSTALL_INCLUDEDIR}/arpack-ng${ITF64SUFFIX}")
+set(ARPACK_INSTALL_INCLUDEDIR "${CMAKE_INSTALL_INCLUDEDIR}/arpack${ITF64SUFFIX}")
 set(ARPACK_INSTALL_CMAKEDIR "${CMAKE_INSTALL_LIBDIR}/cmake/arpackng${LIBSUFFIX}${ITF64SUFFIX}")

 set(CMAKE_LIBRARY_OUTPUT_DIRECTORY ${CMAKE_BINARY_DIR}/lib)
@@ -62,7 +62,7 @@ function(examples list_name)
    foreach(l ${${list_name}})
        get_filename_component(lwe ${l} NAME_WE)
        add_executable(${lwe} ${arpackexample_DIR}/${l} ${examples_EXTRA_SRCS})
-        target_link_libraries(${lwe} arpack BLAS::BLAS LAPACK::LAPACK ${EXTRA_LDFLAGS})
+        target_link_libraries(${lwe} arpack ${LAPACK_BLAS_LIBS} ${EXTRA_LDFLAGS})
        add_test(NAME "${lwe}_ex" COMMAND ${lwe})
    endforeach()
 endfunction(examples)
@@ -76,18 +76,21 @@ function(pexamples list_name)
    endforeach()
 endfunction(pexamples)

-if (PYTHON3)
-    enable_language(C CXX) # Boost requirement.
-    set(CMAKE_CXX_STANDARD 14) # Boost requirement.
+if (PYTHON3 OR EIGEN OR ICB)
+    enable_language(CXX)
+    set(CMAKE_CXX_STANDARD 14)
+    set(CMAKE_CXX_STANDARD_REQUIRED ON)
+endif ()

+if (PYTHON3)
    find_package(PythonInterp 3 REQUIRED)
    find_package(PythonLibs 3 REQUIRED)
    find_package(Boost 1.78 COMPONENTS python${BOOST_PYTHON_LIBSUFFIX} numpy${BOOST_PYTHON_LIBSUFFIX} REQUIRED)

-    set(ICBEXMM "ON")
+    set(EIGEN "ON")
 endif ()

-if (ICBEXMM)
+if (EIGEN)
    find_package(Eigen3 3.3 QUIET)
    if (NOT EIGEN3_FOUND) # If not found, piggy-back pkg-config files.
        message(WARNING "CMake didn't find the Eigen3 package. Try to look for pkg-config file...")
@@ -95,7 +98,6 @@ if (ICBEXMM)
        pkg_check_modules(EIGEN3 REQUIRED eigen3>=3.3)
        set(EIGEN3_INCLUDE_DIR ${EIGEN3_INCLUDE_DIRS})
    endif ()
-    find_program (BASH_PROGRAM bash)

    # Look for headers.
    find_path(EIGEN3_ITERATIVE_SOLVER_DIR NAMES IterativeLinearSolvers PATHS ${EIGEN3_INCLUDE_DIR} PATH_SUFFIXES Eigen)
@@ -140,10 +142,6 @@ endif ()

 # Enable language(s) before any find_package (in particular before MPI find_package).
 if (ICB)
-    enable_language(C CXX) # For testing binding with c/c++.
-
-    set(CMAKE_CXX_STANDARD 11) # OK, since cmake-3.1 only.
-
    file(WRITE ${CMAKE_BINARY_DIR}${CMAKE_FILES_DIRECTORY}/CMakeTmp/PROG_ICB.f90
         "
          PROGRAM PROG_ICB
@@ -165,6 +163,8 @@ else()
    if (SYMBOLSUFFIX)
        if ("${CMAKE_Fortran_COMPILER_ID}" MATCHES "GNU")
          set(CMAKE_Fortran_FLAGS "${CMAKE_Fortran_FLAGS} -cpp -ffixed-line-length-none")
+        elseif ("${CMAKE_Fortran_COMPILER_ID}" STREQUAL "LLVMFlang")
+          set(CMAKE_Fortran_FLAGS "${CMAKE_Fortran_FLAGS} -cpp -ffixed-line-length=none")
        elseif ("${CMAKE_Fortran_COMPILER_ID}" MATCHES "Intel")
          set(CMAKE_Fortran_FLAGS "${CMAKE_Fortran_FLAGS} -fpp -extend-source")
        else ()
@@ -301,6 +301,8 @@ if (NOT TARGET LAPACK::LAPACK) # Search only if not already found by upper CMake
    endif()
 endif()

+set(LAPACK_BLAS_LIBS LAPACK::LAPACK BLAS::BLAS)
+
 # As BLAS/LAPACK does not provide ICB, we may have to deal with symbols the old-fashion-boring-cumbersome-fortran/C way...

 if (SYMBOLSUFFIX)
@@ -427,7 +429,7 @@ set(PYINT "int32" CACHE STRING "int type to be used in python scripts") # PYINT
 if (INTERFACE64)
    set(PYINT "int64" CACHE STRING "int type to be used in python scripts") # PYINT : int used with python.
    set(INTERFACE64 1)
-    if ("${CMAKE_Fortran_COMPILER_ID}" MATCHES "GNU")
+    if ("${CMAKE_Fortran_COMPILER_ID}" STREQUAL "GNU" OR "${CMAKE_Fortran_COMPILER_ID}" STREQUAL "LLVMFlang")
      set(CMAKE_Fortran_FLAGS "${CMAKE_Fortran_FLAGS} -fdefault-integer-8")
    elseif ("${CMAKE_Fortran_COMPILER_ID}" MATCHES "Intel")
      set(CMAKE_Fortran_FLAGS "${CMAKE_Fortran_FLAGS} -i8")
@@ -625,12 +627,28 @@ endif()
 ############################
 # TEST
 ############################
-function(build_tests)  
+
+function(add_test_with_rpath test_name)
+  # Add test which sets a current working directory that doesn't contain the
+  # libarpack library.
+  # Windows doesn't have a mechanism similar to RPATH on Linux or macOS.
+  # On Windows, if a test is running in a working directory that doesn't contain
+  # the ARPACK (and PARPACK) libraries, the path to these libraries has to be
+  # added to the PATH environment variable.
+  add_test(NAME ${test_name} ${ARGN})
+  if (WIN32 AND BUILD_SHARED_LIBS)
+    set_tests_properties(${test_name}
+      PROPERTIES
+      ENVIRONMENT_MODIFICATION "PATH=path_list_prepend:$<TARGET_FILE_DIR:arpack>")
+  endif()
+endfunction()
+
+function(build_tests)
  set(CMAKE_RUNTIME_OUTPUT_DIRECTORY ${CMAKE_BINARY_DIR}/TESTS)

  add_executable(dnsimp_test TESTS/dnsimp.f TESTS/mmio.f TESTS/debug.h)
  set_target_properties( dnsimp_test PROPERTIES OUTPUT_NAME  dnsimp )
-  target_link_libraries(dnsimp_test arpack BLAS::BLAS LAPACK::LAPACK ${EXTRA_LDFLAGS})
+  target_link_libraries(dnsimp_test arpack ${LAPACK_BLAS_LIBS} ${EXTRA_LDFLAGS})
  add_custom_command(TARGET dnsimp_test POST_BUILD
    COMMAND ${CMAKE_COMMAND} -E copy ${CMAKE_CURRENT_SOURCE_DIR}/TESTS/testA.mtx testA.mtx
  )
@@ -639,41 +657,41 @@ function(build_tests)
  if (ICB)
      add_executable(bug_1315_single TESTS/bug_1315_single.c)
      target_include_directories(bug_1315_single PUBLIC ${PROJECT_SOURCE_DIR}/ICB) # Get arpack.h
-      target_link_libraries(bug_1315_single arpack BLAS::BLAS LAPACK::LAPACK ${EXTRA_LDFLAGS})
+      target_link_libraries(bug_1315_single arpack ${LAPACK_BLAS_LIBS} ${EXTRA_LDFLAGS})
      add_test(bug_1315_single_tst ${CMAKE_RUNTIME_OUTPUT_DIRECTORY}/bug_1315_single)

      add_executable(bug_1315_double TESTS/bug_1315_double.c)
      target_include_directories(bug_1315_double PUBLIC ${PROJECT_SOURCE_DIR}/ICB) # Get arpack.h
-      target_link_libraries(bug_1315_double arpack BLAS::BLAS LAPACK::LAPACK ${EXTRA_LDFLAGS})
+      target_link_libraries(bug_1315_double arpack ${LAPACK_BLAS_LIBS} ${EXTRA_LDFLAGS})
      add_test(bug_1315_double_tst ${CMAKE_RUNTIME_OUTPUT_DIRECTORY}/bug_1315_double)
  endif()

  add_executable(bug_1323 TESTS/bug_1323.f)
-  target_link_libraries(bug_1323 arpack BLAS::BLAS LAPACK::LAPACK ${EXTRA_LDFLAGS})
+  target_link_libraries(bug_1323 arpack ${LAPACK_BLAS_LIBS} ${EXTRA_LDFLAGS})
  add_test(bug_1323_tst ${CMAKE_RUNTIME_OUTPUT_DIRECTORY}/bug_1323)

  add_executable(bug_58_double TESTS/bug_58_double.f)
-  target_link_libraries(bug_58_double arpack BLAS::BLAS LAPACK::LAPACK ${EXTRA_LDFLAGS})
+  target_link_libraries(bug_58_double arpack ${LAPACK_BLAS_LIBS} ${EXTRA_LDFLAGS})
  add_test(bug_58_double_tst ${CMAKE_RUNTIME_OUTPUT_DIRECTORY}/bug_58_double)

  add_executable(bug_79_double_complex TESTS/bug_79_double_complex.f)
-  target_link_libraries(bug_79_double_complex arpack BLAS::BLAS LAPACK::LAPACK ${EXTRA_LDFLAGS})
+  target_link_libraries(bug_79_double_complex arpack ${LAPACK_BLAS_LIBS} ${EXTRA_LDFLAGS})
  add_test(bug_79_double_complex_tst ${CMAKE_RUNTIME_OUTPUT_DIRECTORY}/bug_79_double_complex)

  add_executable(bug_142 TESTS/bug_142.f)
-  target_link_libraries(bug_142 arpack BLAS::BLAS LAPACK::LAPACK ${EXTRA_LDFLAGS})
+  target_link_libraries(bug_142 arpack ${LAPACK_BLAS_LIBS} ${EXTRA_LDFLAGS})
  add_test(bug_142_tst ${CMAKE_RUNTIME_OUTPUT_DIRECTORY}/bug_142)

  add_executable(bug_142_gen TESTS/bug_142_gen.f)
-  target_link_libraries(bug_142_gen arpack BLAS::BLAS LAPACK::LAPACK ${EXTRA_LDFLAGS})
+  target_link_libraries(bug_142_gen arpack ${LAPACK_BLAS_LIBS} ${EXTRA_LDFLAGS})
  add_test(bug_142_gen_tst ${CMAKE_RUNTIME_OUTPUT_DIRECTORY}/bug_142_gen)

  if(MPI)
    set(CMAKE_RUNTIME_OUTPUT_DIRECTORY ${CMAKE_BINARY_DIR}/PARPACK/TESTS/MPI)

    add_executable(issue46 PARPACK/TESTS/MPI/issue46.f)
-    target_link_libraries(issue46 parpack arpack BLAS::BLAS LAPACK::LAPACK ${EXTRA_LDFLAGS})
-    add_test(issue46_tst ${CMAKE_RUNTIME_OUTPUT_DIRECTORY}/issue46)
+    target_link_libraries(issue46 parpack arpack ${LAPACK_BLAS_LIBS} ${EXTRA_LDFLAGS})
+    add_test(issue46_tst mpiexec -n 2 ${CMAKE_RUNTIME_OUTPUT_DIRECTORY}/issue46)
  endif()

  if(ICB)
@@ -681,27 +699,35 @@ function(build_tests)

    add_executable(icb_arpack_c TESTS/icb_arpack_c.c)
    target_include_directories(icb_arpack_c PUBLIC ${PROJECT_SOURCE_DIR}/ICB) # Get arpack.h
-    target_link_libraries(icb_arpack_c arpack BLAS::BLAS LAPACK::LAPACK ${EXTRA_LDFLAGS})
+    target_link_libraries(icb_arpack_c arpack ${LAPACK_BLAS_LIBS} ${EXTRA_LDFLAGS})
    add_test(icb_arpack_c_tst ${CMAKE_RUNTIME_OUTPUT_DIRECTORY}/icb_arpack_c)

    add_executable(icb_arpack_cpp TESTS/icb_arpack_cpp.cpp)
    target_include_directories(icb_arpack_cpp PUBLIC ${PROJECT_SOURCE_DIR}/ICB) # Get arpack.hpp
-    target_link_libraries(icb_arpack_cpp arpack BLAS::BLAS LAPACK::LAPACK ${EXTRA_LDFLAGS})
+    target_link_libraries(icb_arpack_cpp arpack ${LAPACK_BLAS_LIBS} ${EXTRA_LDFLAGS})
    add_test(icb_arpack_cpp_tst ${CMAKE_RUNTIME_OUTPUT_DIRECTORY}/icb_arpack_cpp)

-    if (ICBEXMM)
+    if (EIGEN)
+      find_program (BASH_PROGRAM bash)
+
      set(CMAKE_RUNTIME_OUTPUT_DIRECTORY ${CMAKE_BINARY_DIR}/EXAMPLES/MATRIX_MARKET)

      add_executable(arpackmm EXAMPLES/MATRIX_MARKET/arpackmm.cpp)
      target_include_directories(arpackmm PUBLIC ${PROJECT_SOURCE_DIR}/ICB ${EIGEN3_INCLUDE_DIR}) # Get arpack.h + eigen
-      target_link_libraries(arpackmm arpack BLAS::BLAS LAPACK::LAPACK ${EXTRA_LDFLAGS})
+      target_link_libraries(arpackmm arpack ${LAPACK_BLAS_LIBS} ${EXTRA_LDFLAGS})
      configure_file(EXAMPLES/MATRIX_MARKET/As.mtx ${CMAKE_RUNTIME_OUTPUT_DIRECTORY}/As.mtx)
      configure_file(EXAMPLES/MATRIX_MARKET/An.mtx ${CMAKE_RUNTIME_OUTPUT_DIRECTORY}/An.mtx)
      configure_file(EXAMPLES/MATRIX_MARKET/Az.mtx ${CMAKE_RUNTIME_OUTPUT_DIRECTORY}/Az.mtx)
      configure_file(EXAMPLES/MATRIX_MARKET/B.mtx  ${CMAKE_RUNTIME_OUTPUT_DIRECTORY}/B.mtx)
      configure_file(EXAMPLES/MATRIX_MARKET/Bz.mtx ${CMAKE_RUNTIME_OUTPUT_DIRECTORY}/Bz.mtx)
      configure_file(EXAMPLES/MATRIX_MARKET/arpackmm.sh ${CMAKE_RUNTIME_OUTPUT_DIRECTORY}/arpackmm.sh)
-      add_test(NAME arpackmm_tst WORKING_DIRECTORY ${CMAKE_RUNTIME_OUTPUT_DIRECTORY} COMMAND ${BASH_PROGRAM} arpackmm.sh)
+      add_test_with_rpath(arpackmm_tst WORKING_DIRECTORY ${CMAKE_RUNTIME_OUTPUT_DIRECTORY} COMMAND ${BASH_PROGRAM} arpackmm.sh)
+      configure_file(EXAMPLES/MATRIX_MARKET/issue401.mtx ${CMAKE_RUNTIME_OUTPUT_DIRECTORY}/issue401.mtx)
+      configure_file(EXAMPLES/MATRIX_MARKET/issue401.sh ${CMAKE_RUNTIME_OUTPUT_DIRECTORY}/issue401.sh)
+      add_test_with_rpath(issue401_tst WORKING_DIRECTORY ${CMAKE_RUNTIME_OUTPUT_DIRECTORY} COMMAND ${BASH_PROGRAM} issue401.sh)
+      configure_file(EXAMPLES/MATRIX_MARKET/issue215.mtx ${CMAKE_RUNTIME_OUTPUT_DIRECTORY}/issue215.mtx)
+      configure_file(EXAMPLES/MATRIX_MARKET/issue215.sh ${CMAKE_RUNTIME_OUTPUT_DIRECTORY}/issue215.sh)
+      add_test_with_rpath(issue215_tst WORKING_DIRECTORY ${CMAKE_RUNTIME_OUTPUT_DIRECTORY} COMMAND ${BASH_PROGRAM} issue215.sh)
    endif()

    if (PYTHON3)
@@ -717,7 +743,7 @@ function(build_tests)
          ${Boost_INCLUDE_DIRS}
          ${PYTHON_INCLUDE_DIRS})
      target_link_libraries(pyarpack
-          BLAS::BLAS LAPACK::LAPACK ${Boost_LIBRARIES} ${PYTHON_LIBRARIES})
+          ${LAPACK_BLAS_LIBS} ${Boost_LIBRARIES} ${PYTHON_LIBRARIES})
      install(TARGETS pyarpack
              ARCHIVE DESTINATION ${CMAKE_INSTALL_LIBDIR}/pyarpack
              LIBRARY DESTINATION ${CMAKE_INSTALL_LIBDIR}/pyarpack)
@@ -773,12 +799,12 @@ function(build_tests)

      add_executable(icb_parpack_c PARPACK/TESTS/MPI/icb_parpack_c.c)
      target_include_directories(icb_parpack_c PUBLIC ${PROJECT_SOURCE_DIR}/ICB MPI::MPI_C) # Get parpack.h mpi.h
-      target_link_libraries(icb_parpack_c parpack arpack BLAS::BLAS LAPACK::LAPACK ${EXTRA_LDFLAGS} MPI::MPI_C)
+      target_link_libraries(icb_parpack_c parpack arpack ${LAPACK_BLAS_LIBS} ${EXTRA_LDFLAGS} MPI::MPI_C)
      add_test(icb_parpack_c_tst mpiexec -n 2 ${CMAKE_RUNTIME_OUTPUT_DIRECTORY}/icb_parpack_c)

      add_executable(icb_parpack_cpp PARPACK/TESTS/MPI/icb_parpack_cpp.cpp)
      target_include_directories(icb_parpack_cpp PUBLIC ${PROJECT_SOURCE_DIR}/ICB MPI::MPI_CXX) # Get parpack.hpp mpi.h
-      target_link_libraries(icb_parpack_cpp parpack arpack BLAS::BLAS LAPACK::LAPACK ${EXTRA_LDFLAGS} MPI::MPI_CXX)
+      target_link_libraries(icb_parpack_cpp parpack arpack ${LAPACK_BLAS_LIBS} ${EXTRA_LDFLAGS} MPI::MPI_CXX)
      add_test(icb_parpack_cpp_tst mpiexec -n 2 ${CMAKE_RUNTIME_OUTPUT_DIRECTORY}/icb_parpack_cpp)
    endif()
  endif()
@@ -786,7 +812,7 @@ endfunction(build_tests)

 if(TESTS)
    enable_testing()
-    set(CMAKE_CTEST_COMMAND ctest -V)   
+    set(CMAKE_CTEST_COMMAND ctest -V)
    build_tests()
 endif()

@@ -798,7 +824,7 @@ endif()
 set(prefix ${CMAKE_INSTALL_PREFIX})
 set(exec_prefix \${prefix})
 set(libdir ${CMAKE_INSTALL_FULL_LIBDIR})
-set(includedir ${CMAKE_INSTALL_PREFIX}/${CMAKE_INSTALL_INCLUDEDIR})
+set(includedir ${CMAKE_INSTALL_FULL_INCLUDEDIR})
 set(PACKAGE_NAME ${PROJECT_NAME})
 set(PACKAGE_VERSION ${arpack_ng_VERSION})
 set(PACKAGE_URL "https://github.com/opencollab/arpack-ng/")
@@ -821,17 +847,15 @@ foreach(lib ${LAPACK_LIBRARIES} ${MPI_Fortran_LIBRARIES})
 endforeach()
 string(REPLACE ";" " " PARPACK_PC_LIBS_PRIVATE "${PARPACK_PC_LIBS_PRIVATE}")

-configure_file(pkg-config/arpack.pc.in "${PROJECT_BINARY_DIR}/arpack${LIBSUFFIX}${ITF64SUFFIX}.pc" @ONLY)
-configure_file(pkg-config/parpack.pc.in "${PROJECT_BINARY_DIR}/parpack${LIBSUFFIX}${ITF64SUFFIX}.pc" @ONLY)
-configure_file(pkg-config/arpackSolver.pc.in "${PROJECT_BINARY_DIR}/arpackSolver${LIBSUFFIX}${ITF64SUFFIX}.pc" @ONLY)
-
+configure_file(SRC/arpack.pc.in "${PROJECT_BINARY_DIR}/SRC/arpack${LIBSUFFIX}${ITF64SUFFIX}.pc" @ONLY)
+configure_file(PARPACK/SRC/MPI/parpack.pc.in "${PROJECT_BINARY_DIR}/PARPACK/SRC/MPI/parpack${LIBSUFFIX}${ITF64SUFFIX}.pc" @ONLY)

 install(TARGETS arpack
    EXPORT arpackngTargets
    ARCHIVE DESTINATION ${CMAKE_INSTALL_LIBDIR}
    LIBRARY DESTINATION ${CMAKE_INSTALL_LIBDIR}
    RUNTIME DESTINATION ${CMAKE_INSTALL_BINDIR})
-install(FILES "${PROJECT_BINARY_DIR}/arpack${LIBSUFFIX}${ITF64SUFFIX}.pc"
+install(FILES "${PROJECT_BINARY_DIR}/SRC/arpack${LIBSUFFIX}${ITF64SUFFIX}.pc"
    DESTINATION ${CMAKE_INSTALL_LIBDIR}/pkgconfig)

 if (MPI)
@@ -840,7 +864,7 @@ if (MPI)
      ARCHIVE DESTINATION ${CMAKE_INSTALL_LIBDIR}
      LIBRARY DESTINATION ${CMAKE_INSTALL_LIBDIR}
      RUNTIME DESTINATION ${CMAKE_INSTALL_BINDIR})
-  install(FILES "${PROJECT_BINARY_DIR}/parpack${LIBSUFFIX}${ITF64SUFFIX}.pc"
+  install(FILES "${PROJECT_BINARY_DIR}/PARPACK/SRC/MPI/parpack${LIBSUFFIX}${ITF64SUFFIX}.pc"
      DESTINATION ${CMAKE_INSTALL_LIBDIR}/pkgconfig)
 endif ()

@@ -851,9 +875,8 @@ if(ICB)
    install(FILES ICB/parpack.h DESTINATION ${ARPACK_INSTALL_INCLUDEDIR})
    install(FILES ICB/parpack.hpp DESTINATION ${ARPACK_INSTALL_INCLUDEDIR})
  endif()
-  if (ICBEXMM)
+  if (EIGEN)
    install(FILES EXAMPLES/MATRIX_MARKET/arpackSolver.hpp DESTINATION ${ARPACK_INSTALL_INCLUDEDIR})
-    install(FILES "${PROJECT_BINARY_DIR}/arpackSolver${LIBSUFFIX}${ITF64SUFFIX}.pc" DESTINATION ${CMAKE_INSTALL_LIBDIR}/pkgconfig)
  endif()
 endif()

@@ -874,8 +897,10 @@ endif()
 configure_file(arpackdef.h.in "${PROJECT_BINARY_DIR}/arpackdef.h" @ONLY)
 install(FILES "${PROJECT_BINARY_DIR}/arpackdef.h" DESTINATION ${ARPACK_INSTALL_INCLUDEDIR})

-configure_file(arpackicb.h.in "${PROJECT_BINARY_DIR}/arpackicb.h" @ONLY)
-install(FILES "${PROJECT_BINARY_DIR}/arpackicb.h" DESTINATION ${ARPACK_INSTALL_INCLUDEDIR})
+if(ICB)
+  configure_file(arpackicb.h.in "${PROJECT_BINARY_DIR}/arpackicb.h" @ONLY)
+  install(FILES "${PROJECT_BINARY_DIR}/arpackicb.h" DESTINATION ${ARPACK_INSTALL_INCLUDEDIR})
+endif()

 install(EXPORT arpackngTargets
  DESTINATION "${ARPACK_INSTALL_CMAKEDIR}"
@@ -959,7 +984,7 @@ if (MPI)
 endif()
 libsummary("BLAS" "" "${BLAS_LIBRARIES}")
 libsummary("LAPACK" "" "${LAPACK_LIBRARIES}")
-if (ICBEXMM)
+if (EIGEN)
    libsummary("EIGEN3" "${EIGEN3_INCLUDE_DIR}" "")
 endif()
 if (PYTHON3)
@@ -206,6 +206,7 @@ c
      idiag = kl+ku+1
      do 30 j = 1, n
         a(idiag,j) = (4.0E+0, 0.0E+0)  / h2
+         m(idiag,j) = (1.0E+0, 0.0E+0)
  30  continue
 c
 c     %-------------------------------------%
@@ -206,6 +206,7 @@ c
      idiag = kl+ku+1
      do 30 j = 1, n
         a(idiag,j) = (4.0D+0, 0.0D+0)  / h2
+         m(idiag,j) = (1.0D+0, 0.0D+0)
  30  continue
 c
 c     %-------------------------------------%
@@ -4,13 +4,14 @@ AM_CPPFLAGS = -I$(top_builddir) -I$(top_srcdir)/ICB $(EIGEN3_CFLAGS)
 EXTRA_DIST = README

 check_PROGRAMS = arpackmm
-dist_check_SCRIPTS = arpackmm.sh
-dist_check_DATA = An.mtx As.mtx Az.mtx B.mtx Bz.mtx
-TESTS = arpackmm.sh
+dist_check_SCRIPTS = arpackmm.sh issue401.sh issue215.sh
+dist_check_DATA = An.mtx As.mtx Az.mtx B.mtx Bz.mtx issue401.mtx issue215.mtx
+TESTS = arpackmm.sh issue401.sh issue215.sh

 arpackmm_SOURCES = arpackmm.cpp

-if ICBEXMM
+pkgincludedir = $(includedir)/arpack@ITF64SUFFIX@
+if EIGEN
 pkginclude_HEADERS = arpackSolver.hpp
 endif

@@ -34,6 +34,6 @@ To build this utility, you need:
 - to have installed eigen3 (to deal with the RCI).
 - to configure arpack-ng this way:
  - autotools:
-    ~arpack-ng> ./configure --enable-icb-exmm; make all check
+    ~arpack-ng> ./configure --enable-eigen; make all check
  - cmake:
-    ~arpack-ng/build> cmake -D ICBEXMM=ON ..; make all test
+    ~arpack-ng/build> cmake -D EIGEN=ON ..; make all test
@@ -12,6 +12,9 @@
 #include <cassert>
 #include <vector>
 #include <type_traits> // is_same.
+#include <cmath> // abs
+#include <complex>
+#include <limits> // epsilon

 #include <Eigen/Sparse>
 #include <Eigen/IterativeLinearSolvers>
@@ -173,6 +176,7 @@ class arpackSolver {
    int createMatrix(string const & fileName, Eigen::SparseMatrix<RC> & M) {
      // Read matrix from file.

