Hi All,
When running a script with charged particles
inter coulomb 0.7 p3m tunev2 accuracy 1e-3
(see compilation status)
Program Information:
ESPResSo-3.2.0
{ Compilation status {
FFTW } { DPD_MASS_LIN } { LENNARD_JONES } { EXTERNAL_FORCES } { MASS } {
ELECTROSTATICS } { TUNABLE_SLIP } { DPD } { TRANS_DPD } }
the following error message is generated:
fftw: planner.c:822: assertion failed: flags.u == u
[n068:19329] *** Process received signal ***
[n068:19329] Signal: Aborted (6)
[n068:19329] Signal code: (-6)
[n068:19329] [ 0] /lib64/libc.so.6 [0x36c1e302d0]
[n068:19329] [ 1] /lib64/libc.so.6(gsignal+0x35) [0x36c1e30265]
[n068:19329] [ 2] /lib64/libc.so.6(abort+0x110) [0x36c1e31d10]
[n068:19329] [ 3] /usr/lib64/libfftw3.so.3 [0x3f93e137b6]
[n068:19329] [ 4] /usr/lib64/libfftw3.so.3 [0x3f93e15acc]
[n068:19329] [ 5] /usr/lib64/libfftw3.so.3(fftw_import_wisdom_from_file+0x4c) [0x3f93ee98ec]
[n068:19329] [ 6] /opt/espresso/espresso-3.2.0DPD/Espresso [0x4c0cc4]
[n068:19329] [ 7] /opt/espresso/espresso-3.2.0DPD/Espresso [0x47b5e0]
[n068:19329] [ 8] /opt/espresso/espresso-3.2.0DPD/Espresso [0x41430b]
[n068:19329] [ 9] /opt/espresso/espresso-3.2.0DPD/Espresso [0x4082bf]
[n068:19329] [10] /opt/espresso/espresso-3.2.0DPD/Espresso [0x40a2b7]
[n068:19329] [11] /opt/espresso/espresso-3.2.0DPD/Espresso [0x404adc]
[n068:19329] [12] /lib64/libc.so.6(__libc_start_main+0xf4) [0x36c1e1d994]
[n068:19329] [13] /opt/espresso/espresso-3.2.0DPD/Espresso [0x4049a9]
[n068:19329] *** End of error message ***
I notice that Espresso generates the following files:
fftw3_1d_wisdom_back_n22.file
Often a few of these files will be generated (forward/back 22 through 50) before the error message is written and the simulation is aborted. Re-submitting the identical script results in an increasing number of files before aborting. After enough tries, eventually the simulation proceeds with the particle integration.
My solution to this has been to copy the list of fftw3* files into each directory where a new script will be run. This has been successful in avoiding the error messages, but I am wondering what causes the original problem and if this is a correct procedure.
Best Regards,
Vincent Ustach