BICEP likelihoods for Cosmomc

[Antony Lewis]

If anyone wants to try running BICEP2 likelihoods with cosmomc, the "newHL" branch of the cosmomc git repository (access available on request to existing CosmoMC users) has the latest build with support for binned HL likelihood as used by their provided bandpowers (with a little manipulation).

[moslem zarei]

Dear All,

Could you please explain in detail the running of BICEP2 likelihoods with CosmoMC combining with the recent Planck data?

Thank you very much in advance

[Antony Lewis]

If you don't have it, ask me for repository account. Then you can download the branch, and test.ini is an example of how to use with BICEP2 with Planck.

[Feng-Quan Wu]

Dear Antony Lewis,
It seems very interesting, would you please send me an account of the repository? Thanks!

[moslem zarei]

Dear Antony,
Thank you so much for your help.

[bin hu]

Hi Antony, I want one repository account.

Thx, Bin

[Antony Lewis]

The March 2014 version is now also available from the website (further bug fixes and development will continue in the repository version, but known issues are fixed and the binned HL likelihood agrees with the BICEP python code at least for BB likelihoods).

[bin hu]

I got the following report:

---------------
FileUtils.f90(643): catastrophic error: **Internal compiler error: internal abort** Please report this error along with the circumstances in which it occurred in a Software Problem Report. Note: File and line given may not be explicit cause of this error.
compilation aborted for FileUtils.f90 (code 1)
---------------

is this bug due to I use ifort-2013?

Thx in advance!

[Antony Lewis]

Yes, you need ifort 14+

[Subinoy Das]

Hi Antony, I also want one repository account.
can u send me the link at [Log in to view email]

[bin hu]

Yes, you need ifort 14+

Are there some way to keep using ifort 13?
I do not want to use the full power of the newest cosmomc, just for testing
bicep2 lilelihood.

Cheers, Bin

[Jason Dossett]

Hi Bin,

Unfortunately, with the latest release of CosmoMC you must be on ifort 14+. There were a lot of compiler bugs/Standards noncompliance on the older versions of ifort that caused issues with the new OOP structure.

-Jason

[bin hu]

Hi Bin,

Unfortunately, with the latest release of CosmoMC you must be on ifort 14+. There were a lot of compiler bugs/Standards noncompliance on the older versions of ifort that caused issues with the new OOP structure.

-Jason

Hi Jason,

Do I need update Open MPI? Is Open MPI 1.6.4 okay?

Thx, Bin

[Jason Dossett]

Hi Bin,

I don't think there is any new stuff as far as the MPI standards go. So you should be fine with your current version of Open MPI.

-Jason

[Antony Lewis]

Note that if you want to get the lensed BB accurate at low L you need to set

Code: Select all

use_nonlinear_lensing = T

in the .ini file. (the correction is around 7%, depending on parameters). Probably this can also be importance-sampled efficiently after the main chains are run with linear lensing.