CosmoMC: Lyman Alpha

Use of Healpix, camb, CLASS, cosmomc, compilers, etc.
Post Reply
Loison Hoi
Posts: 23
Joined: February 21 2006
Affiliation: McGill University

CosmoMC: Lyman Alpha

Post by Loison Hoi » June 24 2006

Hi,

The SDSS Lya code, SDSSLY-a-v3.f90, requires a linear inflationary spectrum, i.e. dn/dlnk = 0, as commented in the top of the code and in its website http://www.cita.utoronto.ca/~pmcdonal/L ... fchisq.txt; however, there is a patch (see http://www.slosar.com/aslosar/lya.html) for using a running power spectrum or other non-standard LCDM models, e.g. K \not= 0. My question is: Is this patch reliable for an arbitrary inflationary power spectrum?

By the way, is the code lya.f90 reliable for an arbitrary inflationary power spectrum?

Thank you!

Loison Hoi
23 Jun 2006

Patrick McDonald
Posts: 16
Joined: November 06 2004
Affiliation: CITA
Contact:

CosmoMC: Lyman Alpha

Post by Patrick McDonald » June 24 2006

I see why what I wrote in the file might be confusing, but the LyaF chi^2 table (and the codes based on it) are definitely intended for use with dn/dlnk != 0. The running just can't be so large that dn_effective/dlnk is no longer contained within the boundaries of the table (the code will complain if you go off the table - you shouldn't ignore this). As long as you are using WMAP data, this should not be a problem (unless your model does something extreme at high k, like WDM - you can't use the table for models like that).

Loison Hoi
Posts: 23
Joined: February 21 2006
Affiliation: McGill University

CosmoMC: Lyman Alpha

Post by Loison Hoi » June 25 2006

Hi Patrick,

Thanks for your explanation, but I am still confused.

When you said "the LyaF chi2 table (and the codes based on it) are definitely intended for use with dn/dlnk != 0," did you mean the table "http://www.cita.utoronto.ca/~pmcdonal/L ... fchisq.txt" or the patch "http://www.slosar.com/aslosar/lya.html"? I guess you meant the table, but Antony added a statement in SDSSLy-a-v3.f90:

if (CMB%W /= -1. .or. CMB%omnu/=0. .or. CMB%InitPower(3)/= 0.) then
write (*,*) 'SDSS Lya module does not support extended models'
stop
end if

where CMB%InitPower(3) is the running spectral index, so we cannot have dn/dlnk != 0 when using this code in cosmomc.

Anyway, according to your reply, as long as I use the WMAP data and the modified inflationary power spectrum (not standard non-running or running inflationary spectral index model) is not outside the range of the table, the SDSSLy-a-v3.f90 is still reliable, right?

Another question, if the table works for running spectral index, what is the patch for? non-standard modes, such as CMB%W /= -1. .or. CMB%omnu/=0.?

Thank you very much!

Loison Hoi
24 Jun 2006

Patrick McDonald
Posts: 16
Joined: November 06 2004
Affiliation: CITA
Contact:

CosmoMC: Lyman Alpha

Post by Patrick McDonald » June 25 2006

The patch exists simply to connect the \chi^2 table and interpolator to COSMOMC, so that you don't have to do it yourself.

I haven't looked at the patch, but I don't know any reason why it wouldn't work for a running spectral index, w != -1, or a small component of light neutrinos (not WDM). Antony or Anze will have to answer that.

Antony Lewis
Posts: 1501
Joined: September 23 2004
Affiliation: University of Sussex
Contact:

Re: CosmoMC: Lyman Alpha

Post by Antony Lewis » June 25 2006

The SDSS-lya module supplied with cosmomc only has a reduced chi-squared table with it, no running information. Use Anze/Pat's code if you want to do anything more fancy.

Patrick McDonald
Posts: 16
Joined: November 06 2004
Affiliation: CITA
Contact:

CosmoMC: Lyman Alpha

Post by Patrick McDonald » June 26 2006

Antony explained to me that the standard COSMOMC implementation doesn't include the effective running part of the LyaF \chi^2 table. Anze's patch has this.

Just to make things clear: the LyaF contribution to the constraint on running (or any relatively standard parameter) does not come primarily from a direct measurement of effective running, it comes from combining the LyaF amplitude and slope measurements with the CMB measurements on larger scales, so dropping the effective running constraint probably won't actually affect any results much. I included effective running in the table to provide a final, small, \chi^2 "tug" on parameters.

Post Reply