diff --git a/src/mlpack/methods/kde/kde_rules.hpp b/src/mlpack/methods/kde/kde_rules.hpp
index 44934fb341..5243b4a4f3 100644
--- a/src/mlpack/methods/kde/kde_rules.hpp
+++ b/src/mlpack/methods/kde/kde_rules.hpp
@@ -158,6 +158,9 @@ class KDERules
   constexpr static bool kernelIsGaussian =
       std::is_same<KernelType, kernel::GaussianKernel>::value;
 
+  //! Absolute error tolerance available for each reference point.
+  const double absErrorTol;
+
   //! The last query index.
   size_t lastQueryIndex;
 
diff --git a/src/mlpack/methods/kde/kde_rules_impl.hpp b/src/mlpack/methods/kde/kde_rules_impl.hpp
index 89c7a03af4..24ad7f58c6 100644
--- a/src/mlpack/methods/kde/kde_rules_impl.hpp
+++ b/src/mlpack/methods/kde/kde_rules_impl.hpp
@@ -50,6 +50,7 @@ KDERules<MetricType, KernelType, TreeType>::KDERules(
     kernel(kernel),
     monteCarlo(monteCarlo),
     sameSet(sameSet),
+    absErrorTol(absError / referenceSet.n_cols),
     lastQueryIndex(querySet.n_cols),
     lastReferenceIndex(referenceSet.n_cols),
     baseCases(0),
@@ -146,7 +147,6 @@ Score(const size_t queryIndex, TreeType& referenceNode)
   const double bound = maxKernel - minKernel;
 
   // Error tolerance of the current query point and reference node.
-  const double absErrorTol = absError / referenceSet.n_cols;
   const double relErrorTol = relError * minKernel;
   const double errorTolerance = absErrorTol + relErrorTol;
 
@@ -353,7 +353,6 @@ Score(TreeType& queryNode, TreeType& referenceNode)
   const double bound = maxKernel - minKernel;
 
   // Error tolerance of the current nodes combination.
-  const double absErrorTol = absError / referenceSet.n_cols;
   const double relErrorTol = relError * minKernel;
   const double errorTolerance = absErrorTol + relErrorTol;