From: Martin Kronbichler <kronbichler@lnm.mw.tum.de>
Date: Sun, 12 Mar 2017 19:16:24 +0000 (+0100)
Subject: Avoid floating point exception in Lapack test
X-Git-Tag: v8.5.0-rc1~44^2
X-Git-Url: https://gitweb.dealii.org/cgi-bin/gitweb.cgi?a=commitdiff_plain;h=refs%2Fpull%2F4060%2Fhead;p=dealii.git

Avoid floating point exception in Lapack test
---

diff --git a/tests/lapack/full_matrix_14.cc b/tests/lapack/full_matrix_14.cc
index 77e3f1cec9..e1f89e3ccb 100644
--- a/tests/lapack/full_matrix_14.cc
+++ b/tests/lapack/full_matrix_14.cc
@@ -44,6 +44,11 @@ void test(const bool is_singular)
   catch (std::exception &exc)
     {
       deallog << "matrix is singular" << std::endl;
+      // Some implementations of LAPACK do not detect that the vector we pass
+      // down to the compute_factorization call is compatible with a singular
+      // matrix and divide by zero that triggers a floating point exception,
+      // so we add a small number to the diagonal of the matrix.
+      A(n-1,n-1) += 1e-50;
     }
 
   Vector<double> v(n);