Mutation error at generation 16

[D. King]

Hello,
I am using the local submission option for structure prediction. I have set it to run 24 simultaneous single processor calculations for 24 structures in each generation. At the 16th generation I receive the error:
----------------------------------
{ ??? Attempted to access ranking(1); index out of bounds because
numel(ranking)=0.

Error in ==> lattice_atom_Mutation at 13
r1 = order(ranking(1));

Error in ==> atomLatticeMutation at 10
[MUT_LAT,strainMatrix, new_Coord] =
lattice_atom_Mutation(POP_STRUC.POPULATION(ind).COORDINATES,
POP_STRUC.POPULATION(ind).numIons, POP_STRUC.POPULATION(ind).LATTICE,
POP_STRUC.POPULATION(ind).order)
Error in ==> ev_alg at 954
feval(OFF_STRUC.POPULATION(ind).howCome, ind);
}
-----------------------------------------

the log file looks like:

------------------------------------------
Structure 1 generated from Heredity
Structure 2 generated from Heredity
Structure 3 generated from Heredity
Structure 4 generated from Heredity
Structure 5 generated from Heredity
Structure 6 generated from Heredity
Structure 7 generated from Heredity
Structure 8 generated from Heredity
Structure 9 generated from Heredity
Structure 10 generated from Heredity
Structure 11 generated from Heredity
Structure 12 generated from Heredity
Structure 13 generated from Heredity
Structure 14 generated from Heredity
Structure 15 generated with random symmetry: 13
Structure 16 generated with random symmetry: 112
Structure 17 Generated from softmutation -- from database
Structure 18 Generated from softmutation -- from database
Structure 18 generated from softmutation -- from database
Structure 19 Generated from softmutation -- from database
Structure 20 Generated from softmutation -- from database
Structure 21 Generated from softmutation -- from database
-----------------------------------------------

INPUT.txt file attached.

Thanks,
Daniel King

[Qiang Zhu]

Dear Daniel,

This bug has been reported by some other users. To make it work, you need to manually modify the file FunctionFolder/degradeSimilar.m at lines 33-37: 

from

it = it+1;

if (it + bad_rank == length(ranking))

break;

end

end

to

if (it + bad_rank == length(ranking))

break;

end

it = it+1;

end

You can also update USPEX version (9.3.4) from the website. We are going to release a new version which fixed several minor bugs reported by the users.

Thanks again for your report. And we really appreciate everyone's attention and kind feedback.

Qiang Zhu 

--
You received this message because you are subscribed to the Google Groups "USPEX" group.
To unsubscribe from this group and stop receiving emails from it, send an email to uspex+un...@googlegroups.com.
For more options, visit https://groups.google.com/groups/opt_out.
 
 

--

Qiang Zhu (Ph.D)
ESS 365
Department of Geosciences
Stony Brook University
NY 11794-2100

USPEX download Page: http://uspex.stonybrook.edu/

[Kapil Dhaka]

Dear Daniel,

I think you got your answer by Qiang, he is very nicely replied.

But as far as i know use choose abinitioCode

abinitioCode (which code from CommandExecutable shall be used for calculation? )
1 1 1 1
ENDabinit
 

<< this is for VASP

and executable you insert for Gulp

as 

% commandExecutable
gulp <input >output
% EndExecutable

With best

kapil

--

[Qiang Zhu]

Dear Kapil,

If  you choose whichCluster as 1 or 2, you can just ignore the commandExecutable.

Best,

Qiang

[D. King]

Hello Qiang,
After doing this, removing still* and re-running USPEX, I receive the same output in the log and the same error message.

Regards,
Daniel King

[D. King]

Dear Qiang,
Sorry my last post is wrong, my log has a subtle difference to the one before the edit.
____________

>> >> >> >> >> >> >>  
Structure 1 generated from Heredity
Structure 2 generated from Heredity
Structure 3 generated from Heredity
Structure 4 generated from Heredity
Structure 5 generated from Heredity
Structure 6 generated from Heredity
Structure 7 generated from Heredity
Structure 8 generated from Heredity
Structure 9 generated from Heredity
Structure 10 generated from Heredity
Structure 11 generated from Heredity
Structure 12 generated from Heredity
Structure 13 generated from Heredity
Structure 14 generated from Heredity

Structure 15 generated with random symmetry: 6
Structure 16 generated with random symmetry: 45

Structure 17 Generated from softmutation -- from database
Structure 18 Generated from softmutation -- from database

Structure 19 Generated from softmutation -- from database
Structure 20 Generated from softmutation -- from database

Structure 20 generated from softmutation -- from database
______________
But it does have the same error.

Regards,
Daniel King

On Saturday, April 27, 2013 2:10:51 PM UTC+10, Qiang Zhu wrote:

[Qiang Zhu]

Can you pack the whole directory and send it to me. I need these files to detect what the bug is.