+      if (fileName.empty()) { cerr << "Error: matrix file missing" << endl; return 1; }
      a_uint n = 0, m = 0;
      vector<a_uint> i, j;
      vector<RC> Mij;
@@ -229,6 +233,7 @@ class arpackSolver {

    int solve(EM & A, EM const * B = nullptr) {
      stdPb = !B ? true : false;
+      if (nbCV > A.cols()) nbCV = A.cols(); // Cut-off arpack workspace dim.

      dumpParameters();
      if (verbose == 3) {
@@ -244,14 +249,15 @@ class arpackSolver {

      // If needed, transform the initial problem into a new one that arpack can handle.

-      auto eps = numeric_limits<double>::epsilon();
+      auto eps = numeric_limits<FD>::epsilon();
      bool shiftReal = (fabs(sigmaReal) > eps) ? true : false;
      bool shiftImag = (fabs(sigmaImag) > eps) ? true : false;
      bool backTransform = false;
      mode = 0;
      if (stdPb) {
        mode = 1;
-        if (shiftReal && !shiftImag) {
+        // CAUTION: back transform must be done only if mode = 1.
+        if (shiftReal || shiftImag) {
          EM I(A.rows(), A.cols());
          I.setIdentity();
          RC sigma; makeSigma(sigma);
@@ -260,6 +266,7 @@ class arpackSolver {
        }
      }
      else {
+        // CAUTION: back transform must NOT be done if mode > 1.
        mode = 2;
        if (shiftReal || shiftImag) mode = 3;
      }
@@ -290,10 +297,7 @@ class arpackSolver {
      return 0;
    };

-    int checkEigVec(EM const & A, EM const * B = nullptr, double const * diffTol = nullptr) {
-      stdPb = !B ? true : false;
-      double dTol = !diffTol ? sqrt(tol) : *diffTol;
-
+    int checkEigVec(EM const & A, EM const * B = nullptr, double const maxResNorm = 1.e-3) const {
      // Check eigen vectors.

      string rs = schur ? "Schur" : "Ritz";
@@ -315,26 +319,18 @@ class arpackSolver {
          }
        }

-        EigVecZ left = A.template cast<complex<double>>() * V;
-        EigVecZ right = stdPb ? V : B->template cast<complex<double>>() * V;
-        right *= lambda;
-        EigVecZ diff = left - right;
-        if (diff.norm() > dTol) {
-          cerr << endl << "Error: bad vector " << setw(3) << i << " (norm " << V.norm() << "):" << endl;
-          cerr << endl << V << endl;
-          cerr << endl << "Error: left side (A*V - norm " << left.norm() << "):" << endl;
-          cerr << endl << left << endl;
-          cerr << endl << "Error: right side (lambda*" << (stdPb ? "" : "B*") << "V - norm " << right.norm() << "):" << endl;
-          cerr << endl << right << endl;
-          cerr << endl << "Error: diff (norm " << diff.norm() << ", tol " << dTol << "):" << endl;
-          cerr << endl << diff << endl;
+        double resNorm = computeResidualNorm(i, A, B);
+        if (resNorm > maxResNorm) {
+          cout << endl << "arpackSolver:" << endl;
+          cout << endl << rs << " value/vector " << setw(3) << i << ": check KO";
+          cout << ", residual (norm " << resNorm << ", maxResNorm " << maxResNorm << ")" << endl;
          return 1;
        }
        else {
          if (verbose >= 1) {
            cout << endl << "arpackSolver:" << endl;
            cout << endl << rs << " value/vector " << setw(3) << i << ": check OK";
-            cout << ", diff (norm " << diff.norm() << ", tol " << dTol << ")" << endl;
+            cout << ", residual (norm " << resNorm << ", maxResNorm " << maxResNorm << ")" << endl;
          }
        }
      }
@@ -342,17 +338,33 @@ class arpackSolver {
      return 0;
    };

+    double computeResidualNorm(size_t const & idx, EM const & A, EM const * B = nullptr) const {
+      // Check eigen value index.
+      if (idx >= val.size()) return -1.; // Error.
+      if (idx >= vec.size()) return -1.; // Error.
+
+      // Compute residual norm.
+      EigVecZ V = vec[idx];
+      complex<double> lambda = val[idx];
+      EigVecZ left = A.template cast<complex<double>>() * V;
+      EigVecZ right = stdPb ? V : B->template cast<complex<double>>() * V;
+      right *= lambda;
+      EigVecZ residual = left - right;
+
+      return residual.norm();
+    };
+
  // Private methods.

  private:

-    void makeZero(         float  & zero) {zero = 0.f;};
+    void makeConstant(         float  & cst,  float const & val) {cst = val;};

-    void makeZero(        double  & zero) {zero = 0.;};
+    void makeConstant(        double  & cst, double const & val) {cst = val;};

-    void makeZero(complex< float> & zero) {zero = complex<double>(0.f, 0.f);};
+    void makeConstant(complex< float> & cst,  float const & val) {cst = complex<float>(val, val);};

-    void makeZero(complex<double> & zero) {zero = complex<double>(0., 0.);};
+    void makeConstant(complex<double> & cst, double const & val) {cst = complex<double>(val, val);};

    int readMatrixMarket(string const & fileName,
                         a_uint & n, a_uint & m, vector<a_uint> & i, vector<a_uint> & j, vector<RC> & Mij) {
@@ -368,7 +380,7 @@ class arpackSolver {
        string inpLine; getline(inp, inpLine); l++;
        while (isspace(*inpLine.begin())) inpLine.erase(inpLine.begin()); // Suppress leading white spaces.
        if (inpLine.length() == 0) continue; // Empty line.
-        if (inpLine[0] == '%') continue; // Comments skipped, begin reading.
+        if (inpLine[0] == '%' || inpLine[0] == '#') continue; // Comments skipped, begin reading.

        // Read matrix market file.

@@ -378,7 +390,13 @@ class arpackSolver {
          if (!inpSS) {cerr << "Error: bad header (n, m)" << endl; return 1;}
          if (nnz == 0) {
            inpSS >> nnz;
-            if (inpSS) { // OK, (optional) nnz has been provided.
+            if (inpSS && nnz > 0) { // OK, (optional) nnz has been provided.
+              i.reserve(nnz);
+              j.reserve(nnz);
+              Mij.reserve(nnz);
+            }
+            else {
+              nnz = n*m;
              i.reserve(nnz);
              j.reserve(nnz);
              Mij.reserve(nnz);
@@ -387,7 +405,7 @@ class arpackSolver {
        }
        else { // Body.
          a_uint k = 0, l = 0;
-          RC zero; makeZero(zero);
+          RC zero; makeConstant(zero, 0.);
          RC Mkl = zero;
          inpSS >> k >> l >> Mkl;
          if (!inpSS) {cerr << "Error: bad line (" << fileName << ", line " << l << ")" << endl; return 1;}
@@ -400,10 +418,20 @@ class arpackSolver {

      // Handle 1-based -> 0-based.

-      nnz = i.size(); // In case nnz was not provided.
-      if (*max_element(begin(i), end(i)) == n || *max_element(begin(j), end(j)) == m) {
-        for (size_t k = 0; k < nnz; k++) i[k] -= 1;
-        for (size_t k = 0; k < nnz; k++) j[k] -= 1;
+      if (!i.empty() && !j.empty()) {
+        if (*max_element(begin(i), end(i)) == n && *max_element(begin(j), end(j)) == m) {
+          for (size_t k = 0; k < i.size(); k++) { if (i[k] > 0) i[k] -= 1; }
+          for (size_t k = 0; k < j.size(); k++) { if (j[k] > 0) j[k] -= 1; }
+        }
+      }
+
+      // Checking indices.
+
+      for (size_t k = 0; k < i.size(); k++) {
+        if (i[k] >= n) {cerr << "Error: bad index (" << fileName << ", i " << i[k] << ")" << endl; return 1;};
+      }
+      for (size_t k = 0; k < j.size(); k++) {
+        if (j[k] >= m) {cerr << "Error: bad index (" << fileName << ", j " << j[k] << ")" << endl; return 1;};
      }

      return 0;
@@ -654,26 +682,48 @@ class arpackSolver {
      return rc;
    };

-    void restartSolve(string const & fileName,
-                      a_int const & nbDim, RC * rv, bool readNbCV) {
-      if (restartFromFile) {
-        ifstream ifs(fileName.c_str());
-        if (ifs.is_open()) {
-          a_int nbCV = 1;
-          if (readNbCV) {
-            ifs >> nbCV;
-            if (nbCV < nbDim) nbCV = nbDim; // Cut-off in case previous run used different nbDim.
+    template <typename T>
+    int saveSolve(string const & fileName,
+                  a_int const & nbDim, T * rv) {
+      ofstream ofs(fileName.c_str(), ofstream::trunc);
+      if (ofs.is_open()) {
+        ofs << nbDim << endl;
+        for (a_int n = 0; rv && n < nbDim; n++) ofs << rv[n] << endl;
+        ofs.close(); // Make sure the file is written.
+      }
+      return 0;
+    };
+
+    template <typename T>
+    int restartSolve(string const & fileName,
+                     a_int const & nbDim, T * rv,
+                     bool allowZero = true) {
+      ifstream ifs(fileName.c_str());
+      if (ifs.is_open()) {
+        a_int nDim = 0;
+        ifs >> nDim;
+        if (nDim != nbDim) {cerr << "Error: bad dim - restart KO" << endl; return 1;}
+        for (a_int n = 0; rv && n < nbDim; n++) {
+          RC val; makeConstant(val, 0.);
+          ifs >> val;
+          if (abs(val) < 1.e-6 && !allowZero) {
+            // Do NOT let residual be zero: this stops arpack to iterate (info = -9).
+            auto eps = numeric_limits<FD>::epsilon();
+            RC epsilon; makeConstant(epsilon, eps);
+            val = epsilon;
          }
-          for (a_int n = 0; rv && n < nbDim*nbCV; n++) ifs >> rv[n];
-          if (verbose >= 1) {
-            cout << endl << "arpackSolver:" << endl;
-            cout << endl << fileName << ": restart OK" << endl;
-            if (verbose >= 2) {
-              for (a_int n = 0; rv && n < nbDim; n++) cout << rv[n] << endl;
-            }
+          rv[n] = val;
+        }
+
+        if (verbose >= 1) {
+          cout << endl << "arpackSolver:" << endl;
+          cout << endl << fileName << ": restart OK" << endl;
+          if (verbose >= 2) {
+            for (a_int n = 0; rv && n < nbDim; n++) cout << rv[n] << endl;
          }
        }
      }
+      return 0;
    };

    int initPointerSize(a_int & iparamSz, a_int & ipntrSz, string const & aeupd) {
@@ -706,18 +756,20 @@ class arpackSolver {
      char const * gMat = "G";
      char const * bMat = (mode == 1) ? iMat : gMat;
      a_int nbDim = A.rows();
-      RC zero; makeZero(zero);
+      auto eps = numeric_limits<FD>::epsilon();
+      RC epsilon; makeConstant(epsilon, eps);
      if (!resid) {
        resid = new RC[nbDim];
-        for (a_int n = 0; n < nbDim; n++) resid[n] = zero; // Avoid "bad" starting vector.
+        // Do NOT let residual be zero: this stops arpack to iterate (info = -9).
+        for (a_int n = 0; n < nbDim; n++) resid[n] = epsilon;
      };
-      restartSolve("arpackSolver.resid.out", nbDim, resid, false);
      a_int ldv = nbDim;
+      RC cst; makeConstant(cst, 10.);
+      RC v0 = cst * epsilon; // Start with something close to zero but *not* exactly zero.
      if (!v) {
        v = new RC[ldv*nbCV];
-        for (a_int n = 0; n < ldv*nbCV; n++) v[n] = zero; // Avoid "bad" starting vector.
+        for (a_int n = 0; n < ldv*nbCV; n++) v[n] = v0;
      };
-      restartSolve("arpackSolver.v.out", ldv, v, true);
      a_int iparamSz = 0, ipntrSz = 0;
      int rc = initPointerSize(iparamSz, ipntrSz, "aupd");
      if (rc != 0) {cerr << "Error: bad iparam/ipntr initialization for aupd" << endl; return rc;}
@@ -727,12 +779,22 @@ class arpackSolver {
      iparamAupd[3] = 1; // Block size.
      iparamAupd[4] = 0; // Number of ev found by arpack.
      iparamAupd[6] = mode;
+      RC zero; makeConstant(zero, 0.);
      vector<a_int> ipntrAupd(ipntrSz, 0);
      RC * workd = new RC[3*nbDim]; for (a_int n = 0; n < 3*nbDim; n++) workd[n] = zero; // Avoid "bad" X/Y vector.
      a_int lworkl = symPb ? nbCV*nbCV + 8*nbCV : 3*nbCV*nbCV + 6*nbCV;
      RC * workl = new RC[lworkl];
      a_int info = 0; // Use random initial residual vector.
-      if (restartFromFile) info = 1;
+
+      // Handling restart.
+
+      if (restartFromFile) {
+        info = 1; // Restart.
+        rc = restartSolve("arpackSolver.resid.out", nbDim, resid, false);
+        if (rc != 0) {cerr << "Error: bad restart (resid)" << endl; return rc;}
+        rc = restartSolve("arpackSolver.v.out", ldv*nbCV, v);
+        if (rc != 0) {cerr << "Error: bad restart (v)" << endl; return rc;}
+      }

      // Initialize solver.

@@ -826,9 +888,11 @@ class arpackSolver {
      rc = eupd(rvec, howmny, select, z, ldz, bMat, nbDim, which, resid, v, ldv, iparamEupd.data(), ipntrEupd.data(), workd, workl, lworkl, rwork, info);
      if (rc != 0) {cerr << "Error: bad arpack eupd" << endl; return rc;}

+      // Save solve (dump results to file) to allow later restart.
+
      if (dumpToFile) {
-        ofstream rfs("arpackSolver.resid.out"); for (a_int n = 0; n < nbDim; n++) rfs << resid[n] << endl;
-        ofstream vfs("arpackSolver.v.out"); vfs << nbCV << endl; for (a_int n = 0; n < ldv*nbCV; n++) vfs << v[n] << endl;
+        saveSolve("arpackSolver.resid.out", nbDim, resid);
+        saveSolve("arpackSolver.v.out", ldv*nbCV, v);
      }

      // Clean.
@@ -1,4 +1,9 @@
+#include <iostream>
+#include <exception>
+#include <sstream>
 #include <chrono>
+#include <vector>
+#include <tuple>

 #include "arpackSolver.hpp"
 #include "debug_c.hpp"
@@ -9,8 +14,8 @@ using namespace std;
 class options {
 public:
  options() {
-    fileA = "A.mtx";
-    fileB = "N.A.";  // Not available.
+    fileA = "";
+    fileB = "";
    dense = false;
    denseRR = true;
    nbEV = 1;
@@ -28,6 +33,7 @@ class options {
    invert =
        false;  // Eigen value invertion: look for 1./lambda instead of lambda.
    tol = 1.e-06;
+    maxResNorm = 1.e-3;
    maxIt = 100;
    schur = false;  // Compute Ritz vectors.
    slv = "BiCG";
@@ -44,27 +50,39 @@ class options {
  };

  int readCmdLine(int argc, char** argv) {
+    // Convert command line to stream.
+
+    stringstream ssIP; // Independent parameters.
+    for (int a = 1; argv && a < argc; a++) {
+      ssIP << argv[a] << " ";
+    }
+
    // Check for command line independent parameters.

-    for (int a = 1; argv && a < argc; a++) {
-      string clo = argv[a];  // Command line option.
-      if (clo == "--help") return usage(0);
+    string clo;  // Command line option.
+    while (ssIP >> clo) {
+      if (clo == "--help" || clo == "-h") return usage();
+      if (clo.find("--") == std::string::npos) {
+        cerr << "Error: bad " << clo << " - bad argument" << endl;
+        return usage();
+      }
+
      if (clo == "--A") {
-        a++;
-        if (a >= argc) {
+        ssIP >> clo;
+        if (!ssIP) {
          cerr << "Error: bad " << clo << " - need argument" << endl;
          return usage();
        }
-        fileA = argv[a];
+        fileA = clo;
      }
      if (clo == "--dense") {
        dense = true;
-        a++;
-        if (a >= argc) {
+        ssIP >> clo;
+        if (!ssIP) {
          cerr << "Error: bad " << clo << " - need argument" << endl;
          return usage();
        }
-        string rr(argv[a]);
+        string rr = clo;
        if (rr != "true" && rr != "false") {
          cerr << "Error: bad " << clo << " - bad argument" << endl;
          return usage();
@@ -72,12 +90,12 @@ class options {
        denseRR = (rr == "true") ? true : false;
      }
      if (clo == "--nbEV") {
-        a++;
-        if (a >= argc) {
+        ssIP >> clo;
+        if (!ssIP) {
          cerr << "Error: bad " << clo << " - need argument" << endl;
          return usage();
        }
-        stringstream nEV(argv[a]);
+        stringstream nEV(clo);
        nEV >> nbEV;
        if (!nEV) {
          cerr << "Error: bad " << clo << " - bad argument" << endl;
@@ -85,10 +103,6 @@ class options {
        }
        nbCV = 2 * nbEV + 1;
      }
-      if (clo == "--genPb") {
-        stdPb = false;
-        fileB = "B.mtx";
-      }
      if (clo == "--nonSymPb") symPb = false;
      if (clo == "--cpxPb") {
        symPb = false;
@@ -96,15 +110,15 @@ class options {
      }
      if (clo == "--simplePrec") simplePrec = true;
      if (clo == "--mag") {
-        a++;
-        if (a >= argc) {
+        ssIP >> clo;
+        if (!ssIP) {
          cerr << "Error: bad " << clo << " - need argument" << endl;
          return usage();
        }
-        mag = argv[a];  // small mag (likely poor perf) <=> large mag + invert
+        mag = clo;  // small mag (likely poor perf) <=> large mag + invert
                        // (likely good perf).
        bool ok = (mag == "LM" || mag == "SM" || mag == "LR" || mag == "SR" ||
-                   mag == "LI" || mag == "SI")
+                   mag == "LA" || mag == "SA" || mag == "LI" || mag == "SI")
                      ? true
                      : false;
        if (!ok) {
@@ -114,12 +128,12 @@ class options {
      }
      if (clo == "--shiftReal") {
        shiftReal = true;
-        a++;
-        if (a >= argc) {
+        ssIP >> clo;
+        if (!ssIP) {
          cerr << "Error: bad " << clo << " - need argument" << endl;
          return usage();
        }
-        stringstream s(argv[a]);
+        stringstream s(clo);
        s >> sigmaReal;
        if (!s) {
          cerr << "Error: bad " << clo << " - bad argument" << endl;
@@ -128,12 +142,12 @@ class options {
      }
      if (clo == "--shiftImag") {
        shiftImag = true;
-        a++;
-        if (a >= argc) {
+        ssIP >> clo;
+        if (!ssIP) {
          cerr << "Error: bad " << clo << " - need argument" << endl;
          return usage();
        }
-        stringstream s(argv[a]);
+        stringstream s(clo);
        s >> sigmaImag;
        if (!s) {
          cerr << "Error: bad " << clo << " - bad argument" << endl;
@@ -142,47 +156,62 @@ class options {
      }
      if (clo == "--invert") invert = true;
      if (clo == "--tol") {
-        a++;
-        if (a >= argc) {
+        ssIP >> clo;
+        if (!ssIP) {
          cerr << "Error: bad " << clo << " - need argument" << endl;
          return usage();
        }
-        stringstream t(argv[a]);
+        stringstream t(clo);
        t >> tol;
        if (!t) {
          cerr << "Error: bad " << clo << " - bad argument" << endl;
          return usage();
        }
      }
-      if (clo == "--maxIt") {
-        a++;
-        if (a >= argc) {
+      if (clo == "--maxResNorm") {
+        ssIP >> clo;
+        if (!ssIP) {
          cerr << "Error: bad " << clo << " - need argument" << endl;
          return usage();
        }
-        stringstream mi(argv[a]);
+        stringstream mrn(clo);
+        mrn >> maxResNorm;
+        if (!mrn) {
+          cerr << "Error: bad " << clo << " - bad argument" << endl;
+          return usage();
+        }
+      }
+      if (clo == "--maxIt") {
+        ssIP >> clo;
+        if (!ssIP) {
+          cerr << "Error: bad " << clo << " - need argument" << endl;
+          return usage();
+        }
+        stringstream mi(clo);
        mi >> maxIt;
        if (!mi) {
          cerr << "Error: bad " << clo << " - bad argument" << endl;
          return usage();
        }
      }
+      if (clo == "--schur") {
+        schur = true;
+      }
      if (clo == "--slv") {
-        a++;
-        if (a >= argc) {
+        ssIP >> clo;
+        if (!ssIP) {
          cerr << "Error: bad " << clo << " - need argument" << endl;
          return usage();
        }
-        slv = argv[a];
+        slv = clo;
      }
      if (clo == "--slvItrTol") {
-        a++;
-        if (a >= argc) {
+        ssIP >> clo;
+        if (!ssIP) {
          cerr << "Error: bad " << clo << " - need argument" << endl;
          return usage();
        }
-        stringstream t(argv[a]);
-        double tol = 0.;
+        stringstream t(clo);
        t >> slvItrTol;
        if (!t) {
          cerr << "Error: bad " << clo << " - bad argument" << endl;
@@ -190,12 +219,12 @@ class options {
        }
      }
      if (clo == "--slvItrMaxIt") {
-        a++;
-        if (a >= argc) {
+        ssIP >> clo;
+        if (!ssIP) {
          cerr << "Error: bad " << clo << " - need argument" << endl;
          return usage();
        }
-        stringstream mi(argv[a]);
+        stringstream mi(clo);
        int maxIt = 0;
        mi >> slvItrMaxIt;
        if (!mi) {
@@ -204,12 +233,12 @@ class options {
        }
      }
      if (clo == "--slvItrPC") {
-        a++;
-        if (a >= argc) {
+        ssIP >> clo;
+        if (!ssIP) {
          cerr << "Error: bad " << clo << " - need argument" << endl;
          return usage();
        }
-        stringstream pc(argv[a]);
+        stringstream pc(clo);
        pc >> slvItrPC;
        if (!pc) {
          cerr << "Error: bad " << clo << " - bad argument" << endl;
@@ -217,12 +246,12 @@ class options {
        }
      }
      if (clo == "--slvDrtPivot") {
-        a++;
-        if (a >= argc) {
+        ssIP >> clo;
+        if (!ssIP) {
          cerr << "Error: bad " << clo << " - need argument" << endl;
          return usage();
        }
-        stringstream pv(argv[a]);
+        stringstream pv(clo);
        pv >> slvDrtPivot;
        if (!pv) {
          cerr << "Error: bad " << clo << " - bad argument" << endl;
@@ -230,12 +259,12 @@ class options {
        }
      }
      if (clo == "--slvDrtOffset") {
-        a++;
-        if (a >= argc) {
+        ssIP >> clo;
+        if (!ssIP) {
          cerr << "Error: bad " << clo << " - need argument" << endl;
          return usage();
        }
-        stringstream of(argv[a]);
+        stringstream of(clo);
        of >> slvDrtOffset;
        if (!of) {
          cerr << "Error: bad " << clo << " - bad argument" << endl;
@@ -243,12 +272,12 @@ class options {
        }
      }
      if (clo == "--slvDrtScale") {
-        a++;
-        if (a >= argc) {
+        ssIP >> clo;
+        if (!ssIP) {
          cerr << "Error: bad " << clo << " - need argument" << endl;
          return usage();
        }
-        stringstream sc(argv[a]);
+        stringstream sc(clo);
        sc >> slvDrtScale;
        if (!sc) {
          cerr << "Error: bad " << clo << " - bad argument" << endl;
@@ -257,12 +286,12 @@ class options {
      }
      if (clo == "--noCheck") check = false;
      if (clo == "--verbose") {
-        a++;
-        if (a >= argc) {
+        ssIP >> clo;
+        if (!ssIP) {
          cerr << "Error: bad " << clo << " - need argument" << endl;
          return usage();
        }
-        stringstream vb(argv[a]);
+        stringstream vb(clo);
        vb >> verbose;
        if (!vb) {
          cerr << "Error: bad " << clo << " - bad argument" << endl;
@@ -270,12 +299,12 @@ class options {
        }
      }
      if (clo == "--debug") {
-        a++;
-        if (a >= argc) {
+        ssIP >> clo;
+        if (!ssIP) {
          cerr << "Error: bad " << clo << " - need argument" << endl;
          return usage();
        }
-        stringstream dbg(argv[a]);
+        stringstream dbg(clo);
        dbg >> debug;
        if (!dbg) {
          cerr << "Error: bad " << clo << " - bad argument" << endl;
@@ -287,19 +316,29 @@ class options {
                debug, debug, debug, debug, debug);
      }
      if (clo == "--restart") restart = true;
+
+      // Skip dependent parameters.
+      if (clo == "--nbCV") { ssIP >> clo; continue; }
+      if (clo == "--B") { ssIP >> clo; continue; }
+    }
+
+    // Convert command line to stream.
+
+    stringstream ssDP; // Dependent parameters.
+    for (int a = 1; argv && a < argc; a++) {
+      ssDP << argv[a] << " ";
    }

