Segmentation fault (core dumped) error during the 3rd order Force constant calculation

[ballal09]

Dear Dr. Tadano,

I am doing a 3rd order force constant calculation for my structure which have 120 atoms in the supercell. The 3rd order force constant calculation went well up to 3rd nearest neighbor cutoff distance but when I increased the cut off distance to 4th nearest neighbor and above (I am running a convergence test with respect to cut off distance), the Alm code stops giving me the following error message:

terminate called after throwing an instance of 'std::bad_alloc'
what(): std::bad_alloc
Aborted (core dumped)

I know it happens due to memory issue. I have tried running it both in my PC and supercomputing cluster but no luck (same error persists). I have attached my input file and last part of the logfile for your reference. Please let me know your valuable suggestions. Thank you very much for your time.

Best
Ballal

Input file:

&general
PREFIX = LYC114_cubic
MODE = optimize
NAT = 120; NKD = 3
KD = Li Y Cl
/

&optimize
DFSET = DFSET_cubic
FC2XML = LYC114.xml
/

&interaction
NORDER = 2 # 1: harmonic, 2: cubic, ..
/

&cell
1.0 # factor in Bohr unit
21.134085522436940607 -0.0 -0.0 # a1
-10.56704276112398766 18.302654948131557688 -0.0 # a2
-0.0 -0.0 46.401745214615175428 # a3
/

&cutoff
Li-Li None 9.17
Li-Y None 9.17
Li-Cl None 5.102
Y-Y None 13.61
Y-Cl None 9.73
Cl-Cl None 7.28

