Changed the conditions for the application of the (1-x)/(1+x) transformation,

[ginac.git] / ginac / inifcns_nstdsums.cpp

diff --git a/ginac/inifcns_nstdsums.cpp b/ginac/inifcns_nstdsums.cpp
index 3920bf9e32266daeada8c78077eff79b623eb2a6..88f0fa4ee5f313d3a3c2925294616984782c94c4 100644 (file)
--- a/ginac/inifcns_nstdsums.cpp
+++ b/ginac/inifcns_nstdsums.cpp
@@ -2839,10 +2839,16 @@ static ex H_evalf(const ex& x1, const ex& x2)
                        }
                }
 
-               // since the transformations produce a lot of terms, they are only efficient for
-               // argument near one.
-               // no transformation needed -> do summation
-               if (cln::abs(x) < 0.95) {
+               // check for the applicability of transformations
+               // 
+               // first condition: since the transformations produce a lot of terms,
+               //   they are only efficient for argument near the boundary |x| = 1, then no
+               //   transformation is needed
+               // second condition: veto for region around +-I to avoid endless recursion
+               //   with the (1-x)/(1+x) transformation. 1.198 is sqrt(1.4142) is the
+               //   boundary of the problematic transformation.
+               //   
+               if (cln::abs(x) < 0.95 || (cln::abs(x) < 1 && cln::abs(x-1) >= 1.198)) {
                        lst m_lst;
                        lst s_lst;
                        ex pf;