    // Check for command line dependent parameters.

-    for (int a = 1; argv && a < argc; a++) {
-      string clo = argv[a];  // Command line option.
+    while (ssDP >> clo) {
      if (clo == "--nbCV") {
-        a++;
-        if (a >= argc) {
+        ssDP >> clo;
+        if (!ssDP) {
          cerr << "Error: bad " << clo << " - need argument" << endl;
          return usage();
        }
-        stringstream nCV(argv[a]);
+        stringstream nCV(clo);
        nCV >> nbCV;
        if (!nCV) {
          cerr << "Error: bad " << clo << " - bad argument" << endl;
@@ -307,26 +346,20 @@ class options {
        }
      }
      if (clo == "--B") {
-        a++;
-        if (a >= argc) {
+        ssDP >> clo;
+        if (!ssDP) {
          cerr << "Error: bad " << clo << " - need argument" << endl;
          return usage();
        }
-        fileB = argv[a];
+        fileB = clo;
+        stdPb = false; // Generalized problem.
      }
    }

-    // Sanity checks.
-
-    if (!stdPb && fileB.empty()) {
-      cerr << "Error: generalized problem without B matrix" << endl;
-      return usage();
-    }
-
    return 0;
  };

-  int usage(int rc = 1) {
+  int usage() {
    cout << "Usage: running arpack with matrix market files to check for eigen "
            "values/vectors."
         << endl;
@@ -371,8 +404,6 @@ class options {
    cout << "                    default: 1" << endl;
    cout << "  --nbCV:           number of columns of the matrix V." << endl;
    cout << "                    default: 2*nbEV+1" << endl;
-    cout << "  --genPb:          generalized problem." << endl;
-    cout << "                    default: standard problem" << endl;
    cout << "  --nonSymPb:       non symmetric problem (<=> use dn[ae]upd)."
         << endl;
    cout << "                    default: symmetric problem (<=> use ds[ae]upd)"
@@ -388,7 +419,7 @@ class options {
            "[dz]*upd)"
         << endl;
    cout << "  --mag M:          set magnitude of eigen values to look for "
-            "(LM, SM, LR, SR, LI, SI)."
+            "(LM, SM, LR, SR, LA, SA, LI, SI)."
         << endl;
    cout << "                    default: large magnitude (LM)" << endl;
    cout << "  --shiftReal S:    real shift where sigma = S (look for lambda+S "
@@ -405,6 +436,8 @@ class options {
    cout << "                    default: no invert" << endl;
    cout << "  --tol T:          tolerance T." << endl;
    cout << "                    default: 1.e-06" << endl;
+    cout << "  --maxResNorm R:   maximum residual norm R." << endl;
+    cout << "                    default: 1.e-3" << endl;
    cout << "  --maxIt M:        maximum iterations M." << endl;
    cout << "                    default: 100" << endl;
    cout << "  --schur:          compute Schur vectors." << endl;
@@ -497,8 +530,7 @@ class options {
            "computed during a previous run."
         << endl;
    cout << "                    default: false" << endl;
-    if (rc == 0) exit(0);
-    return rc;
+    return 1;
  };

  friend ostream& operator<<(ostream& ostr, options const& opt);
@@ -519,6 +551,7 @@ class options {
                                // lambda+sigma instead of lambda.
  bool invert;  // Eigen value invertion: look for 1./lambda instead of lambda.
  double tol;
+  double maxResNorm;
  int maxIt;
  bool schur;
  string slv;
@@ -535,7 +568,8 @@ class options {
 };

 ostream& operator<<(ostream& ostr, options const& opt) {
-  ostr << "OPT: A " << opt.fileA << ", B " << opt.fileB;
+  ostr << "OPT: A " << opt.fileA;
+  if (!opt.fileB.empty()) ostr << ", B " << opt.fileB;
  if (opt.dense && opt.denseRR)
    ostr << ", dense yes (RR true)";
  else if (opt.dense && !opt.denseRR)
@@ -553,14 +587,25 @@ ostream& operator<<(ostream& ostr, options const& opt) {
  ostr << ", shiftImag " << (opt.shiftImag ? "yes" : "no") << ", sigmaImag "
       << opt.sigmaImag;
  ostr << ", invert " << (opt.invert ? "yes" : "no") << ", tol " << opt.tol
-       << ", maxIt " << opt.maxIt;
+       << ", maxResNorm " << opt.maxResNorm << ", maxIt " << opt.maxIt;
  ostr << ", " << (opt.schur ? "Schur" : "Ritz") << " vectors" << endl;
-  ostr << "OPT: slv " << opt.slv << ", slvItrPC " << opt.slvItrPC
-       << ", slvItrTol " << opt.slvItrTol;
-  ostr << ", slvItrMaxIt " << opt.slvItrMaxIt << ", slvDrtPivot "
-       << opt.slvDrtPivot;
-  ostr << ", slvDrtOffset " << opt.slvDrtOffset << ", slvDrtScale "
-       << opt.slvDrtScale << endl;
+  ostr << "OPT: slv " << opt.slv;
+  bool itrSlv = true;  // Use iterative solvers.
+  if (opt.slv.find("LU") != string::npos ||
+      opt.slv.find("QR") != string::npos ||
+      opt.slv.find("LLT") != string::npos ||
+      opt.slv.find("LDLT") != string::npos)
+    itrSlv = false;
+  if (itrSlv) {
+    ostr << ", slvItrPC " << opt.slvItrPC;
+    ostr << ", slvItrTol " << opt.slvItrTol;
+    ostr << ", slvItrMaxIt " << opt.slvItrMaxIt;
+  } else {
+    ostr << ", slvDrtPivot " << opt.slvDrtPivot;
+    ostr << ", slvDrtOffset " << opt.slvDrtOffset;
+    ostr << ", slvDrtScale " << opt.slvDrtScale;
+  }
+  ostr << endl;
  ostr << "OPT: check " << (opt.check ? "yes" : "no") << ", verbose "
       << opt.verbose << ", debug " << opt.debug;
  ostr << ", restart " << (opt.restart ? "yes" : "no") << endl;
@@ -582,6 +627,7 @@ class output {
  int nbIt;        // Arpack number of iterations.
  double imsTime;  // Init mode solver time.
  double rciTime;  // Reverse communication interface time.
+  vector<tuple<double, double>> res; // Results: eigen values and associated residual.
 };

 template <typename RC, typename FD, typename EM, typename SLV>
@@ -622,8 +668,6 @@ int itrSolve(options& opt, output& out, double const& slvItrILUDropTol,
  cout << endl;
  cout << "INP: create A " << readATime << " s" << endl;

-  if (opt.nbCV > A.cols()) opt.nbCV = A.cols(); /* Cut-off */
-
  EM B;
  if (!opt.stdPb) {
    start = chrono::high_resolution_clock::now();
@@ -657,7 +701,7 @@ int itrSolve(options& opt, output& out, double const& slvItrILUDropTol,
    return rc;
  }
  if (opt.check) {
-    rc = as.checkEigVec(A, opt.stdPb ? nullptr : &B);
+    rc = as.checkEigVec(A, opt.stdPb ? nullptr : &B, opt.maxResNorm);
    if (rc != 0) {
      cerr << "Error: check KO" << endl;
      return rc;
@@ -671,6 +715,11 @@ int itrSolve(options& opt, output& out, double const& slvItrILUDropTol,
  out.nbIt = as.nbIt;
  out.imsTime = as.imsTime;
  out.rciTime = as.rciTime;
+  for (auto idx = 0; idx < as.val.size(); idx++) {
+    double val = norm(as.val[idx]);
+    double res = as.computeResidualNorm(idx, A, opt.stdPb ? nullptr : &B);
+    out.res.push_back(tuple<double, double>{val, res});
+  }

  return 0;
 }
@@ -711,8 +760,6 @@ int drtSolve(options& opt, output& out) {
  cout << endl;
  cout << "INP: create A " << readATime << " s" << endl;

-  if (opt.nbCV > A.cols()) opt.nbCV = A.cols(); /* Cut-off */
-
  EM B;
  if (!opt.stdPb) {
    start = chrono::high_resolution_clock::now();
@@ -746,7 +793,7 @@ int drtSolve(options& opt, output& out) {
    return rc;
  }
  if (opt.check) {
-    rc = as.checkEigVec(A, opt.stdPb ? nullptr : &B);
+    rc = as.checkEigVec(A, opt.stdPb ? nullptr : &B, opt.maxResNorm);
    if (rc != 0) {
      cerr << "Error: check KO" << endl;
      return rc;
@@ -760,6 +807,11 @@ int drtSolve(options& opt, output& out) {
  out.nbIt = as.nbIt;
  out.imsTime = as.imsTime;
  out.rciTime = as.rciTime;
+  for (auto idx = 0; idx < as.val.size(); idx++) {
+    double val = norm(as.val[idx]);
+    double res = as.computeResidualNorm(idx, A, opt.stdPb ? nullptr : &B);
+    out.res.push_back(tuple<double, double>{val, res});
+  }

  return 0;
 }
@@ -822,7 +874,7 @@ int itrSolve(options& opt, output& out) {
  return rc;
 }

-int main(int argc, char** argv) {
+int run(int argc, char** argv) {
  // Check for options.

  options opt;
@@ -851,6 +903,7 @@ int main(int argc, char** argv) {
  if (opt.dense) {
    if (itrSlv) {
      cerr << "Error: dense matrices does not support iterative solvers"
+           << ", specify --slv XX with XX being a direct solver"
           << endl;
      return 1;
    }
@@ -972,9 +1025,30 @@ int main(int argc, char** argv) {
  cout << "STAT: total number of restart steps                               "
       << nrstrt << endl;

+  // Output eigen values and residuals.
+
+  cout << endl;
+  for (auto idx = 0; idx < out.res.size(); idx++) {
+    tuple<double, double> res = out.res[idx];
+    cout << "RES: eigen value " << idx << ": norm " << get<0>(res);
+    cout << " (residual norm " << get<1>(res) << ")";
+    cout << endl;
+  }
+
  return 0;
 }

+int main(int argc, char** argv) {
+  int rc = 1;
+  try {
+    rc = run(argc, argv);
+  }
+  catch (exception& e) {
+    cout << "Error - exception:" << e.what() << endl;
+  }
+  return rc;
+}
+
 // Local Variables:
 // mode: c++
 // c-file-style:"stroustrup"
@@ -1,90 +1,94 @@
 #!/bin/bash -eu

-export CMD="./arpackmm --help" # For coverage
-echo "$CMD"
-eval "$CMD &> arpackmm.run.log"
-echo ""
-echo "========================================================================================"
-echo ""
+trap 'catch' ERR
+catch() {
+  grep OPT arpackmm.run.log
+  grep Error arpackmm.run.log
+  exit 1
+}

-for eigPb in "--A As.mtx" "--nonSymPb --A An.mtx" "--nonSymPb --cpxPb --A Az.mtx --B Bz.mtx"
+# For all these eigen problems, the first eigen value is about 382 or (382, 0.).
+
+for stdPb in "--A As.mtx" "--nonSymPb --A An.mtx" "--nonSymPb --cpxPb --A Az.mtx"
 do
-  for genPb in "" "--genPb"
+  # Choose B matrix according to A.
+  export fileB=""
+  if [[ "$stdPb" == *cpxPb* ]]; then
+    export fileB="--B Bz.mtx"
+  else
+    export fileB="--B B.mtx"
+  fi
+
+  for genPb in "" "$fileB"
  do
-    for smallMag in "" "--mag SM --noCheck" # SM is known to be difficult to converge.
+    # SM may not converge so we can not apply checks.
+    # LM + invert is equivalent to SM and does converge.
+    # Note: this is expected as "power-like" methods are designed to find largest eigen values.
+    for magOpt in "--mag LM" "--mag SM --noCheck"
    do
-      export shiftOpt=""
-      if [[ "$eigPb" == *nonSymPb* ]]; then
-        if [[ "$genPb" == *genPb* ]]; then
-          continue # Skip to ensure stable test: tricky to convergence.
-        else
-          export shiftOpt="--shiftReal 100.0 --shiftImag 100.0"
-        fi
+      export mrn="--maxResNorm 1.e-1" # Relax residual check to get stable tests.
+
+      # Shift slightly to avoid the zero-vector starting problem.
+      export shiftZV=""
+      if [[ "$stdPb" == *cpxPb* ]]; then
+        export shiftZV="--shiftReal 1.0 --shiftImag 1.0"
      else
-        if [[ "$genPb" == *genPb* ]]; then
-          continue # Skip to ensure stable test: tricky to convergence.
-        else
-          export shiftOpt="--shiftReal 100.0"
-        fi
+        export shiftZV="--shiftReal 1.0"
      fi

-      for shiftRI in "" "$shiftOpt"
+      # Shift according to the estimation of the eigen value we may have.
+      export shiftEV=""
+      if [[ "$stdPb" == *cpxPb* ]]; then
+        export shiftEV="--shiftReal 380.0 --shiftImag 1.0"
+      else
+        export shiftEV="--shiftReal 380.0"
+      fi
+
+      for shiftRI in "$shiftZV" "$shiftEV"
      do
        for invert in "" "--invert"
        do
-          for tol in "" "--tol 1.e-5"
+          # Choose solver according to the sym/non-sym type of the problem.
+          export cgSlv=""
+          if [[ "$stdPb" == *nonSymPb* ]]; then
+            export cgSlv="BiCG"
+          else
+            export cgSlv="CG"
+          fi
+
+          for slv in "--slv $cgSlv" "--slv LU"
          do
-            for slv in "--slv BiCG --slvItrTol 1.e-06 --slvItrMaxIt 150" "--slv   CG --slvItrTol 1.e-06 --slvItrMaxIt 150" \
-                       "--slv BiCG --slvItrPC ILU"                       "--slv   CG --slvItrPC ILU#1.e-06#2"              \
-                       "--slv   LU"                                      "--slv   QR --slvDrtPivot 1.e-06"                 \
-                       "--slv  LLT"                                      "--slv  LLT --slvDrtOffset 0."                    \
-                       "--slv LDLT"                                      "--slv LDLT --slvDrtScale 1."
+            for rs in "" "--schur"
            do
-              for rs in "" "--schur"
+              for dsPrec in "" "--simplePrec"
              do
-                for dsPrec in "" "--simplePrec"
+                for dsMat in "" "--dense true"
                do
-                  for dsMat in "" "--dense false" "--dense true"
-                  do
-                    export extraGenPb=""
-                    if [[ "$genPb" == *genPb* ]]; then
-                      export extraGenPb="$shiftOpt" # Force shift if genPb.
-                    fi
-
+                  # Skip not supported cases.
+                  if [[ "$dsMat" == *dense* ]]; then
                    if [[ "$slv" == *CG* ]]; then
-                      if [[ "$eigPb" == *nonSymPb* ]]; then
-                        continue # Skip CG that could fail (CG is meant to deal with sym matrices).
-                      fi
+                      continue # Iterative solvers are not allowed when using dense matrices.
                    fi
+                  fi

-                    if [[ "$slv" == *LLT* ]] || [[ "$slv" == *LDLT* ]]; then
-                      if [[ "$eigPb" == *nonSymPb* ]] || [[ "$genPb" == *genPb* ]]; then
-                        continue # Skip LLT/LDLT that could fail (LLT/LDLT are meant to deal with SPD matrices).
-                      fi
-                    fi
+                  export easeCV="--nbCV 6 --maxIt 200" # Use --nbCV 6 and --maxIt 200 to ease convergence.
+                  echo "CLI: ./arpackmm $stdPb $genPb $magOpt $mrn $shiftRI $invert $slv $rs $dsPrec $dsMat $easeCV"
+                  echo "----------------------------------------------------------------------------------------"

-                    if [[ "$dsMat" == *dense* ]]; then
-                      if [[ "$slv" == *CG* ]]; then
-                        continue # Iterative solvers are not allowed when using dense matrices.
-                      fi
-                    fi
+                  # Run arpackmm.
+                  ./arpackmm "$stdPb" "$genPb" "$magOpt" "$mrn" "$shiftRI" "$invert" "$slv" "$rs" "$dsPrec" "$dsMat" \
+                             "$easeCV" --verbose 3 &> arpackmm.run.log
+                  grep OPT arpackmm.run.log
+                  grep OUT arpackmm.run.log
+                  echo "----------------------------------------------------------------------------------------"

-                    # Run arpackmm: use --nbCV 6 and --maxIt 200 to ease convergence, and, --verbose 3 for debug.
-                    export CMD="./arpackmm $eigPb $genPb $smallMag $shiftRI $invert $tol $slv $rs $dsPrec $dsMat $extraGenPb --nbCV 6 --maxIt 200 --verbose 3 --debug 3"
-                    echo "$CMD"
-                    eval "$CMD &> arpackmm.run.log"
-                    echo ""
-                    echo "========================================================================================"
-                    echo ""
-
-                    # Run arpackmm: re-run with restart.
-                    export CMD="$CMD --restart"
-                    echo "$CMD"
-                    eval "$CMD &> arpackmm.run.log"
-                    echo ""
-                    echo "========================================================================================"
-                    echo ""
-                  done
+                  # Run arpackmm: re-run with restart always with small shift to avoid the zero-starting vector problem.
+                  ./arpackmm "$stdPb" "$genPb" "$magOpt" "$mrn" "$shiftZV" "$invert" "$slv" "$rs" "$dsPrec" "$dsMat" \
+                             --restart                                                                               \
+                             "$easeCV" --verbose 3 &> arpackmm.run.log
+                  grep OPT arpackmm.run.log
+                  grep OUT arpackmm.run.log
+                  echo "========================================================================================"
                done
              done
            done
@@ -95,4 +99,4 @@ do
  done
 done

-echo "OK"
+echo "arpackmm: OK"
@@ -0,0 +1,6 @@
+%% MatrixMarket matrix coordinate double symmetric
+% 4x4 matrix filled with zero
+%
+
+4 4 0
+
@@ -0,0 +1,9 @@
+#!/bin/bash -eu
+
+# Note: need to shift slightly as eigen value is zero.
+./arpackmm --A issue215.mtx --mag LM --nbEV 1 --nbCV 4 --shiftReal 0.1
+if [ "$?" -ne "0" ]; then exit 1; fi
+./arpackmm --A issue215.mtx --mag LM --nbEV 1 --nbCV 4 --shiftReal 0.1 --restart
+if [ "$?" -ne "0" ]; then exit 1; fi
+
+echo "OK"
@@ -0,0 +1,8 @@
+%% MatrixMarket matrix coordinate double symmetric
+% 10x10 matrix with 2 non-zero values
+%
+
+10 10 2
+
+0  1  1.
+1  0  1.
@@ -0,0 +1,10 @@
+#!/bin/bash -eu
+
+./arpackmm --A issue401.mtx --mag LA --nbEV 1 --nbCV 5
+if [ "$?" -ne "0" ]; then exit 1; fi
+./arpackmm --A issue401.mtx --mag LA --nbEV 1 --nbCV 5 --restart
+if [ "$?" -ne "0" ]; then exit 1; fi
+./arpackmm --A issue401.mtx --mag LA --nbEV 1 --nbCV 5 --restart --shiftReal 0.9
+if [ "$?" -ne "0" ]; then exit 1; fi
+
+echo "OK"
@@ -1,5 +1,5 @@
 SUBDIRS = BAND COMPLEX NONSYM SIMPLE SVD SYM
-if ICBEXMM
+if EIGEN
 SUBDIRS += MATRIX_MARKET
 endif

@@ -29,8 +29,9 @@ void exportArpackSparseItr(bp::scope& pySlv, std::string const& dtype) {
          .def("checkEigVec",
               &pyarpackSparseItrSolver<RC, FD, EM, SLV>::checkEigVec,
               (bp::arg("A"), bp::arg("B") = bp::tuple(),
-                bp::arg("diffTol") = 1.e-3),
-               "check eigen vectors accuracy where A and B must be sparse and "
+                bp::arg("maxResNorm") = 1.e-3),
+               "check eigen vectors accuracy (according to max allowed residual norm) "
+               "where A and B must be sparse and "
               "provided in coo format: (dimension, row-indice array, "
               "column-indice array, matrice-value array) tuple")
              ARPACKSOLVERMEMBER(pyarpackSparseItrSolver)
@@ -69,8 +70,9 @@ void exportArpackSparseDrt(bp::scope& pySlv, std::string const& dtype) {
          .def("checkEigVec",
               &pyarpackSparseDrtSolver<RC, FD, EM, SLV>::checkEigVec,
               (bp::arg("A"), bp::arg("B") = bp::tuple(),
-                bp::arg("diffTol") = 1.e-3),
-               "check eigen vectors accuracy where A and B must be sparse and "
+                bp::arg("maxResNorm") = 1.e-3),
+               "check eigen vectors accuracy (according to max allowed residual norm) "
+               "where A and B must be sparse and "
               "provided in coo format: (dimension, row-indice array, "
               "column-indice array, matrice-value array) tuple")
              ARPACKSOLVERMEMBER(pyarpackSparseDrtSolver)
@@ -103,8 +105,9 @@ void exportArpackDenseDrt(bp::scope& pySlv, std::string const& dtype) {
          .def("checkEigVec",
               &pyarpackDenseDrtSolver<RC, FD, EM, SLV>::checkEigVec,
               (bp::arg("A"), bp::arg("B") = bp::tuple(),
-                bp::arg("diffTol") = 1.e-3),
-               "check eigen vectors accuracy where A and B must be dense and "
+                bp::arg("maxResNorm") = 1.e-3),
+               "check eigen vectors accuracy (according to max allowed residual norm) "
+               "where A and B must be dense and "
               "provided in raw format: (n-squared matrice-value array, row or "
               "column ordered boolean)")
              ARPACKSOLVERMEMBER(pyarpackDenseDrtSolver)
@@ -6,24 +6,22 @@ from pyarpack import denseLDLT as pyarpackSlv
 # Build laplacian.

 n = 4
-Aij = np.array([], dtype='complex128')
+Aij = np.array([], dtype='float64')
 for k in range(n):
    for l in range(n):
        if l == k:
-            Aij = np.append(Aij, np.complex128(complex( 200.,  200.))) # Casting value on append is MANDATORY or C++ won't get the expected type.
-        elif l == k-1:
-            Aij = np.append(Aij, np.complex128(complex(-101., -101.))) # Casting value on append is MANDATORY or C++ won't get the expected type.
-        elif l == k+1:
-            Aij = np.append(Aij, np.complex128(complex( -99.,  -99.))) # Casting value on append is MANDATORY or C++ won't get the expected type.
+            Aij = np.append(Aij, np.float64( 200.)) # Casting value on append is MANDATORY or C++ won't get the expected type.
+        elif l == k-1 or l == k+1:
+            Aij = np.append(Aij, np.float64(-100.)) # Casting value on append is MANDATORY or C++ won't get the expected type.
        else:
-            Aij = np.append(Aij, np.complex128(complex(   0.,    0.))) # Casting value on append is MANDATORY or C++ won't get the expected type.
+            Aij = np.append(Aij, np.float64(   0.)) # Casting value on append is MANDATORY or C++ won't get the expected type.
 for idx, val in enumerate(Aij):
    print("A[", idx, "] =", val)
 A = (Aij, False) # raw format: Aij values, row ordered (or not).

