yaoyongxin / cygutz

A versatile Gutzwiller solver to optimize the single Slater determinant and local many-body degrees of freedom simultaneously within the Gutzwiller approximation or equivalently Slave-boson approach.

CyGutz

CyGutz is an efficient Gutzwiller solver, which solves a generic tight-binding model with local interactions using Gutzwiller-rotationally invariant slave-boson method.

Please refer to the online documentation at https://cygutz.readthedocs.io/en/stable/

cygutz's Issues

Segmentation Fault in CyGutz

Dear Yongxin,

I have successfully compiled the code, but I met the segmentation fault error while executing CyGutz. Please see the message below:

forrtl: severe (174): SIGSEGV, segmentation fault occurred
Image PC Routine Line Source
CyGutz 000000000051394D Unknown Unknown Unknown
libpthread-2.23.s 00007F6E894BB390 Unknown Unknown Unknown
CyGutz 0000000000477210 bandstru_mp_rotat 292 BANDSTRU.f90
libiomp5.so 00007F6E88EFF413 __kmp_invoke_micr Unknown Unknown
libiomp5.so 00007F6E88ECF60D __kmp_fork_call Unknown Unknown
libiomp5.so 00007F6E88EA7EE8 kmpc_fork_call Unknown Unknown
CyGutz 0000000000476846 bandstru_mp_rotat 269 BANDSTRU.f90
CyGutz 000000000040D477 MAIN 95 CYGUTZ.f90
CyGutz 000000000040D32E Unknown Unknown Unknown
libc-2.23.so 00007F6E88AB0830 __libc_start_main Unknown Unknown
CyGutz 000000000040D229 Unknown Unknown Unknown

The message seems to involve openmp. I have tested it on two computer (8GB and 16GB mem) with intel-2016 and intel-2017. They both show the same error message. I am wondering if you know the solution for this problem?

Thank you.
Tsung-Han

Segmentation fault in Cygutz

Dear Yongxin,

I have been trying to get the code running for the band insulating compounds Ba2YIrO6 (https://journals.aps.org/prb/abstract/10.1103/PhysRevB.93.035129) and NaIrO3 (http://iopscience.iop.org/article/10.1209/0295-5075/101/27003/meta) compounds. I come across Segmentation fault error at the CyGutz step. If I understand correctly, the correlated space automatically chosen by the code when specifying Ir d as the correlated orbitals, turns out to be too large (40 electrons for Ba2YIrO6, I guess this is because of the strong hybridisation of the Ir d orbitals with the O 2p orbitals ), and hence there seems to be some memory issues. Have you ever came across such issues? and if so how does one solve it? In some sense is there a way to specify the orbital numbers just as it is done in the lda+dmft code.

Thanks and Regards
Vamshi