CosmoMC: Another problem with 16.06 X intel 2017.3

Use of Cobaya. camb, CLASS, cosmomc, compilers, etc.
Post Reply
Ira *Wolfson
Posts: 69
Joined: January 24 2013
Affiliation: MPA
Contact:

CosmoMC: Another problem with 16.06 X intel 2017.3

Post by Ira *Wolfson » May 14 2017

Hi all,

As was mentioned before I am trying to run CosmoMC on ubuntu 16.06 with the Intel compiler 2017 update no. 3

When trying to MCMC I am encountering the followig error:

Code: Select all

forrtl: severe (194): Run-Time Check Failure. The variable 'var$4277' is being used in 'ParamSet.f90(67,5)' without being defined
Image              PC                Routine            Line        Source             
cosmomc_debug      00000000005F33C2  parampointset_mp_          67  ParamSet.f90
cosmomc_debug      0000000000681396  samplecollector_m         108  SampleCollector.f90
cosmomc_debug      000000000064DD3D  montecarlo_mp_tch         178  MCMC.f90
cosmomc_debug      0000000000657505  montecarlo_mp_tfa         440  MCMC.f90
cosmomc_debug      000000000064D024  montecarlo_mp_tch         144  MCMC.f90
cosmomc_debug      00000000006A7BA2  generalsetup_mp_t         137  GeneralSetup.f90
cosmomc_debug      00000000008FE078  MAIN__                    268  driver.F90
cosmomc_debug      0000000000410CAE  Unknown               Unknown  Unknown
libc-2.23.so       00007F226A79F830  __libc_start_main     Unknown  Unknown
cosmomc_debug      0000000000410BA9  Unknown               Unknown  Unknown

Any clues?

Antony Lewis
Posts: 1936
Joined: September 23 2004
Affiliation: University of Sussex
Contact:

Re: CosmoMC: Another problem with 16.06 X intel 2017.3

Post by Antony Lewis » May 14 2017

Not sure, does it run OK if you compile without runtime checks? (i.e. default build)

I see erratic behaviour in 2017.4 - first run gave access violation, subsequent runs with other compile options ran fine. If it's reproducible try printing out the values of this%P and params_used separately.

Ira *Wolfson
Posts: 69
Joined: January 24 2013
Affiliation: MPA
Contact:

CosmoMC: Another problem with 16.06 X intel 2017.3

Post by Ira *Wolfson » May 14 2017

Unfortunately no go wit the original ./cosmomc :

Code: Select all

 CAMB done
 Calling CAMB
           7 Likelihood:    25.29699     Current Like:  1.0000000E+30
forrtl: severe (174): SIGSEGV, segmentation fault occurred
Image              PC                Routine            Line        Source             
cosmomc            000000000078CA11  tbk_trace_stack_i     Unknown  Unknown
cosmomc            000000000078AB4B  tbk_string_stack_     Unknown  Unknown
cosmomc            00000000007281A4  Unknown               Unknown  Unknown
cosmomc            0000000000727FB6  tbk_stack_trace       Unknown  Unknown
cosmomc            00000000006B07B9  for__issue_diagno     Unknown  Unknown
cosmomc            00000000006BA0B6  for__signal_handl     Unknown  Unknown
libpthread-2.23.s  00007FB0E9673390  Unknown               Unknown  Unknown
libiomp5.so        00007FB0EAE3B7B5  Unknown               Unknown  Unknown
cosmomc            00000000006F1025  for_dealloc_alloc     Unknown  Unknown
cosmomc            00000000005C621D  Unknown               Unknown  Unknown
cosmomc            0000000000654F19  Unknown               Unknown  Unknown
cosmomc            0000000000507294  Unknown               Unknown  Unknown
cosmomc            00000000004999B3  Unknown               Unknown  Unknown
cosmomc            00000000004D3AF7  Unknown               Unknown  Unknown
cosmomc            00000000004D4463  Unknown               Unknown  Unknown
cosmomc            00000000004D2A3A  Unknown               Unknown  Unknown
cosmomc            00000000004EDBE5  Unknown               Unknown  Unknown
cosmomc            00000000005A4811  Unknown               Unknown  Unknown
cosmomc            000000000041076E  Unknown               Unknown  Unknown
libc-2.23.so       00007FB0E8FB0830  __libc_start_main     Unknown  Unknown
cosmomc            0000000000410669  Unknown               Unknown  Unknown


Should I maybe upgrade to 2017.4?