 # Get and tune arpack solver.

-arpackSlv = pyarpackSlv.complexDouble() # Caution: complexDouble <=> np.array(..., dtype='complex128')
+arpackSlv = pyarpackSlv.double() # Caution: double <=> np.array(..., dtype='float64')
 arpackSlv.verbose = 3 # Set to 0 to get a quiet solve.
 arpackSlv.debug = 1 # Set to 0 to get a quiet solve.
 arpackSlv.nbEV = 1
@@ -32,7 +30,7 @@ arpackSlv.mag = 'LM'
 arpackSlv.maxIt = 200
 arpackSlv.slvOffset = 0.
 arpackSlv.slvScale = 1.
-arpackSlv.symPb = False
+arpackSlv.symPb = True

 # Solve eigen problem.

@@ -60,22 +58,19 @@ print("\n#######################################################################
 # Build laplacian.

 n = 8
-Aij = np.array([], dtype='complex64')
-Bij = np.array([], dtype='complex64')
+Aij = np.array([], dtype='float32')
+Bij = np.array([], dtype='float32')
 for k in range(n):
    for l in range(n):
        if l == k:
-            Aij = np.append(Aij, np.complex64(complex( 200.,  200.))) # Casting value on append is MANDATORY or C++ won't get the expected type.
-            Bij = np.append(Bij, np.complex64(complex( 33.3,  33.3))) # Casting value on append is MANDATORY or C++ won't get the expected type.
-        elif l == k-1:
-            Aij = np.append(Aij, np.complex64(complex(-101., -101.))) # Casting value on append is MANDATORY or C++ won't get the expected type.
-            Bij = np.append(Bij, np.complex64(complex( 16.6,  16.6))) # Casting value on append is MANDATORY or C++ won't get the expected type.
-        elif l == k+1:
-            Aij = np.append(Aij, np.complex64(complex( -99.,  -99.))) # Casting value on append is MANDATORY or C++ won't get the expected type.
-            Bij = np.append(Bij, np.complex64(complex( 16.6,  16.6))) # Casting value on append is MANDATORY or C++ won't get the expected type.
+            Aij = np.append(Aij, np.float32( 200.)) # Casting value on append is MANDATORY or C++ won't get the expected type.
+            Bij = np.append(Bij, np.float32( 33.3)) # Casting value on append is MANDATORY or C++ won't get the expected type.
+        elif l == k-1 or l == k+1:
+            Aij = np.append(Aij, np.float32(-100.)) # Casting value on append is MANDATORY or C++ won't get the expected type.
+            Bij = np.append(Bij, np.float32( 16.6)) # Casting value on append is MANDATORY or C++ won't get the expected type.
        else:
-            Aij = np.append(Aij, np.complex64(complex(   0.,    0.))) # Casting value on append is MANDATORY or C++ won't get the expected type.
-            Bij = np.append(Bij, np.complex64(complex(   0.,    0.))) # Casting value on append is MANDATORY or C++ won't get the expected type.
+            Aij = np.append(Aij, np.float32(   0.)) # Casting value on append is MANDATORY or C++ won't get the expected type.
+            Bij = np.append(Bij, np.float32(   0.)) # Casting value on append is MANDATORY or C++ won't get the expected type.
 for idx, val in enumerate(Aij):
    print("A[", idx, "] =", val)
 for idx, val in enumerate(Bij):
@@ -85,7 +80,7 @@ B = (Bij,  True) # raw format: Bij values, row ordered (or not).

 # Get and tune arpack solver.

-arpackSlv = pyarpackSlv.complexFloat() # Caution: complexFloat <=> np.array(..., dtype='complex64')
+arpackSlv = pyarpackSlv.float() # Caution: float <=> np.array(..., dtype='float32')
 arpackSlv.verbose = 3 # Set to 0 to get a quiet solve.
 arpackSlv.debug = 1 # Set to 0 to get a quiet solve.
 arpackSlv.nbEV = 2
@@ -95,7 +90,7 @@ arpackSlv.maxIt = 200
 arpackSlv.slvOffset = 0.
 arpackSlv.slvScale = 1.
 arpackSlv.sigmaReal = 1
-arpackSlv.symPb = False
+arpackSlv.symPb = True

 # Solve eigen problem.

@@ -6,22 +6,24 @@ from pyarpack import denseLUPP as pyarpackSlv
 # Build laplacian.

 n = 4
-Aij = np.array([], dtype='float64')
+Aij = np.array([], dtype='complex128')
 for k in range(n):
    for l in range(n):
        if l == k:
-            Aij = np.append(Aij, np.float64( 200.)) # Casting value on append is MANDATORY or C++ won't get the expected type.
-        elif l == k-1 or l == k+1:
-            Aij = np.append(Aij, np.float64(-100.)) # Casting value on append is MANDATORY or C++ won't get the expected type.
+            Aij = np.append(Aij, np.complex128(complex( 200.,  200.))) # Casting value on append is MANDATORY or C++ won't get the expected type.
+        elif l == k-1:
+            Aij = np.append(Aij, np.complex128(complex(-101., -101.))) # Casting value on append is MANDATORY or C++ won't get the expected type.
+        elif l == k+1:
+            Aij = np.append(Aij, np.complex128(complex( -99.,  -99.))) # Casting value on append is MANDATORY or C++ won't get the expected type.
        else:
-            Aij = np.append(Aij, np.float64(   0.)) # Casting value on append is MANDATORY or C++ won't get the expected type.
+            Aij = np.append(Aij, np.complex128(complex(   0.,    0.))) # Casting value on append is MANDATORY or C++ won't get the expected type.
 for idx, val in enumerate(Aij):
    print("A[", idx, "] =", val)
 A = (Aij, False) # raw format: Aij values, row ordered (or not).

 # Get and tune arpack solver.

-arpackSlv = pyarpackSlv.double() # Caution: double <=> np.array(..., dtype='float64')
+arpackSlv = pyarpackSlv.complexDouble() # Caution: complexDouble <=> np.array(..., dtype='complex128')
 arpackSlv.verbose = 3 # Set to 0 to get a quiet solve.
 arpackSlv.debug = 1 # Set to 0 to get a quiet solve.
 arpackSlv.nbEV = 1
@@ -29,6 +31,7 @@ arpackSlv.nbCV = 2*arpackSlv.nbEV + 1
 arpackSlv.mag = 'LM'
 arpackSlv.maxIt = 200
 arpackSlv.slvPvtThd = 1.e-6
+arpackSlv.symPb = False

 # Solve eigen problem.

@@ -56,19 +59,22 @@ print("\n#######################################################################
 # Build laplacian.

 n = 8
-Aij = np.array([], dtype='float32')
-Bij = np.array([], dtype='float32')
+Aij = np.array([], dtype='complex64')
+Bij = np.array([], dtype='complex64')
 for k in range(n):
    for l in range(n):
        if l == k:
-            Aij = np.append(Aij, np.float32( 200.)) # Casting value on append is MANDATORY or C++ won't get the expected type.
-            Bij = np.append(Bij, np.float32( 33.3)) # Casting value on append is MANDATORY or C++ won't get the expected type.
-        elif l == k-1 or l == k+1:
-            Aij = np.append(Aij, np.float32(-100.)) # Casting value on append is MANDATORY or C++ won't get the expected type.
-            Bij = np.append(Bij, np.float32( 16.6)) # Casting value on append is MANDATORY or C++ won't get the expected type.
+            Aij = np.append(Aij, np.complex64(complex( 200.,  200.))) # Casting value on append is MANDATORY or C++ won't get the expected type.
+            Bij = np.append(Bij, np.complex64(complex( 33.3,  33.3))) # Casting value on append is MANDATORY or C++ won't get the expected type.
+        elif l == k-1:
+            Aij = np.append(Aij, np.complex64(complex(-101., -101.))) # Casting value on append is MANDATORY or C++ won't get the expected type.
+            Bij = np.append(Bij, np.complex64(complex( 16.6,  16.6))) # Casting value on append is MANDATORY or C++ won't get the expected type.
+        elif l == k+1:
+            Aij = np.append(Aij, np.complex64(complex( -99.,  -99.))) # Casting value on append is MANDATORY or C++ won't get the expected type.
+            Bij = np.append(Bij, np.complex64(complex( 16.6,  16.6))) # Casting value on append is MANDATORY or C++ won't get the expected type.
        else:
-            Aij = np.append(Aij, np.float32(   0.)) # Casting value on append is MANDATORY or C++ won't get the expected type.
-            Bij = np.append(Bij, np.float32(   0.)) # Casting value on append is MANDATORY or C++ won't get the expected type.
+            Aij = np.append(Aij, np.complex64(complex(   0.,    0.))) # Casting value on append is MANDATORY or C++ won't get the expected type.
+            Bij = np.append(Bij, np.complex64(complex(   0.,    0.))) # Casting value on append is MANDATORY or C++ won't get the expected type.
 for idx, val in enumerate(Aij):
    print("A[", idx, "] =", val)
 for idx, val in enumerate(Bij):
@@ -78,7 +84,7 @@ B = (Bij,  True) # raw format: Bij values, row ordered (or not).

 # Get and tune arpack solver.

-arpackSlv = pyarpackSlv.float() # Caution: float <=> np.array(..., dtype='float32')
+arpackSlv = pyarpackSlv.complexFloat() # Caution: complexFloat <=> np.array(..., dtype='complex64')
 arpackSlv.verbose = 3 # Set to 0 to get a quiet solve.
 arpackSlv.debug = 1 # Set to 0 to get a quiet solve.
 arpackSlv.nbEV = 2
@@ -87,6 +93,7 @@ arpackSlv.mag = 'LM'
 arpackSlv.maxIt = 200
 arpackSlv.slvPvtThd = 1.e-6
 arpackSlv.sigmaReal = 1
+arpackSlv.symPb = False

 # Solve eigen problem.

@@ -44,7 +44,7 @@ class pyarpackSparseDrtSolver: public arpackDrtSolver<RC, FD, EM, SLV> {
      return arpackDrtSolver<RC, FD, EM, SLV>::solve(M, (stdPb ? NULL : &N));
    };

-    int checkEigVec(bp::tuple const & A, bp::tuple const B = bp::tuple(), double const diffTol = 1.e-3) {
+    int checkEigVec(bp::tuple const & A, bp::tuple const B = bp::tuple(), double const maxResNorm = 1.e-3) {
      ARPACKSOLVERDEBUGSTAT();
      EM M;
      int rc = pyarpackServices<RC, EM>::buildSparseMatrice(A, M, debug, "A");
@@ -55,7 +55,7 @@ class pyarpackSparseDrtSolver: public arpackDrtSolver<RC, FD, EM, SLV> {
        rc = pyarpackServices<RC, EM>::buildSparseMatrice(B, N, debug, "B");
        if (rc != 0) {pyarpackThrowError("build matrice from B KO"); return rc;}
      }
-      return arpackDrtSolver<RC, FD, EM, SLV>::checkEigVec(M, (stdPb ? NULL : &N), &diffTol);
+      return arpackDrtSolver<RC, FD, EM, SLV>::checkEigVec(M, (stdPb ? NULL : &N), maxResNorm);
    };

  // Public members.
@@ -101,7 +101,7 @@ class pyarpackDenseDrtSolver: public arpackDrtSolver<RC, FD, EM, SLV> {
      return arpackDrtSolver<RC, FD, EM, SLV>::solve(M, (stdPb ? NULL : &N));
    };

-    int checkEigVec(bp::tuple const & A, bp::tuple const B = bp::tuple(), double const diffTol = 1.e-3) {
+    int checkEigVec(bp::tuple const & A, bp::tuple const B = bp::tuple(), double const maxResNorm = 1.e-3) {
      ARPACKSOLVERDEBUGSTAT();
      EM M;
      int rc = pyarpackServices<RC, EM>::buildDenseMatrice(A, M, debug, "A");
@@ -112,7 +112,7 @@ class pyarpackDenseDrtSolver: public arpackDrtSolver<RC, FD, EM, SLV> {
        rc = pyarpackServices<RC, EM>::buildDenseMatrice(B, N, debug, "B");
        if (rc != 0) {pyarpackThrowError("build matrice from B KO"); return rc;}
      }
-      return arpackDrtSolver<RC, FD, EM, SLV>::checkEigVec(M, (stdPb ? NULL : &N), &diffTol);
+      return arpackDrtSolver<RC, FD, EM, SLV>::checkEigVec(M, (stdPb ? NULL : &N), maxResNorm);
    };

  // Public members.
@@ -43,7 +43,7 @@ class pyarpackSparseItrSolver: public arpackItrSolver<RC, FD, EM, SLV> {
      return arpackItrSolver<RC, FD, EM, SLV>::solve(M, (stdPb ? NULL : &N));
    };

-    int checkEigVec(bp::tuple const & A, bp::tuple const B = bp::tuple(), double const diffTol = 1.e-3) {
+    int checkEigVec(bp::tuple const & A, bp::tuple const B = bp::tuple(), double const maxResNorm = 1.e-3) {
      ARPACKSOLVERDEBUGSTAT();
      EM M;
      int rc = pyarpackServices<RC, EM>::buildSparseMatrice(A, M, debug, "A");
@@ -54,7 +54,7 @@ class pyarpackSparseItrSolver: public arpackItrSolver<RC, FD, EM, SLV> {
        rc = pyarpackServices<RC, EM>::buildSparseMatrice(B, N, debug, "B");
        if (rc != 0) {pyarpackThrowError("build matrice from B KO"); return rc;}
      }
-      return arpackItrSolver<RC, FD, EM, SLV>::checkEigVec(M, (stdPb ? NULL : &N), &diffTol);
+      return arpackItrSolver<RC, FD, EM, SLV>::checkEigVec(M, (stdPb ? NULL : &N), maxResNorm);
    };

  // Public members.
@@ -18,9 +18,9 @@ for k in range(n):
        if l == k:
            Aij = np.append(Aij, np.complex128(complex( 200.,  200.))) # Casting value on append is MANDATORY or C++ won't get the expected type.
        if l == k-1:
-            Aij = np.append(Aij, np.complex128(complex(-101., -101.))) # Casting value on append is MANDATORY or C++ won't get the expected type.
+            Aij = np.append(Aij, np.complex128(complex(-100., -100.))) # Casting value on append is MANDATORY or C++ won't get the expected type.
        if l == k+1:
-            Aij = np.append(Aij, np.complex128(complex( -99.,  -99.))) # Casting value on append is MANDATORY or C++ won't get the expected type.
+            Aij = np.append(Aij, np.complex128(complex( -100.,  100.))) # Casting value on append is MANDATORY or C++ won't get the expected type.
 for k, l, Akl in zip(i, j, Aij):
    print("A[", k, ",", l, "] =", Akl)
 A = (n, i, j, Aij) # coo format: dimension, i 0-based indices, j 0-based indices, Aij values.
@@ -69,7 +69,6 @@ n = 8
 i = np.array([], dtype='@PYINT@')
 j = np.array([], dtype='@PYINT@')
 Aij = np.array([], dtype='complex64')
-Bij = np.array([], dtype='complex64')
 for k in range(n):
    for l in [k-1, k, k+1]:
        if l < 0 or l > n-1:
@@ -78,19 +77,13 @@ for k in range(n):
        j = np.append(j, np.@PYINT@(l+1)) # Casting value on append is MANDATORY or C++ won't get the expected type.
        if l == k:
            Aij = np.append(Aij, np.complex64(complex( 200.,  200.))) # Casting value on append is MANDATORY or C++ won't get the expected type.
-            Bij = np.append(Bij, np.complex64(complex( 33.3,  33.3))) # Casting value on append is MANDATORY or C++ won't get the expected type.
        if l == k-1:
-            Aij = np.append(Aij, np.complex64(complex(-101., -101.))) # Casting value on append is MANDATORY or C++ won't get the expected type.
-            Bij = np.append(Bij, np.complex64(complex( 16.6,  16.6))) # Casting value on append is MANDATORY or C++ won't get the expected type.
+            Aij = np.append(Aij, np.complex64(complex(-100., -100.))) # Casting value on append is MANDATORY or C++ won't get the expected type.
        if l == k+1:
-            Aij = np.append(Aij, np.complex64(complex( -99.,  -99.))) # Casting value on append is MANDATORY or C++ won't get the expected type.
-            Bij = np.append(Bij, np.complex64(complex( 16.6,  16.6))) # Casting value on append is MANDATORY or C++ won't get the expected type.
+            Aij = np.append(Aij, np.complex64(complex( -100.,  100.))) # Casting value on append is MANDATORY or C++ won't get the expected type.
 for k, l, Akl in zip(i, j, Aij):
    print("A[", k, ",", l, "] =", Akl)
-for k, l, Bkl in zip(i, j, Bij):
-    print("B[", k, ",", l, "] =", Bkl)
 A = (n, i, j, Aij) # coo format: dimension, i 1-based indices, j 1-based indices, Aij values.
-B = (n, i, j, Bij) # coo format: dimension, i 1-based indices, j 1-based indices, Bij values.

 # Get and tune arpack solver.

@@ -111,9 +104,9 @@ arpackSlv.symPb = False

 # Solve eigen problem.

-rc = arpackSlv.solve(A, B)
+rc = arpackSlv.solve(A)
 assert rc == 0, "bad solve"
-rc = arpackSlv.checkEigVec(A, B, 1.)
+rc = arpackSlv.checkEigVec(A)
 assert rc == 0, "bad checkEigVec"

 # Print out results (mode selected, eigen vectors, eigen values, ...).
@@ -8,7 +8,7 @@ from pyarpack import sparseLDLT as pyarpackSlv
 n = 4
 i = np.array([], dtype='@PYINT@')
 j = np.array([], dtype='@PYINT@')
-Aij = np.array([], dtype='complex128')
+Aij = np.array([], dtype='float64')
 for k in range(n):
    for l in [k-1, k, k+1]:
        if l < 0 or l > n-1:
@@ -16,18 +16,16 @@ for k in range(n):
        i = np.append(i, np.@PYINT@(k)) # Casting value on append is MANDATORY or C++ won't get the expected type.
        j = np.append(j, np.@PYINT@(l)) # Casting value on append is MANDATORY or C++ won't get the expected type.
        if l == k:
-            Aij = np.append(Aij, np.complex128(complex( 200.,  200.))) # Casting value on append is MANDATORY or C++ won't get the expected type.
-        if l == k-1:
-            Aij = np.append(Aij, np.complex128(complex(-101., -101.))) # Casting value on append is MANDATORY or C++ won't get the expected type.
-        if l == k+1:
-            Aij = np.append(Aij, np.complex128(complex( -99.,  -99.))) # Casting value on append is MANDATORY or C++ won't get the expected type.
+            Aij = np.append(Aij, np.float64( 200.)) # Casting value on append is MANDATORY or C++ won't get the expected type.
+        if l == k-1 or l == k+1:
+            Aij = np.append(Aij, np.float64(-100.)) # Casting value on append is MANDATORY or C++ won't get the expected type.
 for k, l, Akl in zip(i, j, Aij):
    print("A[", k, ",", l, "] =", Akl)
 A = (n, i, j, Aij) # coo format: dimension, i 0-based indices, j 0-based indices, Aij values.

