SIGSEGV, segmentation fault occurred for action =2 in CosmoMC

Use of Healpix, camb, CLASS, cosmomc, compilers, etc.
Post Reply
srijita sinha
Posts: 27
Joined: April 12 2019
Affiliation: Indian Institute of Science Education and Research Kolkata

SIGSEGV, segmentation fault occurred for action =2 in CosmoMC

Post by srijita sinha » October 27 2020

Hi,

I have a problem in the same line of the discussion topic:
viewtopic.php?f=11&t=3207&p=8792&hilit=3+Stopping+as+have+in+cosmomc#p8792

It says there was a problem with very small but non-zero neutrino mass. But in the CosmoMC May 2020 that issue is resolved as I can see from

Code: Select all

! Directly solve to avoid issues with perturbative result when no longer very relativistic
in massive_neutrinos.f90. Or I am wrong?

I am pasting last few lines of my log file:

Code: Select all

 minmizing fast parameters again
 minimizing           21  with rhobeg, rhoend =    3.000000      5.0000001E-02
 minimization points:           43 chi2 converge tol:  5.0000001E-02

    New RHO = 3.8730D-01     Number of function values =    44
    Least value of F =  1.184064627873601D+03         The corresponding X is:
    -1.309099D+01   5.173035D+00  -5.039266D+00   6.611972D-01   5.517555D+00
     6.309067D+00  -1.424951D+00  -5.641655D+00   3.152453D+00   8.680639D-01
     5.144410D+00  -1.531492D+01  -8.690081D-01  -5.031092D+00  -8.863623D-01
     2.798367D+00  -1.676246D+00   1.138297D+00   1.378513D+00   3.423952D+00
    -1.534691D+00
     FVAL_Converge_difference reached

    At the return from BOBYQA     Number of function values =    52
    Least value of F =  1.184061956475163D+03         The corresponding X is:
    -1.311725D+01   5.156628D+00  -5.069693D+00   6.297866D-01   5.404794D+00
     6.269616D+00  -1.451581D+00  -5.641655D+00   3.144074D+00   8.204290D-01
     5.144351D+00  -1.532255D+01  -8.672900D-01  -5.030785D+00  -8.861946D-01
     2.798847D+00  -1.677103D+00   1.137249D+00   1.384507D+00   3.458831D+00
    -1.655565D+00
 fast parameter minimize logLike:    1184.06195647516
           3 Refining minimimum using low temp MCMC
           3 Current logLike:    1184.06195647516
           3 Minimize MCMC with temp  1.000000000000000E-002

    New RHO = 5.0000D-02     Number of function values =   359
    Least value of F =  1.183650267850857D+03         The corresponding X is:
     3.462869D+00  -5.769378D+00  -8.958129D-01   6.153088D+00   9.774858D+00
     1.793702D+01   3.548648D+01  -1.951149D+00   8.361154D+00  -1.702369D+01
     2.696975D+00  -5.933188D+00  -2.312695D-01   4.157892D+00   1.664085D+00
     4.465042D+00   6.437730D-01   1.131438D+00   1.172308D+00  -1.280337D+01
     1.476176D+01   1.767684D+00  -1.069306D+00   2.319110D-01   2.324624D+00
    -1.955839D+00   1.702042D+00   1.804873D+00   2.270645D+00   8.833159D+00
           3 Stopping as have          600  samples.
 slow changes         341  power changes           4
           3 MCMC MaxLike =    1182.45642915725
I am minimising 17 parameters with the default theta parameterization using Planck 2018 likelihood.

I cannot understand which means what. Thank you for your help, in advance.

Best
Srijita

srijita sinha
Posts: 27
Joined: April 12 2019
Affiliation: Indian Institute of Science Education and Research Kolkata

Re: SIGSEGV, segmentation fault occurred for action =2 in CosmoMC

Post by srijita sinha » November 01 2020

Hi,

I have an update.

This problem of segmentation fault is occurring when I am using mpirun, without mpirun

Code: Select all

action =2 
ended successfully. What does that mean?

I am using:
iFort version: 18.0.3 20180410


Any help is much appreciated.

Regards,
Srijita

Antony Lewis
Posts: 1652
Joined: September 23 2004
Affiliation: University of Sussex
Contact:

Re: SIGSEGV, segmentation fault occurred for action =2 in CosmoMC

Post by Antony Lewis » November 02 2020

Try setting
minimize_mcmc_refine_num=0

srijita sinha
Posts: 27
Joined: April 12 2019
Affiliation: Indian Institute of Science Education and Research Kolkata

Re: SIGSEGV, segmentation fault occurred for action =2 in CosmoMC

Post by srijita sinha » November 18 2020

Thank you for your reply.

I did that and it is working fine.
Will that change errors? I do not understand how that was causing problem.

Thanks
Srijita

Post Reply