-- Ira

Antony Lewis
Posts: 1936
Joined: September 23 2004
Affiliation: University of Sussex
Contact:

Re: CosmoMC: Another problem with 16.06 X intel 2017.3

Post by Antony Lewis » May 15 2017

It could be a compiler bug. Try writing out to console this%P and params_used separately before the offending line.

Ira *Wolfson
Posts: 69
Joined: January 24 2013
Affiliation: MPA
Contact:

CosmoMC: Another problem with 16.06 X intel 2017.3

Post by Ira *Wolfson » May 30 2017

The offending line is (I think) in MCMC.f90

There is no This%P or params_used defined there.

Instead of params_used Im printing CurParams.

What is the analogue of This%P?

Ira *Wolfson
Posts: 69
Joined: January 24 2013
Affiliation: MPA
Contact:

CosmoMC: Another problem with 16.06 X intel 2017.3

Post by Ira *Wolfson » May 30 2017

Sorry,
I am writing out Trial%P for each MCMC trial parameters I guess....

Ira *Wolfson
Posts: 69
Joined: January 24 2013
Affiliation: MPA
Contact:

CosmoMC: Another problem with 16.06 X intel 2017.3

Post by Ira *Wolfson » May 30 2017

Updated to Intel - 2017.4 X 16.04 ubuntu

This is the output before the offending line: (with Trial%P[\b]