 # Get and tune arpack solver.

-arpackSlv = pyarpackSlv.complexDouble() # Caution: complexDouble <=> np.array(..., dtype='complex128')
+arpackSlv = pyarpackSlv.double() # Caution: double <=> np.array(..., dtype='float64')
 arpackSlv.verbose = 3 # Set to 0 to get a quiet solve.
 arpackSlv.debug = 1 # Set to 0 to get a quiet solve.
 arpackSlv.nbEV = 1
@@ -36,7 +34,7 @@ arpackSlv.mag = 'LM'
 arpackSlv.maxIt = 200
 arpackSlv.slvOffset = 0.
 arpackSlv.slvScale = 1.
-arpackSlv.symPb = False
+arpackSlv.symPb = True

 # Solve eigen problem.

@@ -66,8 +64,8 @@ print("\n#######################################################################
 n = 8
 i = np.array([], dtype='@PYINT@')
 j = np.array([], dtype='@PYINT@')
-Aij = np.array([], dtype='complex64')
-Bij = np.array([], dtype='complex64')
+Aij = np.array([], dtype='float32')
+Bij = np.array([], dtype='float32')
 for k in range(n):
    for l in [k-1, k, k+1]:
        if l < 0 or l > n-1:
@@ -75,14 +73,11 @@ for k in range(n):
        i = np.append(i, np.@PYINT@(k+1)) # Casting value on append is MANDATORY or C++ won't get the expected type.
        j = np.append(j, np.@PYINT@(l+1)) # Casting value on append is MANDATORY or C++ won't get the expected type.
        if l == k:
-            Aij = np.append(Aij, np.complex64(complex( 200.,  200.))) # Casting value on append is MANDATORY or C++ won't get the expected type.
-            Bij = np.append(Bij, np.complex64(complex( 33.3,  33.3))) # Casting value on append is MANDATORY or C++ won't get the expected type.
-        if l == k-1:
-            Aij = np.append(Aij, np.complex64(complex(-101., -101.))) # Casting value on append is MANDATORY or C++ won't get the expected type.
-            Bij = np.append(Bij, np.complex64(complex( 16.6,  16.6))) # Casting value on append is MANDATORY or C++ won't get the expected type.
-        if l == k+1:
-            Aij = np.append(Aij, np.complex64(complex( -99.,  -99.))) # Casting value on append is MANDATORY or C++ won't get the expected type.
-            Bij = np.append(Bij, np.complex64(complex( 16.6,  16.6))) # Casting value on append is MANDATORY or C++ won't get the expected type.
+            Aij = np.append(Aij, np.float32( 200.)) # Casting value on append is MANDATORY or C++ won't get the expected type.
+            Bij = np.append(Bij, np.float32( 33.3)) # Casting value on append is MANDATORY or C++ won't get the expected type.
+        if l == k-1 or l == k+1:
+            Aij = np.append(Aij, np.float32(-100.)) # Casting value on append is MANDATORY or C++ won't get the expected type.
+            Bij = np.append(Bij, np.float32( 16.6)) # Casting value on append is MANDATORY or C++ won't get the expected type.
 for k, l, Akl in zip(i, j, Aij):
    print("A[", k, ",", l, "] =", Akl)
 for k, l, Bkl in zip(i, j, Bij):
@@ -92,7 +87,7 @@ B = (n, i, j, Bij) # coo format: dimension, i 1-based indices, j 1-based indices

 # Get and tune arpack solver.

-arpackSlv = pyarpackSlv.complexFloat() # Caution: complexFloat <=> np.array(..., dtype='complex64')
+arpackSlv = pyarpackSlv.float() # Caution: float <=> np.array(..., dtype='float32')
 arpackSlv.verbose = 3 # Set to 0 to get a quiet solve.
 arpackSlv.debug = 1 # Set to 0 to get a quiet solve.
 arpackSlv.nbEV = 2
@@ -102,8 +97,7 @@ arpackSlv.maxIt = 200
 arpackSlv.slvOffset = 0.
 arpackSlv.slvScale = 1.
 arpackSlv.sigmaReal = 1
-arpackSlv.sigmaImag = 1
-arpackSlv.symPb = False
+arpackSlv.symPb = True

 # Solve eigen problem.

@@ -2,13 +2,12 @@

 import numpy as np
 from pyarpack import sparseLU as pyarpackSlv
-
 # Build laplacian.

 n = 4
 i = np.array([], dtype='@PYINT@')
 j = np.array([], dtype='@PYINT@')
-Aij = np.array([], dtype='float64')
+Aij = np.array([], dtype='complex128')
 for k in range(n):
    for l in [k-1, k, k+1]:
        if l < 0 or l > n-1:
@@ -16,16 +15,18 @@ for k in range(n):
        i = np.append(i, np.@PYINT@(k)) # Casting value on append is MANDATORY or C++ won't get the expected type.
        j = np.append(j, np.@PYINT@(l)) # Casting value on append is MANDATORY or C++ won't get the expected type.
        if l == k:
-            Aij = np.append(Aij, np.float64( 200.)) # Casting value on append is MANDATORY or C++ won't get the expected type.
-        if l == k-1 or l == k+1:
-            Aij = np.append(Aij, np.float64(-100.)) # Casting value on append is MANDATORY or C++ won't get the expected type.
+            Aij = np.append(Aij, np.complex128(complex( 200.,  200.))) # Casting value on append is MANDATORY or C++ won't get the expected type.
+        if l == k-1:
+            Aij = np.append(Aij, np.complex128(complex(-101., -101.))) # Casting value on append is MANDATORY or C++ won't get the expected type.
+        if l == k+1:
+            Aij = np.append(Aij, np.complex128(complex( -99.,  -99.))) # Casting value on append is MANDATORY or C++ won't get the expected type.
 for k, l, Akl in zip(i, j, Aij):
    print("A[", k, ",", l, "] =", Akl)
 A = (n, i, j, Aij) # coo format: dimension, i 0-based indices, j 0-based indices, Aij values.

 # Get and tune arpack solver.

-arpackSlv = pyarpackSlv.double() # Caution: double <=> np.array(..., dtype='float64')
+arpackSlv = pyarpackSlv.complexDouble() # Caution: complexDouble <=> np.array(..., dtype='complex128')
 arpackSlv.verbose = 3 # Set to 0 to get a quiet solve.
 arpackSlv.debug = 1 # Set to 0 to get a quiet solve.
 arpackSlv.nbEV = 1
@@ -33,6 +34,7 @@ arpackSlv.nbCV = 2*arpackSlv.nbEV + 1
 arpackSlv.mag = 'LM'
 arpackSlv.maxIt = 200
 arpackSlv.slvPvtThd = 1.e-6
+arpackSlv.symPb = False

 # Solve eigen problem.

@@ -62,8 +64,8 @@ print("\n#######################################################################
 n = 8
 i = np.array([], dtype='@PYINT@')
 j = np.array([], dtype='@PYINT@')
-Aij = np.array([], dtype='float32')
-Bij = np.array([], dtype='float32')
+Aij = np.array([], dtype='complex64')
+Bij = np.array([], dtype='complex64')
 for k in range(n):
    for l in [k-1, k, k+1]:
        if l < 0 or l > n-1:
@@ -71,11 +73,14 @@ for k in range(n):
        i = np.append(i, np.@PYINT@(k+1)) # Casting value on append is MANDATORY or C++ won't get the expected type.
        j = np.append(j, np.@PYINT@(l+1)) # Casting value on append is MANDATORY or C++ won't get the expected type.
        if l == k:
-            Aij = np.append(Aij, np.float32( 200.)) # Casting value on append is MANDATORY or C++ won't get the expected type.
-            Bij = np.append(Bij, np.float32( 33.3)) # Casting value on append is MANDATORY or C++ won't get the expected type.
-        if l == k-1 or l == k+1:
-            Aij = np.append(Aij, np.float32(-100.)) # Casting value on append is MANDATORY or C++ won't get the expected type.
-            Bij = np.append(Bij, np.float32( 16.6)) # Casting value on append is MANDATORY or C++ won't get the expected type.
+            Aij = np.append(Aij, np.complex64(complex( 200.,  200.))) # Casting value on append is MANDATORY or C++ won't get the expected type.
+            Bij = np.append(Bij, np.complex64(complex( 33.3,  33.3))) # Casting value on append is MANDATORY or C++ won't get the expected type.
+        if l == k-1:
+            Aij = np.append(Aij, np.complex64(complex(-101., -101.))) # Casting value on append is MANDATORY or C++ won't get the expected type.
+            Bij = np.append(Bij, np.complex64(complex( 16.6,  16.6))) # Casting value on append is MANDATORY or C++ won't get the expected type.
+        if l == k+1:
+            Aij = np.append(Aij, np.complex64(complex( -99.,  -99.))) # Casting value on append is MANDATORY or C++ won't get the expected type.
+            Bij = np.append(Bij, np.complex64(complex( 16.6,  16.6))) # Casting value on append is MANDATORY or C++ won't get the expected type.
 for k, l, Akl in zip(i, j, Aij):
    print("A[", k, ",", l, "] =", Akl)
 for k, l, Bkl in zip(i, j, Bij):
@@ -85,7 +90,7 @@ B = (n, i, j, Bij) # coo format: dimension, i 1-based indices, j 1-based indices

 # Get and tune arpack solver.

-arpackSlv = pyarpackSlv.float() # Caution: float <=> np.array(..., dtype='float32')
+arpackSlv = pyarpackSlv.complexFloat() # Caution: complexFloat <=> np.array(..., dtype='complex64')
 arpackSlv.verbose = 3 # Set to 0 to get a quiet solve.
 arpackSlv.debug = 1 # Set to 0 to get a quiet solve.
 arpackSlv.nbEV = 2
@@ -94,6 +99,8 @@ arpackSlv.mag = 'LM'
 arpackSlv.maxIt = 200
 arpackSlv.slvPvtThd = 1.e-6
 arpackSlv.sigmaReal = 1
+arpackSlv.sigmaImag = 1
+arpackSlv.symPb = False

 # Solve eigen problem.

@@ -1,5 +1,6 @@
 AM_CPPFLAGS = -I$(top_builddir) -I$(top_srcdir)

+pkgincludedir = $(includedir)/arpack@ITF64SUFFIX@
 pkginclude_HEADERS = debug_c.h debug_c.hpp
 pkginclude_HEADERS += stat_c.h stat_c.hpp
 pkginclude_HEADERS += arpack.h arpack.hpp
@@ -1,6 +1,6 @@
 ACLOCAL_AMFLAGS = -I m4

-pkgincludedir = $(includedir)/arpack-ng@ITF64SUFFIX@
+pkgincludedir = $(includedir)/arpack@ITF64SUFFIX@
 pkginclude_HEADERS = debug.h stat.h debugF90.h statF90.h arpackdef.h

 AM_DISTCHECK_CONFIGURE_FLAGS =
@@ -11,8 +11,8 @@ if ICB
 AM_DISTCHECK_CONFIGURE_FLAGS += --enable-icb
 pkginclude_HEADERS += arpackicb.h
 endif
-if ICBEXMM
-AM_DISTCHECK_CONFIGURE_FLAGS += --enable-icb-exmm
+if EIGEN
+AM_DISTCHECK_CONFIGURE_FLAGS += --enable-eigen
 endif

 SUBDIRS = .
@@ -27,7 +27,6 @@ endif
 EXTRA_DIST = README.md PARPACK_CHANGES CHANGES DOCUMENTS VISUAL_STUDIO \
 detect_arpack_bug.m4 CMakeLists.txt

-pkgconfig_DATA = arpack@LIBSUFFIX@@ITF64SUFFIX@.pc parpack@LIBSUFFIX@@ITF64SUFFIX@.pc arpackSolver@LIBSUFFIX@@ITF64SUFFIX@.pc
-
-# Due to the LIBSUFFIX/ITF64SUFFIX, configure doesn't automatically clean this file:
-DISTCLEANFILES = arpack@LIBSUFFIX@@ITF64SUFFIX@.pc parpack@LIBSUFFIX@@ITF64SUFFIX@.pc arpackSolver@LIBSUFFIX@@ITF64SUFFIX@.pc
+cmakedir = $(libdir)/cmake/arpack-ng
+cmake_DATA = cmake/arpackng-config-version.cmake  \
+  cmake/arpackng-config.cmake
@@ -114,7 +114,7 @@ c     %------------------%
 c     | Scalar Arguments |
 c     %------------------%
 c
-      character*2 which
+      character  which*2
      integer    ishift, kev, np
 c
 c     %-----------------%
@@ -120,7 +120,7 @@ c     %------------------%
 c     | Scalar Arguments |
 c     %------------------%
 c
-      character*2 which
+      character  which*2
      integer    ishift, kev, np
 c
 c     %-----------------%
@@ -117,7 +117,7 @@ c     %------------------%
 c     | Scalar Arguments |
 c     %------------------%
 c
-      character*2 which
+      character  which*2
      integer    ishift, kev, np
 c
 c     %-----------------%
@@ -120,7 +120,7 @@ c     %------------------%
 c     | Scalar Arguments |
 c     %------------------%
 c
-      character*2 which
+      character  which*2
      integer    ishift, kev, np
 c
 c     %-----------------%
@@ -117,7 +117,7 @@ c     %------------------%
 c     | Scalar Arguments |
 c     %------------------%
 c
-      character*2 which
+      character  which*2
      integer    ishift, kev, np
 c
 c     %-----------------%
@@ -114,7 +114,7 @@ c     %------------------%
 c     | Scalar Arguments |
 c     %------------------%
 c
-      character*2 which
+      character  which*2
      integer    ishift, kev, np
 c
 c     %-----------------%
@@ -1,17 +1,15 @@
 AUTOMAKE_OPTIONS = subdir-objects # Needed as debug/stat* are not in current directory.

 F77 = $(MPIF77)
-FFLAGS_SAV = @FFLAGS@
-FFLAGS =

 PSRC = pcontext.f
 SSRC = psnaitr.f psnapps.f psnaup2.f psnaupd.f psneigh.f psneupd.f psngets.f \
       pssaitr.f pssapps.f pssaup2.f pssaupd.f psseigt.f psseupd.f pssgets.f \
-       psgetv0.f pslarnv.f psnorm2.f
+       psgetv0.f pslarnv.f psnorm2.f pslamch10.f

 DSRC = pdnaitr.f pdnapps.f pdnaup2.f pdnaupd.f pdneigh.f pdneupd.f pdngets.f \
       pdsaitr.f pdsapps.f pdsaup2.f pdsaupd.f pdseigt.f pdseupd.f pdsgets.f \
-       pdgetv0.f pdlarnv.f pdnorm2.f
+       pdgetv0.f pdlarnv.f pdnorm2.f pdlamch10.f

 CSRC = pcnaitr.f pcnapps.f pcnaup2.f pcnaupd.f pcneigh.f pcneupd.f pcngets.f \
       pcgetv0.f pclarnv.f pscnorm2.f
@@ -28,16 +26,11 @@ endif

 EXTRA_DIST = debug.h stat.h pcontext.h

-noinst_LTLIBRARIES = libparpack@LIBSUFFIX@@ITF64SUFFIX@_noopt.la
-libparpack@LIBSUFFIX@@ITF64SUFFIX@_noopt_la_SOURCES = pslamch10.f pdlamch10.f
-libparpack@LIBSUFFIX@@ITF64SUFFIX@_noopt_la_FFLAGS = -O0
-
 lib_LTLIBRARIES = libparpack@LIBSUFFIX@@ITF64SUFFIX@.la
 libparpack@LIBSUFFIX@@ITF64SUFFIX@_la_SOURCES = $(PSRC) $(SSRC) $(DSRC) $(CSRC) $(ZSRC)
 libparpack@LIBSUFFIX@@ITF64SUFFIX@_la_SOURCES += $(top_builddir)/dbgini.f
 libparpack@LIBSUFFIX@@ITF64SUFFIX@_la_SOURCES += $(top_builddir)/staini.f
-libparpack@LIBSUFFIX@@ITF64SUFFIX@_la_FFLAGS = $(FFLAGS_SAV)
-libparpack@LIBSUFFIX@@ITF64SUFFIX@_la_LIBADD = libparpack@LIBSUFFIX@@ITF64SUFFIX@_noopt.la \
+libparpack@LIBSUFFIX@@ITF64SUFFIX@_la_LIBADD = \
 	$(top_builddir)/PARPACK/UTIL/MPI/libparpackutil.la \
 	$(top_builddir)/SRC/libarpack@LIBSUFFIX@@ITF64SUFFIX@.la \
 	$(LAPACK_LIBS) $(BLAS_LIBS) $(MPI_Fortran_LIBS)
@@ -46,3 +39,8 @@ if ICB
 libparpack@LIBSUFFIX@@ITF64SUFFIX@_la_LIBADD += $(top_builddir)/ICB/libdbgicb.la $(top_builddir)/ICB/libstaicb.la
 libparpack@LIBSUFFIX@@ITF64SUFFIX@_la_CPPFLAGS = $(AM_CPPFLAGS) -I$(top_builddir)
 endif
+
+pkgconfig_DATA = parpack@LIBSUFFIX@@ITF64SUFFIX@.pc
+
+# Due to the LIBSUFFIX/ITF64SUFFIX, configure doesn't automatically clean this file:
+DISTCLEANFILES = parpack@LIBSUFFIX@@ITF64SUFFIX@.pc
@@ -1,7 +1,7 @@
 prefix=@prefix@
 exec_prefix=@exec_prefix@
 libdir=@libdir@
-includedir=@includedir@/arpack-ng@ITF64SUFFIX@
+includedir=@includedir@

 Name: @PACKAGE_NAME@
 Description: Collection of Fortran77 subroutines designed to solve large scale eigenvalue problems
@@ -10,4 +10,4 @@ URL: @PACKAGE_URL@
 Requires.private: arpack@LIBSUFFIX@@ITF64SUFFIX@
 Libs: -L${libdir} -lparpack@LIBSUFFIX@@ITF64SUFFIX@
 Libs.private: @PARPACK_PC_LIBS_PRIVATE@
-Cflags: -I${includedir}
+Cflags: -I${includedir}/arpack@ITF64SUFFIX@
@@ -176,13 +176,13 @@ c     %------------------------%
 c     | Local Scalars & Arrays |
 c     %------------------------%
 c
-      logical    first, orth
+      logical    first, inits, orth
      integer    idist, iseed(4), iter, msglvl, jj, myid, igen
      Real
     &           rnorm0
      Complex
     &           cnorm, cnorm2
-      save       first, iseed, iter, msglvl, orth, rnorm0
+      save       first, iseed, inits, iter, msglvl, orth, rnorm0
 c
      Complex
     &           cnorm_buf, buf2(1)
@@ -203,6 +203,12 @@ c
     &           ccdotc
      external   ccdotc, pscnorm2, slapy2
 c
+c     %-----------------%
+c     | Data Statements |
+c     %-----------------%
+c
+      data       inits /.true./
+c
 c     %-----------------------%
 c     | Executable Statements |
 c     %-----------------------%
@@ -213,26 +219,30 @@ c     | Initialize the seed of the LAPACK |
 c     | random number generator           |
 c     %-----------------------------------%
 c
+      if (inits) then
 c
-c     %-----------------------------------%
-c     | Generate a seed on each processor |
-c     | using process id (myid).          |
-c     | Note: the seed must be between 1  |
-c     | and 4095.  iseed(4) must be odd.  |
-c     %-----------------------------------%
+c        %-----------------------------------%
+c        | Generate a seed on each processor |
+c        | using process id (myid).          |
+c        | Note: the seed must be between 1  |
+c        | and 4095.  iseed(4) must be odd.  |
+c        %-----------------------------------%
 c
-      call MPI_COMM_RANK(comm, myid, ierr)
-      igen = 1000 + 2*myid + 1
-      if (igen .gt. 4095) then
-         write(0,*) 'Error in p_getv0: seed exceeds 4095!'
+         call MPI_COMM_RANK(comm, myid, ierr)
+         igen = 1000 + 2*myid + 1
+         if (igen .gt. 4095) then
+            write(0,*) 'Error in p_getv0: seed exceeds 4095!'
+         end if
+c
+         iseed(1) = igen/1000
+         igen     = mod(igen,1000)
+         iseed(2) = igen/100
+         igen     = mod(igen,100)
+         iseed(3) = igen/10
+         iseed(4) = mod(igen,10)
+c
+         inits = .false.
      end if
-c
-      iseed(1) = igen/1000
-      igen     = mod(igen,1000)
-      iseed(2) = igen/100
-      igen     = mod(igen,100)
-      iseed(3) = igen/10
-      iseed(4) = 7
 c
      if (ido .eq.  0) then
 c
@@ -114,7 +114,7 @@ c     %------------------%
 c     | Scalar Arguments |
 c     %------------------%
 c
-      character*2 which
+      character  which*2
      integer    ishift, kev, np
 c
 c     %-----------------%
@@ -177,11 +177,11 @@ c     %------------------------%
 c     | Local Scalars & Arrays |
 c     %------------------------%
 c
-      logical    first, orth
-      integer    idist, iseed(4), iter, msglvl, jj
+      logical    first, inits, orth
+      integer    idist, iseed(4), iter, msglvl, jj, myid, igen
      Double precision
     &           rnorm0, buf2(1)
-      save       first, iseed, iter, msglvl, orth, rnorm0
+      save       first, iseed, inits, iter, msglvl, orth, rnorm0
 c
      Double precision
     &           rnorm_buf
@@ -206,6 +206,12 @@ c     %---------------------%
 c
      intrinsic    abs, sqrt
 c
+c     %-----------------%
+c     | Data Statements |
+c     %-----------------%
+c
+      data       inits /.true./
+c
 c     %-----------------------%
 c     | Executable Statements |
 c     %-----------------------%
@@ -216,10 +222,30 @@ c     | Initialize the seed of the LAPACK |
 c     | random number generator           |
 c     %-----------------------------------%
 c
-      iseed(1) = 1
-      iseed(2) = 3
-      iseed(3) = 5
-      iseed(4) = 7
+      if (inits) then
+c
+c        %-----------------------------------%
+c        | Generate a seed on each processor |
+c        | using process id (myid).          |
+c        | Note: the seed must be between 1  |
+c        | and 4095.  iseed(4) must be odd.  |
+c        %-----------------------------------%
+c
+         call MPI_COMM_RANK(comm, myid, ierr)
+         igen = 1000 + 2*myid + 1
+         if (igen .gt. 4095) then
+            write(0,*) 'Error in p_getv0: seed exceeds 4095!'
+         end if
+c
+         iseed(1) = igen/1000
+         igen     = mod(igen,1000)
+         iseed(2) = igen/100
+         igen     = mod(igen,100)
+         iseed(3) = igen/10
+         iseed(4) = mod(igen,10)
+c
+         inits = .false.
+      end if
 c
      if (ido .eq.  0) then
 c
@@ -331,7 +331,7 @@ c
         call dlabad( unfl, ovfl )
         ulp = pdlamch10( comm, 'precision' )
         smlnum = unfl*( n / ulp )
-         first = .false.
+         aitr_first = .false.
      end if
 c
      if (ido .eq. 0) then
@@ -120,7 +120,7 @@ c     %------------------%
 c     | Scalar Arguments |
 c     %------------------%
 c
-      character*2 which
+      character  which*2
      integer    ishift, kev, np
 c
 c     %-----------------%
@@ -117,7 +117,7 @@ c     %------------------%
 c     | Scalar Arguments |
 c     %------------------%
 c
-      character*2 which
+      character  which*2
      integer    ishift, kev, np
 c
 c     %-----------------%
@@ -177,11 +177,11 @@ c     %------------------------%
 c     | Local Scalars & Arrays |
 c     %------------------------%
 c
-      logical    first, orth
-      integer    idist, iseed(4), iter, msglvl, jj
+      logical    first, inits, orth
+      integer    idist, iseed(4), iter, msglvl, jj, myid, igen
      Real
     &           rnorm0
-      save       first, iseed, iter, msglvl, orth, rnorm0
+      save       first, iseed, inits, iter, msglvl, orth, rnorm0
 c
      Real
     &           rnorm_buf
@@ -206,20 +206,41 @@ c     %---------------------%
 c
      intrinsic    abs, sqrt
 c
+c     %-----------------%
+c     | Data Statements |
+c     %-----------------%
+c
+      data       inits /.true./
+c
 c     %-----------------------%
 c     | Executable Statements |
 c     %-----------------------%
 c
 c
-c     %-----------------------------------%
-c     | Initialize the seed of the LAPACK |
-c     | random number generator           |
-c     %-----------------------------------%
+      if (inits) then
 c
-      iseed(1) = 1
-      iseed(2) = 3
-      iseed(3) = 5
-      iseed(4) = 7
+c        %-----------------------------------%
+c        | Generate a seed on each processor |
+c        | using process id (myid).          |
+c        | Note: the seed must be between 1  |
+c        | and 4095.  iseed(4) must be odd.  |
+c        %-----------------------------------%
+c
+         call MPI_COMM_RANK(comm, myid, ierr)
+         igen = 1000 + 2*myid + 1
+         if (igen .gt. 4095) then
+            write(0,*) 'Error in p_getv0: seed exceeds 4095!'
+         end if
+c
+         iseed(1) = igen/1000
+         igen     = mod(igen,1000)
+         iseed(2) = igen/100
+         igen     = mod(igen,100)
+         iseed(3) = igen/10
+         iseed(4) = mod(igen,10)
+c
+         inits = .false.
+      end if
 c
      if (ido .eq.  0) then
 c
@@ -120,7 +120,7 @@ c     %------------------%
 c     | Scalar Arguments |
 c     %------------------%
 c
-      character*2 which
+      character  which*2
      integer    ishift, kev, np
 c
 c     %-----------------%
@@ -117,7 +117,7 @@ c     %------------------%
 c     | Scalar Arguments |
 c     %------------------%
 c
-      character*2 which
+      character  which*2
      integer    ishift, kev, np
 c
 c     %-----------------%
@@ -176,13 +176,13 @@ c     %------------------------%
 c     | Local Scalars & Arrays |
 c     %------------------------%
 c
-      logical    first, orth
+      logical    first, inits, orth
      integer    idist, iseed(4), iter, msglvl, jj, myid, igen
      Double precision
     &           rnorm0
      Complex*16
     &           cnorm, cnorm2
-      save       first, iseed, iter, msglvl, orth, rnorm0
+      save       first, iseed, inits, iter, msglvl, orth, rnorm0
 c
      Complex*16
     &           cnorm_buf, buf2(1)
@@ -203,6 +203,12 @@ c
     &           zzdotc
      external   zzdotc , pdznorm2 , dlapy2
 c
+c     %-----------------%
+c     | Data Statements |
+c     %-----------------%
+c
+      data       inits /.true./
+c
 c     %-----------------------%
 c     | Executable Statements |
 c     %-----------------------%
@@ -213,26 +219,30 @@ c     | Initialize the seed of the LAPACK |
 c     | random number generator           |
 c     %-----------------------------------%
 c
+      if (inits) then
 c
-c     %-----------------------------------%
-c     | Generate a seed on each processor |
-c     | using process id (myid).          |
-c     | Note: the seed must be between 1  |
-c     | and 4095.  iseed(4) must be odd.  |
-c     %-----------------------------------%
+c        %-----------------------------------%
+c        | Generate a seed on each processor |
+c        | using process id (myid).          |
+c        | Note: the seed must be between 1  |
+c        | and 4095.  iseed(4) must be odd.  |
+c        %-----------------------------------%
 c
-      call MPI_COMM_RANK(comm, myid, ierr)
-      igen = 1000 + 2*myid + 1
-      if (igen .gt. 4095) then
-         write(0,*) 'Error in p_getv0: seed exceeds 4095!'
+         call MPI_COMM_RANK(comm, myid, ierr)
+         igen = 1000 + 2*myid + 1
+         if (igen .gt. 4095) then
+            write(0,*) 'Error in p_getv0: seed exceeds 4095!'
+         end if
+c
+         iseed(1) = igen/1000
+         igen     = mod(igen,1000)
+         iseed(2) = igen/100
+         igen     = mod(igen,100)
+         iseed(3) = igen/10
+         iseed(4) = mod(igen,10)
+c
+         inits = .false.
      end if
-c
-      iseed(1) = igen/1000
-      igen     = mod(igen,1000)
-      iseed(2) = igen/100
-      igen     = mod(igen,100)
-      iseed(3) = igen/10
-      iseed(4) = 7
 c
      if (ido .eq.  0) then
 c
@@ -114,7 +114,7 @@ c     %------------------%
 c     | Scalar Arguments |
 c     %------------------%
 c
-      character*2 which
+      character  which*2
      integer    ishift, kev, np
 c
 c     %-----------------%
@@ -22,9 +22,9 @@
 * with entries 1000, 999, ... , 2, 1 on the diagonal.
 * */

