Segmentation fault in CC2 excited state calculation
I am seeing a segfault on a development branch build (commit c232be53), using this input file
- system
charge: 0
- memory
available: 60
- do
restart
response
- method
hf
cc2
- cc response
eom
dipole length
transition moments
- solver scf
residual threshold: 1.0d-8
- solver cholesky
threshold: 1.0d-6
- solver cc gs
residual threshold: 1.0d-7
- solver cc multipliers
threshold: 1.0d-7
- solver cc es
residual threshold: 1.0d-5
singlet states: 3
algorithm: davidson
- solver cc response
storage: memory
threshold: 1.0d-3
- geometry
basis: aug-cc-pVTZ
C 8.446233780861300 5.997787185616000 6.096031907228300
O 7.497494553386500 6.809147978929400 6.754271643749300
C 6.337178449431300 6.949331591858100 5.979118238242900
C 5.495431064377300 5.698596819395500 5.989459978772900
O 4.669235886020400 5.433784748007800 5.148638624615200
H 9.306618072140401 5.914933818000000 6.746627205384300
H 8.751472627736399 6.445205474305000 5.150111265780800
H 8.057499402142200 4.998786808235900 5.896156373048800
H 5.744049810666300 7.742562747571700 6.427774088158200
H 6.555574386269100 7.227529374008500 4.947291276689900
H 5.671535586968700 5.046819584072100 6.858892818329200Note that the issue doesn't occur with a smaller basis set (aug-cc-pVDZ). EOM-CCSD and CC3 both work fine on the same system.
With the debug build (using -rcheck as well) I get this backtrace
Program received signal SIGSEGV: Segmentation fault - invalid memory reference.
Backtrace for this error:
#0 0x7f9b857612e2 in ???
#1 0x7f9b85760475 in ???
#2 0x7f9b8510fdbf in ???
#3 0xafedf1 in __doubles_class_MOD_print_dominant_x2_singlet
at /lfs1i3/home/b35a/hollas.b35a/software/eT/dev/eT/src/wavefunctions/doubles/doubles_class.F90:535
#4 0xae7f5b in __doubles_class_MOD_print_dominant_x_amplitudes_singlet_doubles
at /lfs1i3/home/b35a/hollas.b35a/software/eT/dev/eT/src/wavefunctions/doubles/doubles_class.F90:500
#5 0x6ded92 in __ccs_class_MOD_cc_multipliers_summary_ccs
at /lfs1i3/home/b35a/hollas.b35a/software/eT/dev/eT/src/wavefunctions/ccs/multiplier_equation_ccs.F90:159
#6 0x461a71 in __cc_multipliers_task_class_MOD_execute_cc_multipliers_task
at /lfs1i3/home/b35a/hollas.b35a/software/eT/dev/eT/src/engines/cc/tasks/cc_multipliers_task_class.F90:94
#7 0xa25588 in __eom_transition_moment_engine_class_MOD_ignite_eom_transition_moment_engine
at /lfs1i3/home/b35a/hollas.b35a/software/eT/dev/eT/src/engines/cc/eom_transition_moment_engine_class.F90:90
#8 0x42e11b in __et_class_MOD_run_cc_calculation
at /lfs1i3/home/b35a/hollas.b35a/software/eT/dev/eT/src/program/eT_class.F90:509
#9 0x412d4a in __et_class_MOD_run_calculations
at /lfs1i3/home/b35a/hollas.b35a/software/eT/dev/eT/src/program/eT_class.F90:303
#10 0x413a16 in __et_class_MOD_run
at /lfs1i3/home/b35a/hollas.b35a/software/eT/dev/eT/src/program/eT_class.F90:104
#11 0x413b2a in et_program
at /lfs1i3/home/b35a/hollas.b35a/software/eT/dev/eT/src/eT_program.F90:35
#12 0x443d41 in main
at /lfs1i3/home/b35a/hollas.b35a/software/eT/dev/eT/src/eT_program.F90:28Here are the last lines in the output file:
- Finished solving the cc2 multipliers equations
Total wall time (s): 66.386
Total cpu time (s): 530.745
- CC multipliers solver summary:
Largest single amplitudes:
-----------------------------------
a i tbar(a,i)
-----------------------------------
21 20 0.023845859220
27 20 -0.022547309675
10 16 0.021751939062
13 16 0.021060403817
17 18 0.017228056735
65 19 0.017197618201
55 20 -0.016404707226
10 12 -0.016361917965
8 20 0.015998388917
60 20 -0.015426075278
------------------------------------I build eT with gfortran 13.2.1 and MKL with
./setup.py -type Debug -rcheck -G "Unix Makefiles" -clean -int-lib libint -ll $LIBINT2_ROOT -CC $CC -CXX $CXX -FC $FC -F-flags="-g -Og -flto=$NCPUS"