phynexis-deps/armadillo-code
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity

more details

Browse Source
This commit is contained in:
conrad
2020-12-23 00:45:24 +10:00
parent 059fafdbcb
commit 8a7e98acd0
1 changed files with 10 additions and 7 deletions
Download Patch File Download Diff File Expand all files Collapse all files
docs.html
+10 -7
View File
@@ -13233,7 +13233,7 @@ See also:
</ul>
</li>
<br>
<li>The argument <i>sigma</i> is optional; if <i>sigma</i> is given, eigenvalues closest to <i>sigma</i> are found
<li>The argument <i>sigma</i> is optional; if <i>sigma</i> is given, eigenvalues closest to <i>sigma</i> are found via shift-invert mode
<br><b>NOTE:</b> to use <i>sigma</i>, both <i>ARMA_USE_ARPACK</i> and <i>ARMA_USE_SUPERLU</i> must be enabled in <a href="#config_hpp">config.hpp</a>
</li>
<br>
@@ -13272,10 +13272,11 @@ struct eigs_opts
<ul>
<li>the number of obtained eigenvalues/eigenvectors may be lower than requested, depending on the given data</li>
<li>
if the decomposition fails, try first increasing <i>opts.subdim</i> (Krylov subspace dimension)
<br>and, as secondary options, try increasing <i>opts.subdim</i> (maximum number of iterations)
<br>and/or <i>opts.tol</i> (tolerance for convergence) and/or <i>k</i> (number of eigenvalues)
if the decomposition fails, try first increasing <i>opts.subdim</i> (Krylov subspace dimension),
and, as secondary options, try increasing <i>opts.maxiter</i> (maximum number of iterations),
and/or <i>opts.tol</i> (tolerance for convergence), and/or <i>k</i> (number of eigenvalues)
</li>
<li>as an alternative to the "sm" form, use the shift-invert mode with <i>sigma</i> set to zero</li>
</ul>
</li>
<br>
@@ -13357,7 +13358,7 @@ Obtain a limited number of eigenvalues and eigenvectors of <b>sparse</b> general
</ul>
</li>
<br>
<li>The argument <i>sigma</i> is optional; if <i>sigma</i> is given, eigenvalues closest to <i>sigma</i> are found
<li>The argument <i>sigma</i> is optional; if <i>sigma</i> is given, eigenvalues closest to <i>sigma</i> are found via shift-invert mode
<br><b>NOTE:</b> to use <i>sigma</i>, both <i>ARMA_USE_ARPACK</i> and <i>ARMA_USE_SUPERLU</i> must be enabled in <a href="#config_hpp">config.hpp</a>
</li>
<br>
@@ -13401,9 +13402,10 @@ If <i>X</i> is not square sized, a <i>std::logic_error</i> exception is thrown
<li>the number of obtained eigenvalues/eigenvectors may be lower than requested, depending on the given data</li>
<li>
if the decomposition fails, try first increasing <i>opts.subdim</i> (Krylov subspace dimension)
<br>and, as secondary options, try increasing <i>opts.subdim</i> (maximum number of iterations)
<br>and/or <i>opts.tol</i> (tolerance for convergence) and/or <i>k</i> (number of eigenvalues)
and, as secondary options, try increasing <i>opts.maxiter</i> (maximum number of iterations),
and/or <i>opts.tol</i> (tolerance for convergence), and/or <i>k</i> (number of eigenvalues)
</li>
<li>as an alternative to the "sm" form, use the shift-invert mode with <i>sigma</i> set to zero</li>
</ul>
</li>
<br>
@@ -18995,6 +18997,7 @@ List of additions and changes for each version:
<ul>
<li>added <a href="#misc_fns">tgamma()</a></li>
<li>added <a href="#brief_print">.brief_print()</a> for abridged printing of matrices &amp; cubes</li>
<li>expanded <a href="#eigs_sym">eigs_sym()</a> and <a href="#eigs_gen">eigs_gen()</a> with optional shift-invert mode</li>
</ul>
</li>
<br>