Code: Select all

 4    Trial%P:   1.586609288158535E-002  0.109442807501636     
   1.04263347505320       0.126633649688708       0.000000000000000E+000
  6.000000000000000E-002  0.000000000000000E+000  -1.00000000000000     
  0.000000000000000E+000   3.04600000000000       0.240000000000000     
  0.000000000000000E+000  0.500000000000000        1.00000000000000     
  -1.00000000000000       0.000000000000000E+000   3.12457159179742     
  0.874802477625107       0.000000000000000E+000  0.000000000000000E+000
  3.189786566333036E-002  0.000000000000000E+000  0.000000000000000E+000
   1.00000000000000       0.992791001899292        3.08418153596100     
  0.000000000000000E+000 -0.420000000000000        1.59785963732530     
   19.6000000000000      -0.600000000000000       -3.30000000000000     
  0.000000000000000E+000   2.00000000000000        1.00000000000000     
  0.000000000000000E+000  0.000000000000000E+000  0.000000000000000E+000
  0.000000000000000E+000  0.000000000000000E+000  0.000000000000000E+000
  0.000000000000000E+000  0.000000000000000E+000  0.000000000000000E+000
  0.000000000000000E+000  0.000000000000000E+000  0.000000000000000E+000
  0.000000000000000E+000  0.000000000000000E+000  0.000000000000000E+000
  0.000000000000000E+000  0.000000000000000E+000  0.000000000000000E+000
  0.000000000000000E+000  0.000000000000000E+000  0.000000000000000E+000
  0.000000000000000E+000  0.000000000000000E+000  0.000000000000000E+000
  0.000000000000000E+000  0.000000000000000E+000  0.000000000000000E+000
  0.000000000000000E+000  0.000000000000000E+000  0.000000000000000E+000
  0.000000000000000E+000  0.000000000000000E+000  0.000000000000000E+000
  0.000000000000000E+000  0.000000000000000E+000  0.000000000000000E+000
  0.000000000000000E+000  0.000000000000000E+000  0.000000000000000E+000
  0.000000000000000E+000  0.000000000000000E+000  0.000000000000000E+000
  0.000000000000000E+000  0.000000000000000E+000  0.000000000000000E+000
  0.000000000000000E+000  0.000000000000000E+000  0.000000000000000E+000
  0.000000000000000E+000  0.000000000000000E+000  0.000000000000000E+000
  0.000000000000000E+000  0.000000000000000E+000  0.000000000000000E+000
  0.000000000000000E+000  0.000000000000000E+000  0.000000000000000E+000
  0.000000000000000E+000  0.000000000000000E+000  0.000000000000000E+000
  0.000000000000000E+000  0.000000000000000E+000  0.000000000000000E+000
  0.000000000000000E+000  0.000000000000000E+000  0.000000000000000E+000
  0.000000000000000E+000  0.000000000000000E+000  0.000000000000000E+000
  0.000000000000000E+000  0.000000000000000E+000  0.000000000000000E+000
  0.000000000000000E+000  0.000000000000000E+000  0.000000000000000E+000
  0.000000000000000E+000  0.000000000000000E+000  0.000000000000000E+000
  0.000000000000000E+000  0.000000000000000E+000  0.000000000000000E+000
  0.000000000000000E+000  0.000000000000000E+000  0.000000000000000E+000
  0.000000000000000E+000  0.000000000000000E+000  0.000000000000000E+000
  0.000000000000000E+000  0.000000000000000E+000  0.000000000000000E+000
  0.000000000000000E+000  0.000000000000000E+000  0.000000000000000E+000
  0.000000000000000E+000  0.000000000000000E+000  0.000000000000000E+000
  0.000000000000000E+000  0.000000000000000E+000  0.000000000000000E+000
  0.000000000000000E+000  0.000000000000000E+000  0.000000000000000E+000
  0.000000000000000E+000  0.000000000000000E+000  0.000000000000000E+000
  0.000000000000000E+000  0.000000000000000E+000  0.000000000000000E+000
  0.000000000000000E+000  0.000000000000000E+000  0.000000000000000E+000
  0.000000000000000E+000  0.000000000000000E+000  0.000000000000000E+000
  0.000000000000000E+000  0.000000000000000E+000  0.000000000000000E+000