Best

Qiang

[D. King]

Please download the zipped folder from this link
https://www.dropbox.com/s/8aa6e798wu0gnam/Daniel_King_USPEX.zip

[Qiang Zhu]

Dear Daniel,

Your calculation shows very strange behavior. Are you aware that:

 15  347   heredity   [    4  2   ]     -52.313    100.76162  [4  3  3]    2   0.061   3.1446   1.8602

 15  348 softmutation [    4  2   ]     -52.427    100.76359  [4  3  3]   11   0.123   3.1795   1.7858

 15  349   heredity   [    4  2   ]     -52.611    100.35532  [4  3  3]    2   0.050   2.8479   1.7815

 15  350   heredity   [    4  2   ]     -52.613    100.09337  [3  4  3]    2   0.052   2.7922   1.7742

 15  351   heredity   [    4  2   ]     -52.247     99.81670  [4  4  3]    2   0.105   2.8326   1.7280

 15  352   random     [    4  2   ]     -52.023    102.11503  [3  3  3]    2   0.076   2.6944   1.6354

 15  353 softmutation [    4  2   ]     -52.437    100.15052  [4  3  3]    2   0.006   2.8467   1.8021

 15  354 softmutation [    4  2   ]     -53.405     98.82639  [3  3  3]  166   0.020   5.6168   3.4901

 16  355   keptBest   [    4  2   ]     -53.405     98.82639  [3  3  3]  166   0.020   5.6168   3.4901

 16  356   keptBest   [    4  2   ]     -52.941     99.85884  [4  3  3]  139   0.000   4.8252   2.8053

Your calculation at 16th generation only have two valid structures. But all the rest structures are failed during vasp relaxation.  This is quite strange!!! I suspect that your system might undergo some abnormal calculation during a particular period. 

If you just rm still_reading, and re-rerun uspex, the bug would not be fixed. You can only fix the bug with a new calculation from the beginning.

I think you already found the ground state in the calculation. For a system of 6 atoms, you can typically find it within 3 generation. There is no need to continue this calculation. And I believe this bug would disappear if you modified the file which I mentioned in the previous letters.

Best regards,

Qiang

[D. King]

That is very strange indeed. Can i interpret the data as: structure 356 in the .cif file is the lowest energy structure? Because when i view that structure it is not the same as the structure from literature.
Regards,
Daniel

[Qiang Zhu]

In USPEX, the last structure is definitely not the best structure. To analyze data, you should go to results folder and check a few files:

http://uspex.stonybrook.edu/Manual/module_3_2.htm?ms=ARA=&mw=MjQw&st=MA==&sct=MA==

If you look at the file results1/BESTIndividuals, it contains the information for the best structure in each generation.

Gen   ID   Origin Composition  Enthalpy(eV)  Volume(A^3)  KPOINTS  SYMM Q_entropy A_order 

  1   16    random    [    4  2   ]     -53.403           98.65213  [3  3  3]  227    0.0028   6.0963

  2   21   keptBest   [    4  2   ]     -53.403           98.65213  [3  3  3]  227    0.0028   6.0963

  3   48 softmutation [    4  2   ]     -53.403          98.70540  [3  3  3]  227    0.0006   6.2045

  4   69   keptBest   [    4  2   ]     -53.403           98.70540  [3  3  3]  227    0.0006   6.2045

  5   94   keptBest   [    4  2   ]     -53.403           98.70540  [3  3  3]  227    0.0006   6.2045

  6  119   keptBest   [    4  2   ]     -53.403          98.70540  [3  3  3]  227    0.0006   6.2045

  7  141   keptBest   [    4  2   ]     -53.403          98.70540  [3  3  3]  227    0.0006   6.2045

  8  161   keptBest   [    4  2   ]     -53.403          98.70540  [3  3  3]  227    0.0006   6.2045

  9  197 latmutation  [    4  2   ]     -53.404          98.65760  [3  3  3]  166    0.0275   5.5410

 10  209   keptBest   [    4  2   ]     -53.404         98.65760  [3  3  3]  166    0.0275   5.5410

 11  232   keptBest   [    4  2   ]     -53.404         98.65760  [3  3  3]  166    0.0275   5.5410

 12  259   keptBest   [    4  2   ]     -53.404         98.65760  [3  3  3]  166    0.0275   5.5410

 13  294   heredity   [    4  2   ]     -53.404          98.67867  [3  3  3]  166    0.0324   5.3358

 14  305   keptBest   [    4  2   ]     -53.404         98.67867  [3  3  3]  166    0.0324   5.3358

 15  354 softmutation [    4  2   ]     -53.405        98.82639  [3  3  3]  166    0.0197   5.6168

 16  355   keptBest   [    4  2   ]     -53.405         98.82639  [3  3  3]  166    0.0197   5.6168

The enthalpies nearly does not change since the first generation, except a small jump at the 9th generation. Note that the Symmetry also varies there (from fd3m 227 -> R-3m 166). From the enthalpy and volume, I believe Structure 16 and 197 should be very similar. The subtle difference in symmetry might be related to a numerical error bar. Most likely, structure 197 will also become cubic if you use a relative low symmetry tolerance. Otherwise, it indicates a 2nd order phase transition (cubic - hexagonal). Is the structure in literature cubic or hexagonal?

A general comment, you cannot purely rely on the output of .cif file. It is better to extract structures from POSCAR format, visualize it in some software. If you are very experienced crystallorgraphor, you can find the symmetry operators by your eyes. If not, try to determine symmetry in different tolerance values and and then apply symmetry. Symmetry is very subtle and can change the view quite a lot. One should be careful about it. In USPEX, the default setting for Space group determination is:

http://uspex.stonybrook.edu/Manual/module_4_11.htm?ms=AA==&mw=MjQw&st=MA==&sct=MA== 

medium         : SymTolerance (high/medium/low (= 0.01/0.04/0.10))

Best regards,

Qiang Zhu

[D. King]

Ah, thank you Qiang. The structure from literature is cubic.