-void dMatVec(const double* x, double* y) {
+void dMatVec(int start, int N, const double* x, double* y) {
  int i;
-  for (i = 0; i < 1000; ++i) y[i] = ((double)(i + 1)) * x[i];
+  for (i = 0; i < N; ++i) y[i] = ((double)(start + i + 1)) * x[i];
 };

 int ds() {
@@ -57,16 +57,24 @@ int ds() {
  char which[]  = "LM";
  char howmny[] = "A";
  
-  int rank;
-  MPI_Comm_rank(MPI_COMM_WORLD, &rank);
  MPI_Fint MCW = MPI_Comm_c2f(MPI_COMM_WORLD);
-  
+  /// Split problem across each process/////////////////////
+  int rank, nprocs;
+  MPI_Comm_rank(MPI_COMM_WORLD, &rank);
+  MPI_Comm_size(MPI_COMM_WORLD, &nprocs);
+  a_int N_local = N / nprocs;
+  if (rank < N % nprocs) // spread the remaining on each process
+    N_local = N_local + 1;
+  printf("rank: %d, size: %d\n", rank, N_local);
+  int start = rank*(N / nprocs) + ((rank < N % nprocs) ? rank : N % nprocs);
+  printf("rank: %d, start: %d\n", rank, start);
+  //////////////////////////////////////////////////////////
  a_int info = 0, ido = 0;
  do {
-    pdsaupd_c(MCW, &ido, bmat, N, which, nev, tol, resid, ncv, V, ldv, iparam,
+    pdsaupd_c(MCW, &ido, bmat, N_local, which, nev, tol, resid, ncv, V, ldv, iparam,
              ipntr, workd, workl, lworkl, &info);

-    dMatVec(&(workd[ipntr[0] - 1]), &(workd[ipntr[1] - 1]));
+    dMatVec(start, N_local, &(workd[ipntr[0] - 1]), &(workd[ipntr[1] - 1]));
  } while (ido == 1 || ido == -1);

  // check info and number of ev found by arpack.
@@ -75,7 +83,7 @@ int ds() {
    return 1;
  }

-  pdseupd_c(MCW, rvec, howmny, select, d, z, ldz, sigma, bmat, N, which, nev,
+  pdseupd_c(MCW, rvec, howmny, select, d, z, ldz, sigma, bmat, N_local, which, nev,
            tol, resid, ncv, V, ldv, iparam, ipntr, workd, workl, lworkl, &info);
  if (info < 0) {
    printf("Error in seupd: info %d\n", info);
@@ -94,9 +102,9 @@ int ds() {
  return 0;
 }

-void zMatVec(const a_dcomplex* x, a_dcomplex* y) {
+void zMatVec(int start, int N, const a_dcomplex* x, a_dcomplex* y) {
  int i;
-  for (i = 0; i < 1000; ++i) y[i] = x[i] * CMPLXF(i + 1.0, i + 1.0);
+  for (i = 0; i < N; ++i) y[i] = x[i] * CMPLXF(start + i + 1.0, start + i + 1.0);
 };

 int zn() {
@@ -131,16 +139,24 @@ int zn() {
  char which[]  = "LM";
  char howmny[] = "A";
  
-  int rank;
-  MPI_Comm_rank(MPI_COMM_WORLD, &rank);
  MPI_Fint MCW = MPI_Comm_c2f(MPI_COMM_WORLD);
-
+  /// Split problem across each process/////////////////////
+  int rank, nprocs;
+  MPI_Comm_rank(MPI_COMM_WORLD, &rank);
+  MPI_Comm_size(MPI_COMM_WORLD, &nprocs);
+  a_int N_local = N / nprocs;
+  if (rank < N % nprocs) // spread the remaining on each process
+    N_local = N_local + 1;
+  printf("rank: %d, size: %d\n", rank, N_local);
+  int start = rank*(N / nprocs) + ((rank < N % nprocs) ? rank : N % nprocs);
+  printf("rank: %d, start: %d\n", rank, start);
+  //////////////////////////////////////////////////////////
  a_int info = 0, ido = 0;
  do {
-    pznaupd_c(MCW, &ido, bmat, N, which, nev, tol, resid, ncv, V, ldv, iparam,
+    pznaupd_c(MCW, &ido, bmat, N_local, which, nev, tol, resid, ncv, V, ldv, iparam,
              ipntr, workd, workl, lworkl, rwork, &info);

-    zMatVec(&(workd[ipntr[0] - 1]), &(workd[ipntr[1] - 1]));
+    zMatVec(start, N_local, &(workd[ipntr[0] - 1]), &(workd[ipntr[1] - 1]));
  } while (ido == 1 || ido == -1);
  
  // check info and number of ev found by arpack.
@@ -149,7 +165,7 @@ int zn() {
    return 1;
  }

-  pzneupd_c(MCW, rvec, howmny, select, d, z, ldz, sigma, workev, bmat, N, which,
+  pzneupd_c(MCW, rvec, howmny, select, d, z, ldz, sigma, workev, bmat, N_local, which,
            nev, tol, resid, ncv, V, ldv, iparam, ipntr, workd, workl, lworkl,
            rwork, &info);
  if (info < 0) {
@@ -13,6 +13,7 @@

 #include <cmath>
 #include <iostream>
+#include <stdexcept>
 #include <vector>

 #include "parpack.hpp"
@@ -20,9 +21,9 @@
 #include "stat_c.hpp"   // arpack statistics.

 template <typename Real>
-void diagonal_matrix_vector_product(const Real* x, Real* y) {
-  for (int i = 0; i < 1000; ++i) {
-    y[i] = static_cast<float>(i + 1) * x[i];
+void diagonal_matrix_vector_product(int start, int nloc, const Real* x, Real* y) {
+  for (int i = 0; i < nloc; ++i) {
+    y[i] = static_cast<Real>(start + i + 1) * x[i];
  }
 }

@@ -53,18 +54,27 @@ void real_symmetric_runner(double const& tol_check, arpack::which const& ritz_op
  iparam[3] = 1;      // NB, only 1 allowed
  iparam[6] = 1;      // mode

-  int rank;
-  MPI_Comm_rank(MPI_COMM_WORLD, &rank);
  MPI_Fint MCW = MPI_Comm_c2f(MPI_COMM_WORLD);
+  /// Split problem across each process/////////////////////
+  int rank, nprocs;
+  MPI_Comm_rank(MPI_COMM_WORLD, &rank);
+  MPI_Comm_size(MPI_COMM_WORLD, &nprocs);
+  a_int N_local = N / nprocs;
+  if (rank < N % nprocs) // spread the remaining on each process
+    N_local = N_local + 1;
+  std::cout<< "rank: " << rank << ", size :" << N_local << "\n";
+  int start = rank*(N / nprocs) + std::min(rank, N % nprocs);
+  std::cout<< "rank: " <<rank<< ", start:" << start<<"\n";
+  //////////////////////////////////////////////////////////

  a_int info = 0, ido = 0;
  do {
-    arpack::saupd(MCW, ido, arpack::bmat::identity, N,
+    arpack::saupd(MCW, ido, arpack::bmat::identity, N_local,
                  ritz_option, nev, tol, resid.data(), ncv,
                  V.data(), ldv, iparam, ipntr, workd.data(),
                  workl.data(), lworkl, info);

-    diagonal_matrix_vector_product(&(workd[ipntr[0] - 1]), &(workd[ipntr[1] - 1]));
+    diagonal_matrix_vector_product(start, N_local, &(workd[ipntr[0] - 1]), &(workd[ipntr[1] - 1]));
  } while  (ido == 1 || ido == -1);

  // check info and number of ev found by arpack.
@@ -75,7 +85,7 @@ void real_symmetric_runner(double const& tol_check, arpack::which const& ritz_op
  }

  arpack::seupd(MCW, rvec, arpack::howmny::ritz_vectors, select.data(),
-                d.data(), z.data(), ldz, sigma, arpack::bmat::identity, N,
+                d.data(), z.data(), ldz, sigma, arpack::bmat::identity, N_local,
                ritz_option, nev, tol, resid.data(), ncv,
                V.data(), ldv, iparam, ipntr, workd.data(),
                workl.data(), lworkl, info);
@@ -94,11 +104,11 @@ void real_symmetric_runner(double const& tol_check, arpack::which const& ritz_op
 }

 template <typename Real>
-void diagonal_matrix_vector_product(const std::complex<Real>* x, std::complex<Real>* y) {
-  for (int i = 0; i < 1000; ++i) {
+void diagonal_matrix_vector_product(int start, int N_local, const std::complex<Real>* x, std::complex<Real>* y) {
+  for (int i = 0; i < N_local; ++i) {
    // Use complex matrix (i, -i) instead of (i, i): this way "largest_magnitude"
    // and "largest_imaginary" options produce different results that can be checked.
-    y[i] = x[i] * std::complex<Real>{Real(i + 1), -Real(i + 1)};
+    y[i] = x[i] * std::complex<Real>{Real(start + i + 1), -Real(start + i + 1)};
  }
 }

@@ -131,18 +141,26 @@ void complex_nonsymmetric_runner(double const& tol_check, arpack::which const& r
  iparam[3] = 1;       // NB, only 1 allowed
  iparam[6] = 1;       // mode

-  int rank;
-  MPI_Comm_rank(MPI_COMM_WORLD, &rank);
  MPI_Fint MCW = MPI_Comm_c2f(MPI_COMM_WORLD);
-
+  /// Split problem across each process/////////////////////
+  int rank, nprocs;
+  MPI_Comm_rank(MPI_COMM_WORLD, &rank);
+  MPI_Comm_size(MPI_COMM_WORLD, &nprocs);
+  a_int N_local = N / nprocs;
+  if (rank < N % nprocs) // spread the remaining on each process
+    N_local = N_local + 1;
+  std::cout<< "rank: "<<rank<<", size :"<<N_local<<"\n";
+  int start =  rank * (N / nprocs) + std::min(rank, N % nprocs);
+  std::cout<< "rank: " << rank << ", start:" << start << "\n";
+  //////////////////////////////////////////////////////////
  a_int info = 0, ido = 0;
  do {
-    arpack::naupd(MCW, ido, arpack::bmat::identity, N,
+    arpack::naupd(MCW, ido, arpack::bmat::identity, N_local,
                  ritz_option, nev, tol, resid.data(), ncv,
                  V.data(), ldv, iparam, ipntr, workd.data(),
                  workl.data(), lworkl, rwork.data(), info);

-    diagonal_matrix_vector_product(&(workd[ipntr[0] - 1]), &(workd[ipntr[1] - 1]));
+    diagonal_matrix_vector_product(start, N_local, &(workd[ipntr[0] - 1]), &(workd[ipntr[1] - 1]));
  } while (ido == 1 || ido == -1);

  // check info and number of ev found by arpack
@@ -154,7 +172,7 @@ void complex_nonsymmetric_runner(double const& tol_check, arpack::which const& r

  arpack::neupd(MCW, rvec, arpack::howmny::ritz_vectors, select.data(),
                d.data(), z.data(), ldz, sigma, workev.data(),
-                arpack::bmat::identity, N, ritz_option,
+                arpack::bmat::identity, N_local, ritz_option,
                nev, tol, resid.data(), ncv, V.data(), ldv, iparam,
                ipntr, workd.data(), workl.data(), lworkl, rwork.data(), info);
  if (info < 0) throw std::runtime_error("Error in neupd, info " + std::to_string(info));
@@ -38,8 +38,8 @@
 *     ...
 *     ... SPECIFICATIONS FOR LOCAL VARIABLES
      INTEGER            I, J, NDIGIT, K1, K2, LLL
-      CHARACTER*1        ICOL( 3 )
-      CHARACTER*80       LINE
+      CHARACTER          ICOL( 3 )
+      CHARACTER          LINE*80
 *     ...
 *     ... SPECIFICATIONS INTRINSICS
      INTRINSIC          MIN
@@ -34,7 +34,7 @@
 *     ...
 *     ... SPECIFICATIONS FOR LOCAL VARIABLES
      INTEGER            I, NDIGIT, K1, K2, LLL
-      CHARACTER*80       LINE
+      CHARACTER          LINE*80
 *     ...
 *     ... FIRST EXECUTABLE STATEMENT
 *
@@ -34,7 +34,7 @@
 *     ...
 *     ... SPECIFICATIONS FOR LOCAL VARIABLES
 *     .. Scalar Arguments ..
-      CHARACTER*( * )    IFMT
+      CHARACTER          IFMT*( * )
      INTEGER            IDIGIT, LDA, LOUT, M, N
 *     ..
 *     .. Array Arguments ..
@@ -42,7 +42,7 @@
     &                   A( LDA, * )
 *     ..
 *     .. Local Scalars ..
-      CHARACTER*80       LINE
+      CHARACTER          LINE*80
      INTEGER            I, J, K1, K2, LLL, NDIGIT
 *     ..
 *     .. Local Arrays ..
@@ -31,7 +31,7 @@
 *     ...
 *     ... SPECIFICATIONS FOR LOCAL VARIABLES
 *     .. Scalar Arguments ..
-      CHARACTER*( * )    IFMT
+      CHARACTER          IFMT*( * )
      INTEGER            IDIGIT, LOUT, N
 *     ..
 *     .. Array Arguments ..
@@ -39,7 +39,7 @@
     &                   SX( * )
 *     ..
 *     .. Local Scalars ..
-      CHARACTER*80       LINE
+      CHARACTER          LINE*80
      INTEGER            I, K1, K2, LLL, NDIGIT
 *     ..
 *     .. Intrinsic Functions ..
@@ -34,7 +34,7 @@
 *     ...
 *     ... SPECIFICATIONS FOR LOCAL VARIABLES
      INTEGER    I, NDIGIT, K1, K2, LLL
-      CHARACTER*80 LINE
+      CHARACTER  LINE*80
 *     ...
 *     ... SPECIFICATIONS INTRINSICS
      INTRINSIC          MIN
@@ -34,7 +34,7 @@
 *     ...
 *     ... SPECIFICATIONS FOR LOCAL VARIABLES
 *     .. Scalar Arguments ..
-      CHARACTER*( * )    IFMT
+      CHARACTER          IFMT*( * )
      INTEGER            IDIGIT, LDA, LOUT, M, N
 *     ..
 *     .. Array Arguments ..
@@ -42,7 +42,7 @@
     &                   A( LDA, * )
 *     ..
 *     .. Local Scalars ..
-      CHARACTER*80       LINE
+      CHARACTER          LINE*80
      INTEGER            I, J, K1, K2, LLL, NDIGIT
 *     ..
 *     .. Local Arrays ..
@@ -31,7 +31,7 @@
 *     ...
 *     ... SPECIFICATIONS FOR LOCAL VARIABLES
 *     .. Scalar Arguments ..
-      CHARACTER*( * )    IFMT
+      CHARACTER          IFMT*( * )
      INTEGER            IDIGIT, LOUT, N
 *     ..
 *     .. Array Arguments ..
@@ -39,7 +39,7 @@
     &                   SX( * )
 *     ..
 *     .. Local Scalars ..
-      CHARACTER*80       LINE
+      CHARACTER          LINE*80
      INTEGER            I, K1, K2, LLL, NDIGIT
 *     ..
 *     .. Intrinsic Functions ..
@@ -38,8 +38,8 @@
 *     ...
 *     ... SPECIFICATIONS FOR LOCAL VARIABLES
      INTEGER            I, J, NDIGIT, K1, K2, LLL
-      CHARACTER*1        ICOL( 3 )
-      CHARACTER*80       LINE
+      CHARACTER          ICOL( 3 )
+      CHARACTER          LINE*80
 *     ...
 *     ... SPECIFICATIONS INTRINSICS
      INTRINSIC          MIN
@@ -34,7 +34,7 @@
 *     ...
 *     ... SPECIFICATIONS FOR LOCAL VARIABLES
      INTEGER            I, NDIGIT, K1, K2, LLL
-      CHARACTER*80       LINE
+      CHARACTER          LINE*80
 *     ...
 *     ... FIRST EXECUTABLE STATEMENT
 *
@@ -37,8 +37,8 @@
 *     ...
 *     ... SPECIFICATIONS FOR LOCAL VARIABLES
      INTEGER            I, J, NDIGIT, K1, K2, LLL
-      CHARACTER*1        ICOL( 3 )
-      CHARACTER*80       LINE
+      CHARACTER          ICOL( 3 )
+      CHARACTER          LINE*80
 *     ...
 *     ... SPECIFICATIONS INTRINSICS
      INTRINSIC          MIN
@@ -33,7 +33,7 @@
 *     ...
 *     ... SPECIFICATIONS FOR LOCAL VARIABLES
      INTEGER            I, NDIGIT, K1, K2, LLL
-      CHARACTER*80       LINE
+      CHARACTER          LINE*80
 *     ...
 *     ... FIRST EXECUTABLE STATEMENT
 *
@@ -33,7 +33,7 @@
 *     ...
 *     ... SPECIFICATIONS FOR LOCAL VARIABLES
 *     .. Scalar Arguments ..
-      CHARACTER*( * )    IFMT
+      CHARACTER          IFMT*( * )
      INTEGER            IDIGIT, LDA, LOUT, M, N
 *     ..
 *     .. Array Arguments ..
@@ -41,7 +41,7 @@
     &                   A( LDA, * )
 *     ..
 *     .. Local Scalars ..
-      CHARACTER*80       LINE
+      CHARACTER          LINE*80
      INTEGER            I, J, K1, K2, LLL, NDIGIT
 *     ..
 *     .. Local Arrays ..
@@ -30,7 +30,7 @@
 *     ...
 *     ... SPECIFICATIONS FOR LOCAL VARIABLES
 *     .. Scalar Arguments ..
-      CHARACTER*( * )    IFMT
+      CHARACTER          IFMT*( * )
      INTEGER            IDIGIT, LOUT, N
 *     ..
 *     .. Array Arguments ..
@@ -38,7 +38,7 @@
     &                   SX( * )
 *     ..
 *     .. Local Scalars ..
-      CHARACTER*80       LINE
+      CHARACTER          LINE*80
      INTEGER            I, K1, K2, LLL, NDIGIT
 *     ..
 *     .. Intrinsic Functions ..
@@ -33,7 +33,7 @@
 *     ...
 *     ... SPECIFICATIONS FOR LOCAL VARIABLES
      INTEGER    I, NDIGIT, K1, K2, LLL
-      CHARACTER*80 LINE
+      CHARACTER  LINE*80
 *     ...
 *     ... SPECIFICATIONS INTRINSICS
      INTRINSIC          MIN
@@ -33,7 +33,7 @@
 *     ...
 *     ... SPECIFICATIONS FOR LOCAL VARIABLES
 *     .. Scalar Arguments ..
-      CHARACTER*( * )    IFMT
+      CHARACTER          IFMT*( * )
      INTEGER            IDIGIT, LDA, LOUT, M, N
 *     ..
 *     .. Array Arguments ..
@@ -41,7 +41,7 @@
     &                   A( LDA, * )
 *     ..
 *     .. Local Scalars ..
-      CHARACTER*80       LINE
+      CHARACTER          LINE*80
      INTEGER            I, J, K1, K2, LLL, NDIGIT
 *     ..
 *     .. Local Arrays ..
@@ -30,7 +30,7 @@
 *     ...
 *     ... SPECIFICATIONS FOR LOCAL VARIABLES
 *     .. Scalar Arguments ..
-      CHARACTER*( * )    IFMT
+      CHARACTER          IFMT*( * )
      INTEGER            IDIGIT, LOUT, N
 *     ..
 *     .. Array Arguments ..
@@ -38,7 +38,7 @@
     &                   SX( * )
 *     ..
 *     .. Local Scalars ..
-      CHARACTER*80       LINE
+      CHARACTER          LINE*80
      INTEGER            I, K1, K2, LLL, NDIGIT
 *     ..
 *     .. Intrinsic Functions ..
@@ -37,8 +37,8 @@
 *     ...
 *     ... SPECIFICATIONS FOR LOCAL VARIABLES
      INTEGER            I, J, NDIGIT, K1, K2, LLL
-      CHARACTER*1        ICOL( 3 )
-      CHARACTER*80       LINE
+      CHARACTER          ICOL( 3 )
+      CHARACTER          LINE*80
 *     ...
 *     ... SPECIFICATIONS INTRINSICS
      INTRINSIC          MIN
@@ -33,7 +33,7 @@
 *     ...
 *     ... SPECIFICATIONS FOR LOCAL VARIABLES
      INTEGER            I, NDIGIT, K1, K2, LLL
-      CHARACTER*80       LINE
+      CHARACTER          LINE*80
 *     ...
 *     ... FIRST EXECUTABLE STATEMENT
 *
@@ -42,11 +42,32 @@ About ILP64 support:
 - Sequential arpack supports [ILP64](https://www.intel.com/content/www/us/en/develop/documentation/onemkl-linux-developer-guide/top/linking-your-application-with-onemkl/linking-in-detail/linking-with-interface-libraries/using-the-ilp64-interface-vs-lp64-interface.html), but, parallel arpack doesn't.
 - Reminder: you can NOT mix `ILP64` with `LP64`. If you compile `arpack-ng` with `ILP64` (resp. `LP64`) support, you MUST insure your BLAS/LAPACK is compliant with `ILP64` (resp. `LP64`).
 - Set `INTERFACE64` at configure time.
+- As this can not be automated, you need to make sure that the BLAS/LAPACK libraries you pass to `arpack-ng`:
+  - Are built with ILP64 support.
+    In case these ILP64 BLAS/LAPACK libraries have a non-usual name, you must specify them like so:
+    ```
+        >> ./configure --with-blas=openblas64_ ...
+    ```
+  - Do export ILP64-flavored symbols. In case these symbols have a non-usual name, you must:
+    - Check for ILP64 symbols exported by the ILP64 BLAS/LAPACK libraries:
+      ```
+         >> nm /path/to/libopenblas64_.so | grep scopy64_
+            0000000000000000 T scopy64_
+      ```
+    - Specify them to `arpack-ng` using `SYMBOLSUFFIX`:
+      ```
+         >> SYMBOLSUFFIX=64_ ./configure ...
+            ...
+            Configuration summary for ARPACK-NG
+            ...
+            FFLAGS              :  -fdefault-integer-8 -Dscopy=scopy64_ ...
+      ```
+      FFLAGS used by `arpack-ng` (built from `SYMBOLSUFFIX` - reported in configure log) *must* redirect the usual symbols to the non-usual ones exported by the ILP64 BLAS/LAPACK libraries: this check is the responsability of the user.