  0.000000000000000E+000  0.000000000000000E+000  0.000000000000000E+000
  0.000000000000000E+000  0.000000000000000E+000  0.000000000000000E+000
  0.000000000000000E+000  0.000000000000000E+000  0.000000000000000E+000
  0.000000000000000E+000  0.000000000000000E+000  0.000000000000000E+000
  0.000000000000000E+000  0.000000000000000E+000  0.000000000000000E+000
  0.000000000000000E+000  0.000000000000000E+000  0.000000000000000E+000
  0.000000000000000E+000  0.000000000000000E+000  0.000000000000000E+000
  0.000000000000000E+000  0.000000000000000E+000  0.000000000000000E+000
  0.000000000000000E+000  0.000000000000000E+000  0.000000000000000E+000
  0.000000000000000E+000  0.000000000000000E+000  0.000000000000000E+000
  0.000000000000000E+000  0.000000000000000E+000  0.000000000000000E+000
  0.000000000000000E+000  0.000000000000000E+000  0.000000000000000E+000
  0.000000000000000E+000  0.000000000000000E+000  0.000000000000000E+000
  0.000000000000000E+000  0.000000000000000E+000  0.000000000000000E+000
  0.000000000000000E+000  0.000000000000000E+000  0.000000000000000E+000
  0.000000000000000E+000  0.000000000000000E+000  0.000000000000000E+000
  0.000000000000000E+000  0.000000000000000E+000  0.000000000000000E+000
  0.000000000000000E+000  0.000000000000000E+000  0.000000000000000E+000
  0.000000000000000E+000  0.000000000000000E+000  0.000000000000000E+000
  0.000000000000000E+000  0.000000000000000E+000  0.000000000000000E+000
  0.000000000000000E+000  0.000000000000000E+000  0.000000000000000E+000
  0.000000000000000E+000  0.000000000000000E+000  0.000000000000000E+000
  0.000000000000000E+000  0.000000000000000E+000  0.000000000000000E+000
  0.000000000000000E+000  0.000000000000000E+000  0.000000000000000E+000
  0.000000000000000E+000  0.000000000000000E+000  0.000000000000000E+000
  0.000000000000000E+000  0.000000000000000E+000  0.000000000000000E+000
  0.000000000000000E+000  0.000000000000000E+000  0.000000000000000E+000
  0.000000000000000E+000  0.000000000000000E+000  0.000000000000000E+000
  0.000000000000000E+000  0.000000000000000E+000  0.000000000000000E+000
  0.000000000000000E+000  0.000000000000000E+000  0.000000000000000E+000
  0.000000000000000E+000  0.000000000000000E+000  0.000000000000000E+000
  0.000000000000000E+000  0.000000000000000E+000  0.000000000000000E+000
  0.000000000000000E+000  0.000000000000000E+000
forrtl: severe (174): SIGSEGV, segmentation fault occurred
Image              PC                Routine            Line        Source             
cosmomc            00000000006BA784  for__signal_handl     Unknown  Unknown
libpthread-2.23.s  00007F84C9507390  Unknown               Unknown  Unknown
libiomp5.so        00007F84CACE9185  Unknown               Unknown  Unknown
cosmomc            00000000006F15A5  for_dealloc_alloc     Unknown  Unknown
cosmomc            00000000005C67BD  Unknown               Unknown  Unknown
cosmomc            0000000000655319  Unknown               Unknown  Unknown
cosmomc            0000000000507804  Unknown               Unknown  Unknown
cosmomc            0000000000499E33  Unknown               Unknown  Unknown
cosmomc            00000000004D405B  Unknown               Unknown  Unknown
cosmomc            00000000004D49D3  Unknown               Unknown  Unknown
cosmomc            00000000004D2EFA  Unknown               Unknown  Unknown
cosmomc            00000000004EE155  Unknown               Unknown  Unknown
cosmomc            00000000005A4DB1  Unknown               Unknown  Unknown
cosmomc            0000000000410BEE  Unknown               Unknown  Unknown
libc-2.23.so       00007F84C8E44830  __libc_start_main     Unknown  Unknown
cosmomc            0000000000410AE9  Unknown   

