Author Topic: Paths and parallelization  (Read 2856 times)

faq_user

  • Member
  • *
  • Posts: 23
    • View Profile
Paths and parallelization
« on: November 15, 2013, 05:48:17 PM »
Dear Ilya,
 
meanwhile I'm using spinach 1.3.1980 + matlab R2013a. In order to prepare for some 2D experiments, I tried to run nmr_liquids/hmqc_strychnine.m without any changes. Instead of a 2D plot I obtain the following error text at the end of the protocol output:
 
[...allel_function > (parfor body) > basis > banner ]  ===========================================
[...allel_function > (parfor body) > basis > banner ]  =                                         =
[...allel_function > (parfor body) > basis > banner ]  =                BASIS SET                =
[...allel_function > (parfor body) > basis > banner ]  =                                         =
[...allel_function > (parfor body) > basis > banner ]  ===========================================
[...allel_function > (parfor body) > basis > banner ]   
[...ine > parallel_function > (parfor body) > basis ]  spherical tensor basis set using Liouville space matrix formalism.
[...ine > parallel_function > (parfor body) > basis ]  IK-2 approximation - spin correlations involving all nearest neighbours of each spin on the coupling graph.
[...ine > parallel_function > (parfor body) > basis ]  IK-2 approximation - spin correlations up to order 1 between all spins within 4 Angstrom of each other.
[...ine > parallel_function > (parfor body) > basis ]  scalar couplings will be used to build the coupling graph.
[...ine > parallel_function > (parfor body) > basis ]  coupling tensors with norm below 1e-05 will be ignored.
[...ine > parallel_function > (parfor body) > basis ]  connectivity matrix density 11.1531%
[...ine > parallel_function > (parfor body) > basis ]  proximity matrix density 4.3478%
[...ine > parallel_function > (parfor body) > basis ]  WARNING - the system contains 9 non-interacting subsystems.
[...ine > parallel_function > (parfor body) > basis ]  23 subgraphs generated from coupling data.
[...ine > parallel_function > (parfor body) > basis ]  23 subgraphs generated from proximity data.
[...ine > parallel_function > (parfor body) > basis ]  removing identical subgraphs...
[...ine > parallel_function > (parfor body) > basis ]  generated 23 subgraphs of size 1
[...ine > parallel_function > (parfor body) > basis ]  generated 7 subgraphs of size 2
[...ine > parallel_function > (parfor body) > basis ]  generated 7 subgraphs of size 3
[...ine > parallel_function > (parfor body) > basis ]  generated 2 subgraphs of size 4
[...ine > parallel_function > (parfor body) > basis ]  building basis descriptor...
Error using Composite/subsref (line 64)
An invalid indexing request was made.

Error in basis (line 219)
        basis_spec=vertcat(basis_spec{:});

Error in hmqc_strychnine>(parfor body) (line 47)
    subsystem=basis(subsystems{n},bas);

Error in hmqc_strychnine (line 44)
parfor n=1:numel(subsystems)

Caused by:
    Error using Composite/subsref (line 64)
    An invalid indexing request was made.
        Error using Composite/errorIfRemoteOperationInProgress (line 170)
        The Composite value cannot be accessed because
        a remote operation is in progress.
 
This error appears also with any of the other 2D examples.
Because in some 1D runs, matlab R2013a denied the use of the name "spectrum" (obviously they want to give up their function spectrum), I changed "spectrum" to "spectrum1" in my pulse_acquire.m-files runs and got nice spectra.
If I try this workaround with hmqc_strychnine.m I obtain immediately
 
>> hmqc_strychnine
Error using textscan
Invalid file identifier.  Use fopen to generate a valid file identifier.

Error in g03_parse (line 25)
g03_output=textscan(file_id,'%s','bufsize',65536,'delimiter','\n');

Error in hmqc_strychnine_mt (line 12)
[sys,inter]=g03_to_spinach(g03_parse('..\standard_systems\strychnine.log'),...
 
>>
 
I don't know how to solve the problem. Might there be difficulties in the interaction of spinach with this matlab version?

kuprov

  • Administrator
  • Member
  • *****
  • Posts: 58
    • View Profile
Re: Paths and parallelization
« Reply #1 on: November 15, 2013, 05:49:41 PM »
Please try the following:

1.   Restarting Matlab – the parallelization error looks like Matlab’s parallel pool getting stuck.
2.   Switching the slash symbols from ‘\’to ‘/’ – if you are running Linux or MacOS, that might be it, the error looks like it simply didn’t find the strychnine Gaussian log.
3.   Make sure Spinach paths occur first on the Matlab path list – that would prevent the spectrum name collision, MathWorks have promised to eliminate the offending function in the future release.