 Note for F77/F90 developers:

 - All files which needs `ILP64` support must include `"arpackicb.h"`.
- When coding, use `i_int` (defined in `arpackicb.h`) instead of `c_int`. `i_int` stands for `iso_c_binding int`: it's `#defined` to `c_int` or `c_int64_t` according to the architecture.
+- When coding, use `i_int` (defined in `arpackicb.h`) instead of `c_int`. `i_int` stands for ISO_C_BINDING integer: it's `#defined` to `c_int` or `c_int64_t` according to the architecture.

 Note for C/C++ developers:

@@ -72,11 +93,16 @@ About ISO_C_BINDING support:
 - The install will now provide `arpack.h/hpp`, `parpack.h/hpp` and friends.
 - Examples of use can be found in `./TESTS` and` ./PARPACK/TESTS/MPI`.

-Those who are interested in `ISO_C_BINDING` support can checkout following links for more informations.
+ISO_C_BINDING is a feature of modern Fortran meant to handle safely interoperability between Fortran and C (in practice, no more need to use ugly tricks to link F77 functions to C code using "underscored" symbols). Basically, ISO_C_BINDING make sure all fortran variables are typed (which may not always be the case when using `implicit` keyword in fortran): this way, C compilers can link properly. For more information on ISO_C_BINDING, you can checkout the following links:

 - <http://fortranwiki.org/fortran/show/ISO_C_BINDING>
 - <http://fortranwiki.org/fortran/show/Generating+C+Interfaces>
- <https://www.roguewave.com/sites/rw/files/attachments/StandardizedMixedLanguageProgrammingforCandFortran.pdf>
+
+Using ICB is seamless:
+
+- Compile `arpack-ng` with ISO_C_BINDING: you'll get both old-fashion fortran symbols and new ISO_C_BINDING symbols available for linking.
+- Add `#include "arpack.h"` in your C code.
+- Replace all [sdcz][ae]upd calls by [sdcz][ae]upd_c: functions suffixed with _c are ISO_C_BINDING compliant (exposing same arguments than original fortran functions).

 **Example**: to test arpack with ISO_C_BINDING

@@ -87,7 +113,7 @@ $ cmake -D ICB=ON

 ## Eigen support

-`arpack-ng` provides C++ eigensolver based on `eigen`.
+`arpack-ng` provides C++ eigensolver based on both ISO_C_BINDING and `eigen`.

 Check out `./EXAMPLES/MATRIX_MARKET/README` for more details.

@@ -96,7 +122,7 @@ Check out `./EXAMPLES/MATRIX_MARKET/README` for more details.
 ```bash
 $ mkdir build
 $ cd build
-$ cmake -D EXAMPLES=ON -D ICB=ON -D ICBEXMM=ON ..
+$ cmake -D EXAMPLES=ON -D ICB=ON -D EIGEN=ON ..
 $ make all check
 ```

@@ -112,7 +138,7 @@ Check out `./EXAMPLES/PYARPACK/README` for more details.
 ```bash
 $ mkdir build
 $ cd build
-$ cmake -D EXAMPLES=ON -D ICB=ON -D ICBEXMM=ON -D PYTHON3=ON ..
+$ cmake -D EXAMPLES=ON -D ICB=ON -D EIGEN=ON -D PYTHON3=ON ..
 $ make all check
 ```

@@ -156,7 +182,7 @@ $ make all check
 ### Getting arpack-ng

 Unlike ARPACK, ARPACK-NG is providing autotools and cmake based build system. In addition, `ARPACK-NG` also provides
-`iso_c_binding` support, which enables to call fortran subroutines natively from C or C++.
+ISO_C_BINDING support, which enables to call fortran subroutines natively from C or C++.

 First, obtain the source code 📥 from github:

@@ -277,10 +303,23 @@ $ LIBS="-framework Accelerate" FFLAGS="-ff2c -fno-second-underscore" FCFLAGS="-f

 ### Windows support

-`arpack-ng` runs on windows using `mingw64`.
+`arpack-ng` can be installed on Windows as a MinGW-w64 package via various distribution, for example through [MSYS2](https://packages.msys2.org/package/mingw-w64-x86_64-arpack) with `pacman -S mingw-w64-x86_64-arpack`. It can also be built and installed through [vcpkg](https://github.com/microsoft/vcpkg) with `vcpkg install arpack-ng`.

 ## Using arpack-ng from your own codebase

+The `*.pc` and `*.cmake` files provided by `arpack-ng` are only pointing to arpack libraries.
+If you need other libraries (like MPI), you must add them alongside arpack (see CMake example below).
+
+Typically, if you need
+
+- ARPACK: at compile/link time, you'll need to provide BLAS and LAPACK.
+
+- ARPACK with eigen support (arpackSolver): at compile/link time, you'll need to provide BLAS, LAPACK and Eigen.
+
+- PARPACK: at compile/link time, you'll need to provide BLAS, LAPACK and MPI.
+
+Examples are provided in `tstCMakeInstall.sh` and `tstAutotoolsInstall.sh` generated after running cmake/configure.
+
 ### With autotools

 First, set `PKG_CONFIG_PATH` to the location in the installation directory where `arpack.pc` lies.
@@ -299,7 +338,7 @@ Note: make sure you have installed `pkg-config`.
 You can use arpack in your CMake builds by using `ARPACK::ARPACK` target. For example,

 ```cmake
-FIND_PACKAGE(arpack-ng)
+FIND_PACKAGE(arpackng)
 ADD_EXECUTABLE(main main.f)
 TARGET_INCLUDE_DIRECTORIES(main PUBLIC ARPACK::ARPACK)
 TARGET_LINK_LIBRARIES(main ARPACK::ARPACK)
@@ -308,19 +347,51 @@ TARGET_LINK_LIBRARIES(main ARPACK::ARPACK)
 To use PARPACK in your Cmake builds, use `PARPACK::PARPACK` target:

