cosmomc: Reionization_zreFromOptDepth: Did not converge to optical depth

Use of Healpix, camb, CLASS, cosmomc, compilers, etc.
Post Reply
Carlo Contaldi
Posts: 16
Joined: November 19 2004
Affiliation: Imperial College

cosmomc: Reionization_zreFromOptDepth: Did not converge to o

Post by Carlo Contaldi » June 12 2015

I keep getting this error running test.ini out of the box with the latest version of cosmomc. Is this a known issue and is there an easy fix? It seems to be drifting to huge taus

Reionization_zreFromOptDepth: Did not converge to optical depth
tau = 0.447991623908533 optical_depth = 0.451455236169879
50.0000000000000 49.9984741210938

Antony Lewis
Posts: 1636
Joined: September 23 2004
Affiliation: University of Sussex
Contact:

Re: cosmomc: Reionization_zreFromOptDepth: Did not converge

Post by Antony Lewis » June 12 2015

Did you include lowTEB? The default test.ini does not use any Planck clik files and does not constrain tau (out of the box it should just calculate one test likelihood)

Carlo Contaldi
Posts: 16
Joined: November 19 2004
Affiliation: Imperial College

cosmomc: Reionization_zreFromOptDepth: Did not converge to o

Post by Carlo Contaldi » June 12 2015

thanks Antony, I hadn't realised the planck bits were commented out. Is it ok to include both planck and B2KPlanck?

Carlo Contaldi
Posts: 16
Joined: November 19 2004
Affiliation: Imperial College

cosmomc: Reionization_zreFromOptDepth: Did not converge to o

Post by Carlo Contaldi » June 12 2015

Actually confused now! which combination of batch ini files should I include to do a basic planck run? It seems all the ones included in the new test.ini look for files that are not included in the public plc package. So should I comment out BKPlanck and all the others and go back to the "CAMSpec" file used in the older versions pre-Planck2015?

Antony Lewis
Posts: 1636
Joined: September 23 2004
Affiliation: University of Sussex
Contact:

Re: cosmomc: Reionization_zreFromOptDepth: Did not converge

Post by Antony Lewis » June 12 2015

You can certainly include batch1/ .ini files to use 2013. Or if you wait until the end of next week there *may* (finally) be 2015 likelihoods.

Luke Hart
Posts: 61
Joined: July 13 2015
Affiliation: University of Manchester

cosmomc: Reionization_zreFromOptDepth: Did not converge to o

Post by Luke Hart » July 21 2015

What would happen if you fixed the \tau value such as below

param[tau] = 0.09

in your submitted .ini file (test.ini)

Would this work if you couldn't get hold of the TEB data?

LH

Antony Lewis
Posts: 1636
Joined: September 23 2004
Affiliation: University of Sussex
Contact:

Re: cosmomc: Reionization_zreFromOptDepth: Did not converge

Post by Antony Lewis » July 21 2015

You can certainly do that, or some fixed lower value (2015 lowL likelihoods are also now available).

Post Reply