/

&position
1 0.0000000000000000 0.3429361544215348 0.0000000000000000
1 0.0000000000000000 0.3429361544215348 0.2500000000000000
1 0.0000000000000000 0.3429361544215348 0.5000000000000000
1 0.0000000000000000 0.3429361544215348 0.7500000000000000
1 0.6570638455784722 0.6570638455784722 -0.0000000000000000
1 0.6570638455784722 0.6570638455784722 0.2500000000000000
1 0.6570638455784722 0.6570638455784722 0.5000000000000000
1 0.6570638455784722 0.6570638455784722 0.7500000000000000
1 0.3429361544215348 0.0000000000000000 -0.0000000000000000
1 0.3429361544215348 0.0000000000000000 0.2500000000000000
1 0.3429361544215348 0.0000000000000000 0.5000000000000000
1 0.3429361544215348 0.0000000000000000 0.7500000000000000
1 0.0000000000000000 0.6612740425289326 -0.0000000000000000
1 0.0000000000000000 0.6612740425289326 0.2500000000000000
1 0.0000000000000000 0.6612740425289326 0.5000000000000000
1 0.0000000000000000 0.6612740425289326 0.7500000000000000
1 0.3387259874710628 0.3387259874710628 -0.0000000000000000
1 0.3387259874710628 0.3387259874710628 0.2500000000000000
1 0.3387259874710628 0.3387259874710628 0.5000000000000000
1 0.3387259874710628 0.3387259874710628 0.7500000000000000
1 0.6612740425289326 0.0000000000000000 -0.0000000000000000
1 0.6612740425289326 0.0000000000000000 0.2500000000000000
1 0.6612740425289326 0.0000000000000000 0.5000000000000000
1 0.6612740425289326 0.0000000000000000 0.7500000000000000
1 0.0000000000000000 0.6785788635649793 0.1250000000000000
1 0.0000000000000000 0.6785788635649793 0.3750000000000000
1 0.0000000000000000 0.6785788635649793 0.6250000000000000
1 0.0000000000000000 0.6785788635649793 0.8750000000000000
1 0.3214211664350302 0.3214211664350302 0.1250000000000000
1 0.3214211664350302 0.3214211664350302 0.3750000000000000
1 0.3214211664350302 0.3214211664350302 0.6250000000000000
1 0.3214211664350302 0.3214211664350302 0.8750000000000000
1 0.6785788635649793 0.0000000000000000 0.1250000000000000
1 0.6785788635649793 -0.0000000000000000 0.3750000000000000
1 0.6785788635649793 -0.0000000000000000 0.6250000000000000
1 0.6785788635649793 0.0000000000000000 0.8750000000000000
2 0.3333329860000021 0.6666669849999991 0.1271999316413147
2 0.3333329860000021 0.6666669849999991 0.3771999316413148
2 0.3333329860000021 0.6666669849999991 0.6271999316413147
2 0.3333329860000021 0.6666669849999991 0.8771999316413147
2 0.6666669849999991 0.3333329860000021 0.1228000763586859
2 0.6666669849999991 0.3333329860000021 0.3728000683586852
2 0.6666669849999991 0.3333329860000021 0.6228000683586853
2 0.6666669849999991 0.3333329860000021 0.8728000683586853
2 0.0000000000000000 -0.0000000000000000 -0.0000000000000000
2 0.0000000000000000 -0.0000000000000000 0.2500000000000000
2 0.0000000000000000 -0.0000000000000000 0.5000000000000000
2 0.0000000000000000 0.0000000000000000 0.7500000000000000
3 0.1067786918816598 0.5605779415444466 0.1886622725678447
3 0.1067786918816598 0.5605779415444466 0.4386622725678446
3 0.1067786918816598 0.5605779415444466 0.6886622725678448
3 0.1067786918816598 0.5605779415444466 0.9386622725678448
3 0.4394220584555533 0.5462007503371774 0.1886622725678447
3 0.4394220584555533 0.5462007503371774 0.4386622725678446
3 0.4394220584555533 0.5462007503371774 0.6886622725678448
3 0.4394220584555533 0.5462007503371774 0.9386622725678448
3 0.4537992496628226 0.8932213081183402 0.1886622725678447
3 0.4537992496628226 0.8932213081183402 0.4386622725678446
3 0.4537992496628226 0.8932213081183402 0.6886622725678448
3 0.4537992496628226 0.8932213081183402 0.9386622725678448
3 0.5605779415444466 0.1067786918816598 0.0613377314321557
3 0.5605779415444466 0.1067786918816598 0.3113377274321554
3 0.5605779415444466 0.1067786918816598 0.5613377274321552
3 0.5605779415444466 0.1067786918816598 0.8113377274321552
3 0.5462007503371774 0.4394220584555533 0.0613377314321557
3 0.5462007503371774 0.4394220584555533 0.3113377274321554
3 0.5462007503371774 0.4394220584555533 0.5613377274321552
3 0.5462007503371774 0.4394220584555533 0.8113377274321552
3 0.8932213081183402 0.4537992496628226 0.0613377314321557
3 0.8932213081183402 0.4537992496628226 0.3113377274321554
3 0.8932213081183402 0.4537992496628226 0.5613377274321552
3 0.8932213081183402 0.4537992496628226 0.8113377274321552
3 0.1171934047856111 0.2317844472320864 0.1932276147681892
3 0.1171934047856111 0.2317844472320864 0.4432276147681891
3 0.1171934047856111 0.2317844472320864 0.6932275847681870
3 0.1171934047856111 0.2317844472320864 0.9432275847681870
3 0.7682155827679160 0.8854089435535448 0.1932276147681892
3 0.7682155827679160 0.8854089435535448 0.4432276147681891
3 0.7682155827679160 0.8854089435535448 0.6932275847681870
3 0.7682155827679160 0.8854089435535448 0.9432275847681870
3 0.1145910344464533 0.8828065802143913 0.1932276147681892
3 0.1145910344464533 0.8828065802143913 0.4432276147681891
3 0.1145910344464533 0.8828065802143913 0.6932275847681870
3 0.1145910344464533 0.8828065802143913 0.9432275847681870
3 0.2317844472320864 0.1171934047856111 0.0567723892318111
3 0.2317844472320864 0.1171934047856111 0.3067723852318109
3 0.2317844472320864 0.1171934047856111 0.5567724152318130
3 0.2317844472320864 0.1171934047856111 0.8067724152318130
3 0.8854089435535448 0.7682155827679160 0.0567723892318111
3 0.8854089435535448 0.7682155827679160 0.3067723852318109
3 0.8854089435535448 0.7682155827679160 0.5567724152318130
3 0.8854089435535448 0.7682155827679160 0.8067724152318130
3 0.8828065802143913 0.1145910344464533 0.0567723892318111
3 0.8828065802143913 0.1145910344464533 0.3067723852318109
3 0.8828065802143913 0.1145910344464533 0.5567724152318130
3 0.8828065802143913 0.1145910344464533 0.8067724152318130
3 0.2115893656058012 0.7718956086565362 0.0667422980436697
3 0.2115893656058012 0.7718956086565362 0.3167422980436698
3 0.2115893656058012 0.7718956086565362 0.5667422980436696
3 0.2115893656058012 0.7718956086565362 0.8167422980436696
3 0.2281043913434639 0.4396937419492816 0.0667422980436697
3 0.2281043913434639 0.4396937419492816 0.3167422980436698
3 0.2281043913434639 0.4396937419492816 0.5667422980436696
3 0.2281043913434639 0.4396937419492816 0.8167422980436696
3 0.5603062580507113 0.7884106493941964 0.0667422980436697
3 0.5603062580507113 0.7884106493941964 0.3167422980436698
3 0.5603062580507113 0.7884106493941964 0.5667422980436696
3 0.5603062580507113 0.7884106493941964 0.8167422980436696
3 0.7718956086565362 0.2115893656058012 0.1832577019563303
3 0.7718956086565362 0.2115893656058012 0.4332577019563302
3 0.7718956086565362 0.2115893656058012 0.6832577019563304
3 0.7718956086565362 0.2115893656058012 0.9332577019563304
3 0.4396937419492816 0.2281043913434639 0.1832577019563303
3 0.4396937419492816 0.2281043913434639 0.4332577019563302
3 0.4396937419492816 0.2281043913434639 0.6832577019563304
3 0.4396937419492816 0.2281043913434639 0.9332577019563304
3 0.7884106493941964 0.5603062580507113 0.1832577019563303
3 0.7884106493941964 0.5603062580507113 0.4332577019563302
3 0.7884106493941964 0.5603062580507113 0.6832577019563304
3 0.7884106493941964 0.5603062580507113 0.9332577019563304

