certain parameters/simulations encountered (unphysical) high values for two-site correlations and/or two-site density matrices with trace much bigger than 1. Up-to-now this only is a problem for intel-fortran compiler with optimization -O2 and -O3 (ifort
14.0.0 on CSM cluster mio, and 15.0.090). Workarounds could be:
using gfortran
ifort with flag "-O0" or "-O1"
ifort with flag in makefileTemplate "-O3 -fp-model strict" or "-O2 -fp-model strict". According to David Vargas' MPI simulation this is faster than -O0/-O1.
The optimization level can be changed in MPSFortLib/makefileTemplate.
We will work on resolving this problem with optimization -O2/3.
Kind regards,
Daniel
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
Dear all,
certain parameters/simulations encountered (unphysical) high values for two-site correlations and/or two-site density matrices with trace much bigger than 1. Up-to-now this only is a problem for intel-fortran compiler with optimization -O2 and -O3 (ifort
14.0.0 on CSM cluster mio, and 15.0.090). Workarounds could be:
makefileTemplate"-O3 -fp-model strict" or "-O2 -fp-model strict". According to David Vargas' MPI simulation this is faster than -O0/-O1.The optimization level can be changed in MPSFortLib/makefileTemplate.
We will work on resolving this problem with optimization -O2/3.
Kind regards,
Daniel