From: Jean-Paul Pelteret <jppelteret@gmail.com>
Date: Tue, 14 May 2019 12:02:15 +0000 (+0200)
Subject: Band-aid for a failing AD test
X-Git-Tag: v9.1.0-rc1~8^2
X-Git-Url: https://gitweb.dealii.org/cgi-bin/gitweb.cgi?a=commitdiff_plain;h=36ebc38da4872a8499967adbcab7b0601b6ea162;p=dealii.git

Band-aid for a failing AD test

Sacado::Rad::Advar<Sacado::Fad::Dfad<double>> fails for the second derivatives (nan) with the current evaluation point. This fixes the test, but one still wonders what the underlying issue is :-/
---

diff --git a/tests/ad_common_tests/helper_scalar_coupled_4_components_01.h b/tests/ad_common_tests/helper_scalar_coupled_4_components_01.h
index 082b1de741..982c86f08d 100644
--- a/tests/ad_common_tests/helper_scalar_coupled_4_components_01.h
+++ b/tests/ad_common_tests/helper_scalar_coupled_4_components_01.h
@@ -417,7 +417,7 @@ test_symmetric_tensor_tensor_vector_scalar_coupled()
   SymmetricTensor<2, dim, ScalarNumberType> st =
     unit_symmetric_tensor<dim, ScalarNumberType>();
   for (unsigned int i = 0; i < st.n_independent_components; ++i)
-    st[st.unrolled_to_component_indices(i)] += 0.72 * (1.5 * i + 0.12);
+    st[st.unrolled_to_component_indices(i)] += 0.1 * i;
   Tensor<2, dim, ScalarNumberType> t =
     unit_symmetric_tensor<dim, ScalarNumberType>();
   for (unsigned int i = 0; i < t.n_independent_components; ++i)