 ```cmake
-FIND_PACKAGE(arpack-ng)
+FIND_PACKAGE(arpackng)
+FIND_PACKAGE(MPI REQUIRED COMPONENTS Fortran)
 ADD_EXECUTABLE(main main.f)
 TARGET_INCLUDE_DIRECTORIES(main PUBLIC PARPACK::PARPACK)
 TARGET_LINK_LIBRARIES(main PARPACK::PARPACK)
+TARGET_INCLUDE_DIRECTORIES(main PUBLIC MPI::MPI_Fortran)
+TARGET_LINK_LIBRARIES(main MPI::MPI_Fortran)
 ```

-Note: Make sure to update `CMAKE_MODULE_PATH` env variable (otheriwse, `find_package` won't find arpack-ng cmake file).
+Note: Make sure to update `CMAKE_MODULE_PATH` env variable (otherwise, `find_package` won't find arpack-ng cmake file).

-## Using MKL instead of BLAS / LAPACK
+### FAQ
+
+- Where can I find ARPACK user's guide?
+
+  http://li.mit.edu/Archive/Activities/Archive/CourseWork/Ju_Li/MITCourses/18.335/Doc/ARPACK/Lehoucq97.pdf
+
+- Calling arpack's aupd methods returns `info = -9 - Starting vector is zero.`: why?
+
+  Residuals are null. Try to set `resid` to small values (like epsilon machine magnitude) but *not exactly* zero.
+  Residuals `resid = A*v - lamdba*v` target *exactly* the zero vector.
+  When `resid` is close enough to zero, the iterative procedure stops.
+
+- Say I have an estimate of an eigen value, how to give this information to arpack?
+
+  You need to shift of an amount of about this estimate of `lambda`. Grep `backTransform` in `arpackSolver.hpp` to see an example.
+  For more information, checkout "NUMERICAL METHODS FOR LARGE EIGENVALUE PROBLEMS" by Yousef Saad: https://www-users.cse.umn.edu/~saad/eig_book_2ndEd.pdf (paragraph 4.1.2. and section 4.1.).
+
+- Say I have an estimate of an eigen vector, how to give this information to arpack?
+
+  You need to copy this eigen vector estimate in `v` (not `resid`) and set `info` to 1 before calling aupd methods.
+  The `v` vector targets a non-null vector such that `resid = 0`, that is, such that `A*v = lambda*v`.
+
+- Using PARPACK, I get incorrect eigen values.
+
+  Make sure each MPI processor handles a subpart of the eigen system (matrices) only.
+  ARPACK handles and solves the whole eigen problem (matrices) at once.
+  PARPACK doesn't: each MPI processor must handle and solve a subpart of the eigen system (matrices) only (independently from the other processors).
+  See examples for Fortran in folder `PARPACK/EXAMPLES/MPI`, and for C/C++ examples in `PARPACK/TESTS/MPI/icb_parpack_c.c` and `PARPACK/TESTS/MPI/icb_parpack_cpp.cpp`
+
+## Using MKL (also known as Intel oneAPI or oneMKL) instead of BLAS / LAPACK

 How to use arpack-ng with Intel MKL:

- Let autotools/cmake find MKL for you based on pkg-config files (setting `PKG_CONFIG_PATH`) or cmake options (`BLA_VENDOR=Intel`).
- Refers to the Intel Link Advisor: <https://www.intel.com/content/www/us/en/developer/tools/oneapi/onemkl-link-line-advisor.html>.
+- Let autotools/cmake find MKL for you based on pkg-config files (setting `PKG_CONFIG_PATH`) or cmake options (`BLA_VENDOR=Intel10_64lp` for lp64, `BLA_VENDOR=Intel10_64ilp` for ilp64).
+- Refers to the Intel Link Advisor: <https://www.intel.com/content/www/us/en/developer/tools/oneapi/onemkl-link-line-advisor.html>. Whatever build system you use, you must make sure (e.g. with `make VERBOSE=1`) in the very end that the compile/link options used by the compiler/linker are _exactly_ what Intel libraries expect.

 ## Good luck and enjoy 🎊
@@ -33,3 +33,8 @@ if ICB
 libarpack@LIBSUFFIX@@ITF64SUFFIX@_la_LIBADD += $(top_builddir)/ICB/libdbgicb.la $(top_builddir)/ICB/libstaicb.la
 libarpack@LIBSUFFIX@@ITF64SUFFIX@_la_CPPFLAGS = $(AM_CPPFLAGS) -I$(top_builddir)
 endif
+
+pkgconfig_DATA = arpack@LIBSUFFIX@@ITF64SUFFIX@.pc
+
+# Due to the LIBSUFFIX/ITF64SUFFIX, configure doesn't automatically clean this file:
+DISTCLEANFILES = arpack@LIBSUFFIX@@ITF64SUFFIX@.pc
@@ -1,7 +1,7 @@
 prefix=@prefix@
 exec_prefix=@exec_prefix@
 libdir=@libdir@
-includedir=@includedir@/arpack-ng@ITF64SUFFIX@
+includedir=@includedir@

 Name: @PACKAGE_NAME@
 Description: Collection of Fortran77 subroutines designed to solve large scale eigenvalue problems
@@ -9,4 +9,4 @@ Version: @PACKAGE_VERSION@
 URL: @PACKAGE_URL@
 Libs: -L${libdir} -larpack@LIBSUFFIX@@ITF64SUFFIX@
 Libs.private: @ARPACK_PC_LIBS_PRIVATE@
-Cflags: -I${includedir}
+Cflags: -I${includedir}/arpack@ITF64SUFFIX@
@@ -156,13 +156,13 @@ c     %------------------------%
 c     | Local Scalars & Arrays |
 c     %------------------------%
 c
-      logical    first, orth
+      logical    first, inits, orth
      integer    idist, iseed(4), iter, msglvl, jj
      Real
     &           rnorm0
      Complex
     &           cnorm
-      save       first, iseed, iter, msglvl, orth, rnorm0
+      save       first, iseed, inits, iter, msglvl, orth, rnorm0
 c
 c     %----------------------%
 c     | External Subroutines |
@@ -180,6 +180,12 @@ c
     &           ccdotc
      external   ccdotc, scnrm2, slapy2
 c
+c     %-----------------%
+c     | Data Statements |
+c     %-----------------%
+c
+      data       inits /.true./
+c
 c     %-----------------------%
 c     | Executable Statements |
 c     %-----------------------%
@@ -190,10 +196,13 @@ c     | Initialize the seed of the LAPACK |
 c     | random number generator           |
 c     %-----------------------------------%
 c
-      iseed(1) = 1
-      iseed(2) = 3
-      iseed(3) = 5
-      iseed(4) = 7
+      if (inits) then
+          iseed(1) = 1
+          iseed(2) = 3
+          iseed(3) = 5
+          iseed(4) = 7
+          inits = .false.
+      end if
 c
      if (ido .eq.  0) then
 c
@@ -98,7 +98,7 @@ c     %------------------%
 c     | Scalar Arguments |
 c     %------------------%
 c
-      character*2 which
+      character  which*2
      integer    ishift, kev, np
 c
 c     %-----------------%
@@ -64,7 +64,7 @@ c     %------------------%
 c     | Scalar Arguments |
 c     %------------------%
 c
-      character*2 which
+      character  which*2
      logical    apply
      integer    n
 c
@@ -157,11 +157,11 @@ c     %------------------------%
 c     | Local Scalars & Arrays |
 c     %------------------------%
 c
-      logical    first, orth
+      logical    first, inits, orth
      integer    idist, iseed(4), iter, msglvl, jj
      Double precision
     &           rnorm0
-      save       first, iseed, iter, msglvl, orth, rnorm0
+      save       first, iseed, inits, iter, msglvl, orth, rnorm0
 c
 c     %----------------------%
 c     | External Subroutines |
@@ -183,6 +183,12 @@ c     %---------------------%
 c
      intrinsic    abs, sqrt
 c
+c     %-----------------%
+c     | Data Statements |
+c     %-----------------%
+c
+      data       inits /.true./
+c
 c     %-----------------------%
 c     | Executable Statements |
 c     %-----------------------%
@@ -193,10 +199,13 @@ c     | Initialize the seed of the LAPACK |
 c     | random number generator           |
 c     %-----------------------------------%
 c
-      iseed(1) = 1
-      iseed(2) = 3
-      iseed(3) = 5
-      iseed(4) = 7
+      if (inits) then
+          iseed(1) = 1
+          iseed(2) = 3
+          iseed(3) = 5
+          iseed(4) = 7
+          inits = .false.
+      end if
 c
      if (ido .eq.  0) then
 c
@@ -106,7 +106,7 @@ c     %------------------%
 c     | Scalar Arguments |
 c     %------------------%
 c
-      character*2 which
+      character  which*2
      integer    ishift, kev, np
 c
 c     %-----------------%
@@ -71,7 +71,7 @@ c     %------------------%
 c     | Scalar Arguments |
 c     %------------------%
 c
-      character*2 which
+      character  which*2
      logical    apply
      integer    lda, n, na
 c
@@ -103,7 +103,7 @@ c     %------------------%
 c     | Scalar Arguments |
 c     %------------------%
 c
-      character*2 which
+      character  which*2
      integer    ishift, kev, np
 c
 c     %-----------------%
@@ -69,7 +69,7 @@ c     %------------------%
 c     | Scalar Arguments |
 c     %------------------%
 c
-      character*2 which
+      character  which*2
      logical    apply
      integer    n
 c
@@ -62,7 +62,7 @@ c     %------------------%
 c     | Scalar Arguments |
 c     %------------------%
 c
-      character*2 which
+      character  which*2
      logical    apply
      integer    n
 c
@@ -157,11 +157,11 @@ c     %------------------------%
 c     | Local Scalars & Arrays |
 c     %------------------------%
 c
-      logical    first, orth
+      logical    first, inits, orth
      integer    idist, iseed(4), iter, msglvl, jj
      Real
     &           rnorm0
-      save       first, iseed, iter, msglvl, orth, rnorm0
+      save       first, iseed, inits, iter, msglvl, orth, rnorm0
 c
 c     %----------------------%
 c     | External Subroutines |
@@ -183,6 +183,12 @@ c     %---------------------%
 c
      intrinsic    abs, sqrt
 c
+c     %-----------------%
+c     | Data Statements |
+c     %-----------------%
+c
+      data       inits /.true./
+c
 c     %-----------------------%
 c     | Executable Statements |
 c     %-----------------------%
@@ -193,10 +199,13 @@ c     | Initialize the seed of the LAPACK |
 c     | random number generator           |
 c     %-----------------------------------%
 c
-      iseed(1) = 1
-      iseed(2) = 3
-      iseed(3) = 5
-      iseed(4) = 7
+      if (inits) then
+          iseed(1) = 1
+          iseed(2) = 3
+          iseed(3) = 5
+          iseed(4) = 7
+          inits = .false.
+      end if
 c
      if (ido .eq.  0) then
 c
@@ -106,7 +106,7 @@ c     %------------------%
 c     | Scalar Arguments |
 c     %------------------%
 c
-      character*2 which
+      character  which*2
      integer    ishift, kev, np
 c
 c     %-----------------%
@@ -71,7 +71,7 @@ c     %------------------%
 c     | Scalar Arguments |
 c     %------------------%
 c
-      character*2 which
+      character  which*2
      logical    apply
      integer    lda, n, na
 c
@@ -103,7 +103,7 @@ c     %------------------%
 c     | Scalar Arguments |
 c     %------------------%
 c
-      character*2 which
+      character  which*2
      integer    ishift, kev, np
 c
 c     %-----------------%
@@ -69,7 +69,7 @@ c     %------------------%
 c     | Scalar Arguments |
 c     %------------------%
 c
-      character*2 which
+      character  which*2
      logical    apply
      integer    n
 c
@@ -62,7 +62,7 @@ c     %------------------%
 c     | Scalar Arguments |
 c     %------------------%
 c
-      character*2 which
+      character  which*2
      logical    apply
      integer    n
 c
@@ -156,13 +156,13 @@ c     %------------------------%
 c     | Local Scalars & Arrays |
 c     %------------------------%
 c
-      logical    first, orth
+      logical    first, inits, orth
      integer    idist, iseed(4), iter, msglvl, jj
      Double precision
     &           rnorm0
      Complex*16
     &           cnorm
-      save       first, iseed, iter, msglvl, orth, rnorm0
+      save       first, iseed, inits, iter, msglvl, orth, rnorm0
 c
 c     %----------------------%
 c     | External Subroutines |
@@ -180,6 +180,12 @@ c
     &           zzdotc
      external   zzdotc, dznrm2, dlapy2
 c
+c     %-----------------%
+c     | Data Statements |
+c     %-----------------%
+c
+      data       inits /.true./
+c
 c     %-----------------------%
 c     | Executable Statements |
 c     %-----------------------%
@@ -190,10 +196,13 @@ c     | Initialize the seed of the LAPACK |
 c     | random number generator           |
 c     %-----------------------------------%
 c
-      iseed(1) = 1
-      iseed(2) = 3
-      iseed(3) = 5
-      iseed(4) = 7
+      if (inits) then
+          iseed(1) = 1
+          iseed(2) = 3
+          iseed(3) = 5
+          iseed(4) = 7
+          inits = .false.
+      end if
 c
      if (ido .eq.  0) then
 c
@@ -98,7 +98,7 @@ c     %------------------%
 c     | Scalar Arguments |
 c     %------------------%
 c
-      character*2 which
+      character  which*2
      integer    ishift, kev, np
 c
 c     %-----------------%
@@ -64,7 +64,7 @@ c     %------------------%
 c     | Scalar Arguments |
 c     %------------------%
 c
-      character*2 which
+      character  which*2
      logical    apply
      integer    n
 c
@@ -11,6 +11,7 @@

 #include <cmath>
 #include <iostream>
+#include <stdexcept>
 #include <vector>

 #include "arpack.hpp"
@@ -641,7 +641,7 @@ c
      integer indx(*)
      integer jndx(*)
      integer i, rows, cols, nnz, nnzreq, ounit
-      character*(*)rep,field,symm
+      character  rep*(*),field*(*),symm*(*)
 c
 c Test input qualifiers:
 c
@@ -753,9 +753,9 @@ c Convert uppercase letters to lowercase letters in string with
 c starting position pos and length len.
 cccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccc
      integer pos, len
-      character*(*) string
+      character string*(*)

-      character*26 lcase, ucase
+      character     lcase*26, ucase*26
      save lcase,ucase
      data lcase/'abcdefghijklmnopqrstuvwxyz'/
      data ucase/'ABCDEFGHIJKLMNOPQRSTUVWXYZ'/
@@ -780,8 +780,8 @@ c
 c 30-Oct-96   Bug fix: fixed non-ansi zero stringlength
 cccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccc
      integer slen, start, next, begin, space, wlen
-      character*(*) word
-      character*(*) string
+      character  word*(*)
+      character  string*(*)

      begin = start
      do 5 i=start,slen
@@ -809,7 +809,7 @@ c     Countwd counts the number of words in string starting
 c     at position start.  On return, count is the number of words.
 c 30-Oct-96   Routine added
 cccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccc
-      character*(*) string
+      character string*(*)
      integer slen, start, next, wordlength, count
      character tmp2*2

@@ -31,8 +31,8 @@
 *     ...
 *     ... SPECIFICATIONS FOR LOCAL VARIABLES
      INTEGER            I, J, NDIGIT, K1, K2, LLL
-      CHARACTER*1        ICOL( 3 )
-      CHARACTER*80       LINE
+      CHARACTER          ICOL( 3 )
+      CHARACTER          LINE*80
 *     ...
 *     ... SPECIFICATIONS INTRINSICS
      INTRINSIC          MIN
@@ -30,7 +30,7 @@ c
 *     ...
 *     ... SPECIFICATIONS FOR LOCAL VARIABLES
      INTEGER            I, NDIGIT, K1, K2, LLL
-      CHARACTER*80       LINE
+      CHARACTER          LINE*80
 *     ...
 *     ... FIRST EXECUTABLE STATEMENT
 *
@@ -24,14 +24,14 @@
 *     ...
 *     ... SPECIFICATIONS FOR LOCAL VARIABLES
 *     .. Scalar Arguments ..
-      CHARACTER*( * )    IFMT
+      CHARACTER          IFMT*( * )
      INTEGER            IDIGIT, LDA, LOUT, M, N
 *     ..
 *     .. Array Arguments ..
      DOUBLE PRECISION   A( LDA, * )
 *     ..
 *     .. Local Scalars ..
-      CHARACTER*80       LINE
+      CHARACTER          LINE*80
      INTEGER            I, J, K1, K2, LLL, NDIGIT
 *     ..
 *     .. Local Arrays ..
@@ -21,14 +21,14 @@
 *     ...
 *     ... SPECIFICATIONS FOR LOCAL VARIABLES
 *     .. Scalar Arguments ..
-      CHARACTER*( * )    IFMT
+      CHARACTER          IFMT*( * )
      INTEGER            IDIGIT, LOUT, N
 *     ..
 *     .. Array Arguments ..
      DOUBLE PRECISION   SX( * )
 *     ..
 *     .. Local Scalars ..
-      CHARACTER*80       LINE
+      CHARACTER          LINE*80
      INTEGER            I, K1, K2, LLL, NDIGIT
 *     ..
 *     .. Intrinsic Functions ..
@@ -23,7 +23,7 @@ C     ... SPECIFICATIONS FOR ARGUMENTS
 C     ...
 C     ... SPECIFICATIONS FOR LOCAL VARIABLES
      INTEGER    I, NDIGIT, K1, K2, LLL
-      CHARACTER*80 LINE
+      CHARACTER  LINE*80
 *     ...
 *     ... SPECIFICATIONS INTRINSICS
      INTRINSIC          MIN
@@ -27,8 +27,8 @@
 *     ...
 *     ... SPECIFICATIONS FOR LOCAL VARIABLES
      INTEGER            I, J, NDIGIT, K1, K2, LLL
-      CHARACTER*1        ICOL( 3 )
-      CHARACTER*80       LINE
+      CHARACTER          ICOL( 3 )
+      CHARACTER          LINE*80
 *     ...
 *     ... SPECIFICATIONS INTRINSICS
      INTRINSIC          MIN
@@ -24,7 +24,7 @@
 *     ...
 *     ... SPECIFICATIONS FOR LOCAL VARIABLES
      INTEGER            I, NDIGIT, K1, K2, LLL
-      CHARACTER*80       LINE
+      CHARACTER          LINE*80
 *     ...
 *     ... FIRST EXECUTABLE STATEMENT
 *
@@ -31,8 +31,8 @@
 *     ...
 *     ... SPECIFICATIONS FOR LOCAL VARIABLES
      INTEGER            I, J, NDIGIT, K1, K2, LLL
-      CHARACTER*1        ICOL( 3 )
-      CHARACTER*80       LINE
+      CHARACTER          ICOL( 3 )
+      CHARACTER          LINE*80
 *     ...
 *     ... SPECIFICATIONS INTRINSICS
      INTRINSIC          MIN
@@ -30,7 +30,7 @@ c
 *     ...
 *     ... SPECIFICATIONS FOR LOCAL VARIABLES
      INTEGER            I, NDIGIT, K1, K2, LLL
-      CHARACTER*80       LINE
+      CHARACTER          LINE*80
 *     ...
 *     ... FIRST EXECUTABLE STATEMENT
 *
--- a/Show More
+++ b/Show More
Author	SHA1	Message	Date
Markus Mützel	9304416095	Select different targets for building in CI on Alpine build-alpine / alpine (armv7) (push) Waiting to run build-alpine / alpine (loongarch64) (push) Waiting to run build-alpine / alpine (x86) (push) Waiting to run cross-build-ubuntu / cross armhf (push) Waiting to run cross-build-ubuntu / cross ppc64el (push) Waiting to run cross-build-ubuntu / cross riscv64 (push) Waiting to run cross-build-ubuntu / cross s390x (push) Waiting to run arpack-ng / ubuntu_latest_cmake (push) Waiting to run arpack-ng / ubuntu_latest_cmake_install (push) Waiting to run arpack-ng / ubuntu_latest_autotools (push) Waiting to run arpack-ng / ubuntu_latest_autotools_install (push) Waiting to run arpack-ng / ubuntu_latest_cmake_python (push) Waiting to run arpack-ng / ubuntu_latest_autotools_ilp64 (push) Waiting to run arpack-ng / macos_latest_cmake (push) Waiting to run arpack-ng / macos_latest_cmake_python (push) Waiting to run arpack-ng / macos_latest_autotools (push) Waiting to run arpack-ng / MinGW-w64 CLANG64 INTERFACE64=OFF/MPI=OFF (push) Waiting to run arpack-ng / MinGW-w64 UCRT64 INTERFACE64=OFF/MPI=OFF (push) Waiting to run arpack-ng / MinGW-w64 CLANG64 INTERFACE64=OFF/MPI=ON (push) Waiting to run arpack-ng / MinGW-w64 UCRT64 INTERFACE64=OFF/MPI=ON (push) Waiting to run arpack-ng / MinGW-w64 CLANG64 INTERFACE64=ON/MPI=OFF (push) Waiting to run arpack-ng / MinGW-w64 UCRT64 INTERFACE64=ON/MPI=OFF (push) Waiting to run The Alpine repository for the PowerPC target got more and more unreliable recently. Additionally, the PowerPC target is covered with the cross-build rules on Ubuntu now. Drop some (expensive) targets to reduce the overlap in coverage with the workflow that cross-builds on Ubuntu. Add LoongArch64 target which is not (yet?) available for Ubuntu. Keep targets that can be build natively (which should be reasonably fast). Use an Ubuntu on ARM runner for the armv7 target. Rename workflow because out of the three remaining targets only one is actually emulated - the others can execute natively.	2026-08-16 14:01:08 +02:00
Markus Mützel	9f90017bf8	Add CI workflow that cross-builds for several targets on Ubuntu Cross build ARPACK for some targets on Ubuntu using their cross-compilation toolchains and packages. Run the test suite using `qemu` if necessary.	2026-08-16 14:01:08 +02:00
Markus Mützel	6ae2072b0c	PARPACK: Initialize machine dependent constants only once in `PDNAITR` The current implementation checks the value of the variable `aitr_first` but sets `first`. Change that to set `aitr_first` to false after the initialization has run. This also brings that part of the implementation of `PDNAITR` in line with the corresponding logic that is used in the implementation of the other `P?NAITR`.	2026-08-01 15:59:19 +02:00
Markus Mützel	8a4ede774e	CI: Use VecLibFort as shim library for Apple Accelerate/VecLib on macOS. Currently, the CI rules are building ARPACK using the old `f2c` Fortran ABI. However, all packages of applications written in Fortran that are distributed by Homebrew are using the default GNU Fortran ABI. Testing the package with the ABI expected by most macOS users would probably be better than what is currently done in the CI setup. Instead of building a version of ARPACK using that old, non-default `f2c` Fortran ABI, use a library that "translates" from the default GNU Fortran ABI to the old `f2c` ABI used by Apple Accelerate/VecLib.	2026-07-26 18:28:41 +02:00
dependabot[bot]	7217471833	Bump actions/checkout from 2 to 7 Bumps [actions/checkout](https://github.com/actions/checkout) from 2 to 7. - [Release notes](https://github.com/actions/checkout/releases) - [Changelog](https://github.com/actions/checkout/blob/main/CHANGELOG.md) - [Commits](https://github.com/actions/checkout/compare/v2...v7) --- updated-dependencies: - dependency-name: actions/checkout dependency-version: '7' dependency-type: direct:production update-type: version-update:semver-major ... Signed-off-by: dependabot[bot] <support@github.com>	2026-07-25 17:22:12 +02:00
dependabot[bot]	96892b7938	Bump actions/cache from 4 to 6 Bumps [actions/cache](https://github.com/actions/cache) from 4 to 6. - [Release notes](https://github.com/actions/cache/releases) - [Changelog](https://github.com/actions/cache/blob/main/RELEASES.md) - [Commits](https://github.com/actions/cache/compare/v4...v6) --- updated-dependencies: - dependency-name: actions/cache dependency-version: '6' dependency-type: direct:production update-type: version-update:semver-major ... Signed-off-by: dependabot[bot] <support@github.com>	2026-07-24 19:48:01 +02:00
Markus Mützel	b36bddadd2	CI: Use dependabot to periodically check for updated actions Currently, a warning like the following is emitted for each job in the CI: > Node.js 20 is deprecated. The following actions target Node.js 20 but are being forced to run on Node.js 24: actions/checkout@v2. For more information see: https://github.blog/changelog/2025-09-19-deprecation-of-node-20-on-github-actions-runners/ Let dependabot check for updates to avoid the chore of manually updating the versions of GitHub actions every time the Node.js version on the runner images is updated.	2026-07-24 15:18:16 +02:00
evanbiederstedt	f284b0682d	Fix "Warning: Obsolescent feature: Old-style character length at (1)", found in /UTIL (#482 ) * fix obsolescent character length warning, arpack/utils * updates to changelog, other fixes to avoid warning * try brew install open-mpi * use brew install open-mpi instead * remove F77 syntax, TESTS/mmio.f * Update CHANGES	2026-06-06 12:47:15 +02:00
Kyle GuinnandFranck HOUSSEN	f73592dd1e	Fix header and pkg-config file installation (#426 ) * Don't install pkg-config files for unbuilt libraries The autotools build is configured to skip entire directories depending on the configuration. Each directory's Makefile is responsible for installing what it needs, and moving the .pc files to the toplevel broke that. In the pkg-config files, leave ${includedir} unmodified from how configure defines it. * Install arpackicb.h only when ICB is ON. --------- Co-authored-by: Franck HOUSSEN <fghoussen@users.noreply.github.com>	2025-10-31 15:05:59 +01:00
Theodore	07dbd37e52	Reorder library linking for robustness (#478 ) * fix: reorder library linking for consistency in CMakeLists.txt * fix: unify LAPACK and BLAS library linking across targets * feat: set C++ standard to C++14 for improved language features * fix: update C++ standard to C++14 for Python3, Eigen, and ICB support * fix: improve dependency linking and bump minimum C++ standard version in CMake * Remove redundant C language as it is already enabled previously	2025-10-12 12:03:39 +02:00
Sylvestre Ledru	804fa3149a	Document the recent change	2024-10-21 09:03:36 +02:00
Sylvestre Ledru	b63d37f4b6	Merge pull request #473 from szhorvat/fix/obsolescent-character-length fix: fix obsolescent character length warnings, fixes #472	2024-10-21 09:01:08 +02:00
Szabolcs Horvát	7ada6520f5	fix: fix obsolescent character length warnings, fixes #472	2024-10-20 19:31:46 +00:00
Sylvestre Ledru	9431fd5ea6	Merge pull request #461 from mmuetzel/cmake Make ctests platform independent	2024-09-23 10:14:56 +02:00
Sylvestre Ledru	bc1944851f	Merge pull request #471 from vaerksted/master fix typos	2024-09-23 10:14:09 +02:00
Markus Mützel	2c1203e23f	Update CHANGELOG.	2024-09-22 18:34:27 +02:00
Markus Mützel	32070150c7	CMake: Add function to add tests that change working directory. Add a new CMake function that helps to add tests that change their working directory to somewhere else than the directory with the libarpack library. In that case, the environment variable PATH is set to include the directory with the library on Windows while running the tests. Use that new CMake function for the three ctests involving the Eigen library.	2024-09-22 18:09:41 +02:00
Markus Mützel	7671cebf29	CI: Remove work-around from rules for MinGW that is no longer needed	2024-09-22 17:59:20 +02:00
Markus Mützel	44fe87f3fc	Make ctests platform independent Windows doesn't have a mechanism similar to rpaths. Instead, shared libraries are searched in the current working directory and in the directory with the executable followed by directories in the environment variable PATH. Add the path to the shared libarpack library to PATH for ctests that run executables that are located in a different directory and that set the current working directory not to the directory with the libarpack.dll. Use a generator expression that should be working independent on the CMake generator.	2024-09-22 17:59:20 +02:00
musvaage	04b192a2b1	typos	2024-09-22 10:03:26 -05:00
Franck HOUSSEN	f2921f232b	arpackmm: improve command line for testing. (#470 )	2024-09-19 18:58:07 +02:00
Franck HOUSSEN	43d3e5606b	Fix arpack solver (#468 ) * arpackmm: improve command line. * [BUG FIX] arpackmm: run arpack with real or imag shift. * [BUG FIX] arpackmm: make sure the restart file is written. * [BUG FIX] arpackmm: rewrite the test. - Do NOT use eval in bash script which does NOT return command exit code. - Use a dedicated maxResNorm to test the residual (no point to deduce it from tol). * Update .gitignore. * Update CHANGELOG.	2024-09-18 21:55:38 +02:00
Henri Menke	e55d3b2dcb	Add missing stdexcept header (#467 )	2024-09-14 12:05:42 +02:00
Sylvestre Ledru	0b3038d139	Merge pull request #462 from mmuetzel/ci-macos CI: Enable eigen on macOS runners	2024-06-27 14:30:51 +02:00
Markus Mützel	081e13cd3f	CI: Enable eigen on macOS runners	2024-06-27 13:26:14 +02:00
Sylvestre Ledru	77ba50e800	Merge pull request #459 from mmuetzel/ci CI: Add CI for CLANG64 environment of MSYS2	2024-06-22 14:00:26 +02:00
Markus Mützel	aea67a3426	CI: Re-run tests on error with more output for debugging	2024-06-22 13:04:24 +02:00
Markus Mützel	ca979855b4	CI: Install Eigen on MSYS2 runners	2024-06-22 13:04:24 +02:00
Markus Mützel	8d58c1ac1d	CI: Setup workflow for MSYS2 CLANG64	2024-06-22 13:04:24 +02:00
Sylvestre Ledru	1ebaf744b6	Merge pull request #460 from mmuetzel/ci-macos CI: Install numpy from Homebrew on macOS	2024-06-22 13:04:11 +02:00
Markus Mützel	e5643dce21	CI: Use clearer settings for BLAS/LAPACK implementation on macOS With the previous configuration it was unclear whether the reference implementation (from Netlib) or Apple's implementation of the BLAS and LAPACK libraries should be used. (`LIBS` pointed to Apple's implementation, but `BLA_VENDOR` selected the reference implementation.) Use only `BLA_VENDOR` to select Apple's implementation.	2024-06-22 11:51:22 +02:00
Markus Mützel	0efe8a6113	CI: Install numpy from Homebrew on macOS	2024-06-22 11:51:19 +02:00
Sylvestre Ledru	9d9e1bf36d	Merge pull request #458 from mmuetzel/ci Add CI runners on emulated hardware	2024-06-21 11:48:28 +02:00
Markus Mützel	416011c701	CI (Alpine Linux): Use ccache The compilation (especially of the C++ objects) is pretty slow with qemu. Use ccache to speed up build time on subsequent runs.	2024-06-17 11:15:40 +02:00
Markus Mützel	fec63adf5f	Add CI runners on emulated hardware Use a GitHub action to install Alpine Linux for different, less common architectures that are emulated with qemu (apart from x86). Additionally, Alpine Linux is one of those distributions that are based on musl (instead of glibc). This increases coverage for different configurations in CI.	2024-06-12 19:14:45 +02:00
Franck HOUSSEN	217e25ac7c	Improve documentation. (#456 )	2024-05-08 13:02:16 +02:00
Franck HOUSSEN	7f25013f29	[CI] Fix macos job. (#457 ) * [CI] Fix macos job.	2024-05-08 12:35:10 +02:00
Franck HOUSSEN	4caa8f4c70	Add documentation on ILP64. (#453 ) * Add documentation on ILP64. * Rephrase.	2024-04-01 20:37:14 +02:00
Kyle Guinn	51ce061ff7	Update changelog	2024-01-31 09:59:28 +01:00
Kyle Guinn	ef90e416cf	Build all PARPACK files with optimization	2024-01-31 09:59:28 +01:00
Sylvestre Ledru	ba440d1322	Merge pull request #446 from turboencabulator/quoting Quote the default pkgconfigdir	2024-01-21 08:39:45 +01:00
Kyle Guinn	243a1011e5	Quote the default pkgconfigdir The second argument to AC_SUBST is placed on the right-hand-side of a shell variable assignment. Without the single quotes, ${libdir} is expanded when that variable assignment is executed, which is earlier than we want.	2024-01-20 21:25:36 -06:00
wo80	0eaf38bd51	Fix examples BAND _nbdr2 returning error info = -9 Both examples access the M matrix, so we initialize M to be the identity matrix. See #443 for discussion.	2023-12-20 09:41:27 +01:00
Sam James	31b50762a5	configure.ac: fix bashisms configure scripts need to be runnable with a POSIX-compliant /bin/sh. On many (but not all!) systems, /bin/sh is provided by Bash, so errors like this aren't spotted. Notably Debian defaults to /bin/sh provided by dash which doesn't tolerate such bashisms as '=='. This retains compatibility with bash. Fixes configure warnings/errors like: ``` checking for cheev_ in -llapack... yes ./configure: 8590: test: x: unexpected operator checking how to print strings... printf ```	2023-12-20 08:57:16 +01:00
Sylvestre Ledru	569a3859c1	Merge pull request #438 from sylvestre/cmake configure should still install cmake files	2023-10-15 13:46:22 +02:00
Sylvestre Ledru	3371fdfc8e	configure should still install cmake files	2023-10-14 14:48:34 +02:00
Sylvestre Ledru	40329031ae	also update cmake configure arpack-ng / ubuntu_latest_cmake (push) Waiting to run arpack-ng / ubuntu_latest_cmake_install (push) Waiting to run arpack-ng / ubuntu_latest_autotools (push) Waiting to run arpack-ng / ubuntu_latest_autotools_install (push) Waiting to run arpack-ng / ubuntu_latest_cmake_python (push) Waiting to run arpack-ng / ubuntu_latest_autotools_ilp64 (push) Waiting to run arpack-ng / macos_latest_cmake (push) Waiting to run arpack-ng / macos_latest_cmake_python (push) Waiting to run arpack-ng / macos_latest_autotools (push) Waiting to run arpack-ng / windows_latest_cmake (push) Waiting to run	2023-10-14 13:55:37 +02:00
Sylvestre Ledru	85c35d5d57	update of the version to 3.9.1	2023-10-14 13:38:28 +02:00
Sylvestre Ledru	a80ff0d017	Update of the changelog	2023-10-14 13:38:01 +02:00
Fabien Péan	0f7c32bf78	Update changelog	2023-09-14 12:12:59 +02:00
Fabien Péan	93c048fae3	Fix solver parameters for `pyarpackSparse{LDLT/LU}`	2023-09-14 12:12:59 +02:00
Fabien Péan	447b4387e9	Swap `pyarpackDenseLDLT` and `pyarpackDenseLUPP` tests DenseLUPP tests solves now - complex128 non-symmetric problem - complex64 non-symmetric general problem DenseLDLT solves now - float64 symmetric problem - float32 symmetric general problem	2023-09-14 12:12:59 +02:00
Fabien Péan	12300c1d4b	Change `pyarpackCGILU` to solve hermitian non-general problem	2023-09-14 12:12:59 +02:00
Fabien Péan	de52ee6e32	Swap `pyarpackSparseLDLT` and `pyarpackSparseLU` tests SparseLU tests solves now - complex128 non-symmetric problem - complex64 non-symmetric general problem SparseLDLT solves now - float64 symmetric problem - float32 symmetric general problem	2023-09-14 12:12:59 +02:00
Fabien Péan	bdb881c973	Fix error report in ctest when calling bash script	2023-09-03 18:54:28 +02:00
Fabien Péan	cdeaaeafa0	Update changelog	2023-09-03 15:29:46 +02:00
Fabien Péan	c783dbbb0a	Enable non symmetric tests with BiCG solver in arpackmm test suite	2023-09-03 15:29:46 +02:00
VinceNeede	3c48b9508b	Update README.md MKL options cmake option for mkl BLA_VENDOR=INTEL works for obsolete verdions, replaced with INTEL10_64lp and INTEL10_64ilp for lp64 and ilp64 respectively.	2023-09-02 19:30:16 +02:00
Franck HOUSSEN	e7cf106bc2	Merge pull request #425 from fghoussen/fix-install Fix installation tests: compile and run icb_arpack_c.	2023-08-14 12:26:23 +02:00
Franck HOUSSEN	c08859052b	README: document use of .pc and .cmake files.	2023-08-13 11:04:17 +02:00
Franck HOUSSEN	5a2f942473	CMake test install: add comments + make it symmetric with autotools's one.	2023-08-13 10:59:18 +02:00
Franck HOUSSEN	a857b80721	[CI] Increase CMake version to avoid deprecation messages in logs.	2023-08-13 10:56:55 +02:00
Franck HOUSSEN	cdcb80a0ff	Export LD_LIBRARY_PATH before running binaries.	2023-08-12 21:33:05 +02:00
Franck HOUSSEN	a0ee97487e	README: document use of .pc and .cmake files.	2023-08-12 21:05:52 +02:00
Franck HOUSSEN	4bd39d7a39	[CI] Increase CMake version to avoid deprecation messages in logs.	2023-08-12 20:57:24 +02:00
Franck HOUSSEN	798f87698f	[CI] Add eigen to test installs.	2023-08-12 20:55:42 +02:00
Franck HOUSSEN	031ea2554d	Update change log.	2023-08-12 19:18:35 +02:00
Franck HOUSSEN	7f0e2dca4a	Fix installation tests: compile and run arpackmm.	2023-08-12 19:18:22 +02:00
Franck HOUSSEN	2239bf22e6	Fix installation tests: compile and run icb_parpack_cpp.	2023-08-12 18:02:03 +02:00
Franck HOUSSEN	f4b2f31a64	Fix installation tests: compile and run icb_parpack_c.	2023-08-12 17:40:31 +02:00
Franck HOUSSEN	c46da4f87a	Fix installation tests: compile and run icb_arpack_cpp.	2023-08-12 13:55:09 +02:00
Franck HOUSSEN	1b525b649c	Fix installation tests: compile and run icb_arpack_c.	2023-08-12 13:26:52 +02:00
Fabien Péan	c18d153f20	Update FAQ	2023-08-12 12:39:18 +02:00
Fabien Péan	46d1dd36d0	Fix icb_parpack_[c\|cpp] to split the problem across MPI processes	2023-08-12 12:39:18 +02:00
Szabolcs Horvát	e7ae5f818d	chore: update changelog	2023-08-12 12:39:18 +02:00
Szabolcs Horvát	b29d5630fc	fix: ensure separate seeds on parallel threads, fixes #411	2023-08-12 12:39:18 +02:00
Szabolcs Horvát	9fe9d462a6	fix: propagte RNG state in PARPACK as well	2023-08-12 12:39:18 +02:00
Szabolcs Horvát	0996200c7a	fix: ensure that LAPACK RNG state is propagated	2023-08-12 12:39:18 +02:00
Bálint Aradi	6d625dbf34	Fix package name in CMake find_package() examples (#419 )	2023-05-27 14:09:41 +02:00
Ivan Rostov	3e3e12fef2	Update CMakeLists.txt , add mpiexec to issue46_tst run (#418 )	2023-05-25 22:47:39 +02:00
Vonfry	662cd51fc1	Fix windows icb macros (#416 ) * Distinct new intel llvm compiler on windows intel llvm compiler uses clang-cl who also defines _MSC_VER, which let icx find msc complex.h instead intel copmlex.h * Undef I in windows complex.h I macro is defined by windows complex.h which is used as variable name in arpackSolver as well.	2023-05-18 12:26:31 +02:00
مهدي شينون (Mehdi Chinoune)	22172ff212	Add LLVMFlang flag to build with INTERFACE64 (#412 ) * LLVMFlang: Add required flag to build with INTERFACE64 * LLVMFlang: Add compile flags Add compile flags for preprocessing and for supporting lines longer than 72	2023-04-15 08:40:44 +02:00
Sylvestre Ledru	25524b70c2	Merge pull request #413 from FabienPean/patch-1 Add details on Windows installation	2023-04-10 10:52:38 +02:00
Fabien Péan	2219bef7ea	Update changelog	2023-04-08 19:35:58 +00:00
Fabien Péan	98cd927215	Update README.md	2023-04-08 19:59:20 +02:00
Sylvestre Ledru	396b90a015	Merge pull request #409 from fghoussen/readme README: document where to find arpack user's guide.	2023-03-11 10:07:40 +01:00
Franck HOUSSEN	b13558c527	README: document where to find arpack user's guide.	2023-03-05 22:58:48 +01:00
Franck HOUSSEN	e124fcabc2	[BUG FIX] arpackmm: fix restart. (#408 ) * Fixing issue401. * Fixing issue215 with shift.	2023-02-28 19:29:32 +01:00
Franck HOUSSEN	e144e946ec	README: document how to use ICB.	2023-02-25 10:49:55 +01:00
Franck HOUSSEN	5a963c4d31	Rename icbexmm option into eigen option.	2023-02-18 10:21:19 +01:00
Franck HOUSSEN	ae350abfed	arpackSolver: cut-off arpack workspace dim. (#404 )	2023-02-18 10:20:38 +01:00
Franck HOUSSEN	cd49a7fb98	arpackmm: allow for using LA/SA magnitudes. (#402 )	2023-02-17 22:30:40 +01:00
Franck HOUSSEN	3a9a9ce8c0	Add comments to README. (#400 )	2023-02-15 22:39:19 +01:00
Franck HOUSSEN	459f46a6dd	[BUG FIX] Fix install: headers in /path/to/local/include/arpack (#398 )	2023-02-12 15:29:33 +01:00