/

The Alm code stops after the following section in the log file

OPTIMIZATION

LMODEL = least-squares

Training data file (DFSET) : DFSET_cubic

NSTART = 1; NEND = 2147
2147 entries will be used for training.

Total Number of Parameters : 22051
Total Number of Free Parameters : 3353

Entering fitting routine: SVD with constraints considered algebraically.

[ttadano]

To store the sensing matrix A, you will need, at least, 3353 x 120 x 3 x 2147 x 8 (Byte) = 20.7 GB of RAM.

Here are suggested solutions:

• If you used the finite-difference approach for generating the displaced configurations, many elements of the displacements (in DFSET) are zero. In this case, you can reduce the RAM requirement by using a sparse solver (Please set SPARSE = 1 in the &optimize field.)
• Use a computer node having more RAM. I believe many modern systems have RAM of, at least, 128 GB/node.
• Use compressive sensing option (LMODEL = enet | adaptive-lasso) in combination with DFSET generated for randomly sampled structures. This method enables us to reduce the number of #displacement patterns dramatically, which also reduce the RAM requirement. I have not described details of this option in tutorial, but you can have a look at the materials at https://github.com/ttadano/alamode/tree/gh-pages/files.

[ttadano]

Probably, you will have to rebuild OpenBlas with USE_OPENMP=1. If you have build alamode using an anaconda environment, which version of libopenblas are you using? (This can be checked by conda list).

[ballal09]

Dear Dr. Tadano

I see. I will give a try to rebuild OpenBlas with USE_OPENMP=1. FYI, the version of my libopenblas is 0.2.20 . Thank you very much for your quick response.

Best
Ballal Ahammed

[chunweizhu]

Hi @ballal09
OpenBLAS Warning : Detect OpenMP Loop and this application may hang. Please rebuild the library with USE_OPENMP=1 option. Have you solved this problem? I am also facing the same problem, can you give me some guidance? Thanks!
the version of my libopenblas is
libopenblas 0.3.23 pthreads_h80387f5_0 conda-forge
Best

[ballal09]

Hi @chunweizhu
I tried to rebuild OpenBlas with USE_OPENMP=1, but it didn't work for me. I am sorry and I hope you find some ways. Thank you.

[chunweizhu]

Hi @ballal09 , Regardless, thank you for your response.😊