Top Page > Browsing

core dumped in the exchange coupling calculation (Jx) for Mn12 Date: 2020/01/15 11:35 Name: Po-Hao   <chang.pohao@gmail.com> Dear Prof. Ozaki and developers, I have been recently using exchange coupling calculator jx that comes with openmx 3.9 to test several materials and all seem to work well. However when I test it on larger molecules such as Mn12 (the one from the work folder) I was getting core dumped with the following message. -------------------------- -------------------- [c1021:141131] *** Process received signal *** [c1021:141131] Signal: Segmentation fault (11) [c1021:141131] Signal code: (0) [c1021:141131] Failing at address: 0x10000000000000 [c1021:141131] [ 0] /lib64/libc.so.6[0x36e0e32570] [c1021:141131] [ 1] ./jx[0x41c217] [c1021:141131] [ 2] ./jx[0x41f2d5] [c1021:141131] [ 3] /lib64/libc.so.6(__libc_start_main+0x100)[0x36e0e1ed20] [c1021:141131] [ 4] ./jx[0x402e29] [c1021:141131] *** End of error message *** -------------------------------------------------------------------------- Primary job terminated normally, but 1 process returned a non-zero exit code. Per user-direction, the job has been aborted. Also, not sure if it is relevant, I noticed that I chose Cluster solver for Mn12 for self-consistent calculation, but the output from jx calculation still shows Band. The same thing does not happen in simple H-He-H toy system. I have tried to increase the memory, but doesn't seem to help. I wonder if there is any way to work around this issue. Thanks! Regards, Po-Hao

Page: [1]

Re: core dumped in the exchange coupling calculation (Jx) for Mn12 ( No.1 ) Date: 2020/01/16 17:39 Name: Asako Terasawa Hello, I am a developer of jx and happy to hear that jx seems to work mostly fine in your environment. As you pointed out, I found the problem that the eigenvalue solver is read as Band even though the openmx did Cluster calculations. It does make a problem in jx execution. The origin of the problem may be located at the routine that generates *.scfout rather than jx and related routines, because jx reads the type of eigenvalue solver from *scfout file. I will discuss this with other developers and will fix it as soon as possible. I'm sorry for your inconvenience, and appreciate you very much for your feedback. Many thanks, Asako Terasawa Re: core dumped in the exchange coupling calculation (Jx) for Mn12 ( No.2 ) Date: 2020/01/20 16:07 Name: Asako Terasawa After several tests, we found that we should modify some options in Mn12.dat in order to make cluster calculations in jx, as follows: <Atoms.UnitVectors # unit=Ang. 25.0 0.0 0.0 0.0 25.0 0.0 0.0 0.0 25.0 Atoms.UnitVectors> scf.maxIter 100 # default=40 After this modification, jx executes Jij result by cluster calculation as intended.

Page: [1]