ttadano / alamode

Ab initio simulator for thermal transport and lattice anharmonicity
http://sourceforge.net/projects/alamode
MIT License
137 stars 54 forks source link

Error with SCPH+BUBBLE #188

Closed RicardoCecconello closed 5 months ago

RicardoCecconello commented 5 months ago

Dear Dr. Tadano,

I'm having an error when I do a SCPH calculation with BUBBLE > 0 (1, 2 and 3). The error is

free( ): invalid next size (normal) anphon:574864 terminated with signal 6 at PC=7fad8ac25ce1 SP=7fff12713b70. Backtrace: /lib/x86_64-linux-gnu/libc.so.6(gsignal+0x141)[0x7fad8ac25ce1] /lib/x86_64-linux-gnu/libc.so.6(abort+0x123)[0x7fad8ac0f537] /lib/x86_64-linux-gnu/libc.so.6(+0x7a3a8)[0x7fad8ac673a8] /lib/x86_64-linux-gnu/libc.so.6(+0x8169a)[0x7fad8ac6e69a] /lib/x86_64-linux-gnu/libc.so.6(+0x82b6c)[0x7fad8ac6fb6c] /home/caperott/stuff/alamode/_build/anphon/anphon(_ZN7PHON_NS4Scph11postprocessEPPPPSt7complexIdES6_S6_PKNS_17KpointMeshUniformEPPPNS_15MinimumDistListEbi+0x1567)[0x5c9b57] /home/caperott/stuff/alamode/_build/anphon/anphon(_ZN7PHON_NS4Scph9exec_scphEv+0xa57)[0x5ace97] /home/caperott/stuff/alamode/_build/anphon/anphon(_ZN7PHON_NS4PHONC1EiPPci+0x66c)[0x55e9dc] /home/caperott/stuff/alamode/_build/anphon/anphon(main+0x72)[0x510242] /lib/x86_64-linux-gnu/libc.so.6(__libc_start_main+0xea)[0x7fad8ac10d0a] /home/caperott/stuff/alamode/_build/anphon/anphon(_start+0x2a)[0x455bea]

The error terminates the calculation of the DOS part of SCPH (KPMODE = 2), and the last line printed on scph_dos.log is

prefix.scph+bubble(0)_dos : SCPH+Bubble(0) DOS

My input file is

&general
  PREFIX = prefix
  MODE = SCPH
  NKD = 1
  KD = *
  MASS = *
  FCSXML = prefix.xml
  TMIN = 0; TMAX = 50; DT = 1
  EMIN = -300; EMAX = 1200; DELTA_E = 1.0
/

&scph
 SELF_OFFDIAG = 0
 MAXITER = 1000
 MIXALPHA = 0.1
 KMESH_INTERPOLATE = 4 4 2 # supercela harmônica
 KMESH_SCPH = 4 4 2 # múltiplo de kmesh_interpolate
 LOWER_TEMP = 1 # 0 - tmin para tmax
 TOL_SCPH = 1.0e-12
 BUBBLE = 1
/

&cell
...
/

&kpoint
  2  # KPMODE = 2: uniform mesh mode
  20 20 10
/

There are no errors when I do the calculations without BUBBLE, and the calculations with KPMODE = 1 and BUBBLE run with no issues as well. I'm getting huge negative frequencies at the .scph+bubble(*)_bands files. Could this be the cause of the error?

Thanks in advance, Ricardo.

RicardoCecconello commented 5 months ago

The cause is indeed the negative frequencies calculated because the calculation goes normally for other IFCs that I've calculated.