the params are :

Code: Select all

local_dir = %LOCALDIR%
data_dir = %LOCALDIR%data/
accuracy_level = 1
stop_on_error = T
file_root = test1
root_dir = chains/
feedback = 2
generic_mcmc = F
cosmology_calculator = CAMB
CAMB_timing = F
highL_theory_cl_template = %DATASETDIR%HighL_lensedCls.dat
highL_unlensed_cl_template = ./camb/HighLExtrapTemplate_lenspotentialCls.dat
k_eta_max_scalar = -1.00000000000000
accurate_BB = F
halofit_version = 4
compute_tensors = T
CMB_lensing = T
use_nonlinear_lensing = T
pivot_k = 0.05
tensor_pivot_k = 5.000000000000000E-002
inflation_consistency = T
bbn_consistency = T
neutrino_hierarchy = degenerate
num_massive_neutrinos = 1
lmax_computed_cl = 4500
lmin_computed_cl = 2000
lmin_store_all_cmb = 2500
lmax_tensor = 600
action = 0
use_fast_slow = T
test_likelihood = F
temperature = 1
MPI_Converge_Stop = 0.01
MPI_LearnPropose = T
MPI_R_StopProposeUpdate = 0
MPI_Max_R_ProposeUpdate = 30
MPI_Max_R_ProposeUpdateNew = 50
MPI_Check_Limit_Converge = T
MPI_Limit_Converge = 0.025
MPI_Limit_Converge_Err = 0.2
MPI_Limit_Param = 0
indep_sample = 0
sampling_method = 7
oversample_fast = 1
propose_scale = 2
burn_in = 0
samples = 4000000
checkpoint = T
flush_write = T
start_at_bestfit = T
no_log = F
rand_seed = -1
get_sigma8 = T
cmb_dataset[BKPLANCK] = %DATASETDIR%BKPlanck/BKPlanck_detset_comb_dust.dataset
cmb_dataset[lensing] = %DATASETDIR%planck_lensing/smica_g30_ftl_full_pp.dataset
cmb_dataset[BKPLANCK,maps_use] = B2K_B P217_B P353_B
cmb_dataset[BKPLANCK,use_min] = 1
cmb_dataset[BKPLANCK,use_max] = 5
cmb_dataset_speed[BKPLANCK] = -1
cmb_dataset_speed[lensing] = -1
use_clik = F
use_HST = F
use_SN = F
use_mpk = F
use_BAO = F
use_SZ = F
use_WL = F
num_threads = 0
estimate_propose_matrix = F
parameterization = theta
H0_min = 40
H0_max = 100
use_min_zre = 0
sterile_mphys_max = 10.0000000000000
param[omegabh2] = 0.0221 0.005 0.1 0.0001 0.0001
param[omegach2] = 0.12 0.001 0.99 0.001 0.0005
param[theta] = 1.0411 0.5 10 0.0004 0.0002
param[tau] = 0.09 0.01 0.8 0.01 0.005
param[omegak] = 0
param[mnu] = 0.06
param[meffsterile] = 0
param[w] = -1
param[wa] = 0
param[nnu] = 3.046
param[yhe] = 0.24
param[alpha1] = 0
param[deltazrei] = 0.5
param[Alens] = 1
param[Alensf] = -1
param[fdm] = 0
param[logA] = 3.1 2 4 0.001 0.001
param[ns] = 0.96 0.8 1.2 0.004 0.002
param[nrun] = 0
param[nrunrun] = 0
param[r] = 0.03 0 2 0.04 0.04
param[nt] = 0
param[ntrun] = 0
param[Aphiphi] = 1
param[calPlanck] = 1 0.9 1.1 0.002 0.0005
param[BBdust] = 3 0 15 0.1 0.1
param[BBsync] = 0.
param[BBalphadust] = -0.42
param[BBbetadust] = 1.59 1.04 2.14 0.02 0.02
param[BBTdust] = 19.6
param[BBalphasync] = -0.6
param[BBbetasync] = -3.3
param[BBdustsynccorr] = 0.0
param[EEtoBB_dust] = 2
param[EEtoBB_sync] = 1
include_fixed_parameter_priors = F
prior[calPlanck] = 1 0.0025
prior[BBbetadust] = 1.59 0.11
fast_param_index = 25
block_semi_fast = F
block_fast_likelihood_params = T
propose_matrix = planck_covmats/base_TT_lowTEB_plik.covmat
max_like_radius = 0.05
max_like_iterations = 5000
minimization_points_factor = 2
minimize_loglike_tolerance = 0.05
minimize_separate_fast = T
minimize_mcmc_refine_num = 20
minimize_refine_temp = 0.01
minimize_temp_scale_factor = 5
minimize_random_start_pos = T
CosmoMC_Version = Nov2016
Compiled_CAMB_version = Jan17
Compiled_Recombination = Recfast_1.5.2
Compiled_Equations = equations_ppf-Jan15
Compiled_Reionization = CAMB_reionization
Compiled_InitialPower = power_tilt


Hope this helps to figure this out :|

Antony Lewis
Posts: 1936
Joined: September 23 2004
Affiliation: University of Sussex
Contact:

Re: CosmoMC: Another problem with 16.06 X intel 2017.3

Post by Antony Lewis » May 31 2017

The parameter vector looks OK. I don't know, you'd need to track down in more detail what exactly is wrong and causing the crash.

I fixed one issue with a log(0) derived parameter in some cases when doing tensor runs. Could try this fix in case it makes a difference (master branch on github).

Ira *Wolfson
Posts: 69
Joined: January 24 2013
Affiliation: MPA
Contact:

CosmoMC: Another problem with 16.06 X intel 2017.3

Post by Ira *Wolfson » June 04 2017

Unfortunately no cigar....

What is the latest stable configuration?
I will scrap everything and just use that.

-- Ira

Antony Lewis
Posts: 1936
Joined: September 23 2004
Affiliation: University of Sussex
Contact:

Re: CosmoMC: Another problem with 16.06 X intel 2017.3

Post by Antony Lewis » June 04 2017

The master branch on github is usually fairly stable.

Ira *Wolfson
Posts: 69
Joined: January 24 2013
Affiliation: MPA
Contact:

CosmoMC: Another problem with 16.06 X intel 2017.3

Post by Ira *Wolfson » June 06 2017

It is most probably a compiler compatibility problem.
Either somewhere in the code or with Intel's 2017.4 compiler suite.

The code (current master branch) works fine on 14.04 X 2016.2, 2016.1, 2016.3

Post Reply