Using ACT or SPT alone

Use of Cobaya. camb, CLASS, cosmomc, compilers, etc.
Post Reply
Will Kinney
Posts: 15
Joined: September 14 2005
Affiliation: Univ. at Buffalo, SUNY
Contact:

Using ACT or SPT alone

Post by Will Kinney » May 31 2013

I think I have this figured out, but would like confirmation from a guru that this will work (and perhaps there is a much simpler way...)

What I would like to do is run likelihoods using Planck+ACT and Planck+SPT separately. I have done the following to accomplish this:

(1) In the Planck likelihood directory, I have created a file act.params, which looks like:

Code: Select all

# prepare a file for calling the wmap full likelihood through clik
res_object = actspt.clik
actspt_data = actfull_3yr_spt_v1
#optional
# l ranges for the act data
lmin11 = 1000
lmin12 = 1500
lmin22 = 1500
lmax11 = 10000
lmax12 = 10000
lmax22 = 10000
#lmax used for the CMB
tt_lmax_mc = 5000
# add or remove datasets
use_act_south = 1
use_act_equa = 1
use_spt_highell = 0
# set to one to copy the data in the clik file, 0 to simply refer to it.
include = 1
(2) I run plc-1.0/bin/prepare_actspt using the above parameter file, to create an actspt.clik directory. I rename this act.clik. The file act.clik/clik/lkl_0/_mdb reads:

Code: Select all

tt_lmax_mc int 5000
lmax22 int 10000
use_spt_highell int 0
lkl_type str actspt
use_act_south int 1
lmax12 int 10000
lmax11 int 10000
lmax int 5000
external_dir str .
use_act_equa int 1
lmin22 int 1500
lmin11 int 1000
lmin int 2
unit int 1
lmin12 int 1500
(3) I create a file CAMspec_ACT_defaults.ini as follows:


Code: Select all

#clik likelihood file, when compiling with cliklike
clik_data_actspt = data/clik/act.clik
clik_params_actspt =data/highL.paramnames
clik_speed_actspt = 6

lmax_computed_cl = 4500

#this contains most of the standard parameters so can be used with various different runs
propose_matrix = planck_CAMspec_highL_merged.covmat

DEFAULT(batch1/CAMspec_defaults.ini)
#CAMspec nuisance parameters now use same priors from above

param[cibr]= 0.95 0.0 1.0 0.04 0.04

#ACT/SPT nuisance parameters
#If combined with CAMspec or PLik, fix the common nuisance parameters here and set them above instead
param[a_ps_act_148] = 10 0 30 0.6 0.6
param[a_ps_act_217] = 75 0 200 4 4
param[a_ps_spt_95] =  0 0 0 0 0
param[a_ps_spt_150] = 0 0 0 0 0
param[a_ps_spt_220] = 0 0 0 0 0
param[r_ps_spt_95x150] =  0 0 0 0 0
param[r_ps_spt_95x220] = 0 0 0 0 0
param[r_ps_150x220] = 0.94 0 1.0 0.02 0.02
param[act_dust_s] = 0.4 0 5 0.1 0.1
param[act_dust_e] = 0.8 0 5 0.1 0.1
prior[act_dust_s] = 0.4 0.2
prior[act_dust_e] = 0.8 0.2
param[cal_acts_148] = 1.00 0.8 1.3 0.01 0.01
param[cal_acts_217] = 1.00 0.8 1.3 0.01 0.01
param[cal_acte_148] = 1.00 0.8 1.3 0.01 0.01
param[cal_acte_217] = 1.00 0.8 1.3 0.01 0.01
param[cal_spt_95] = 0 0 0 0 0
param[cal_spt_150] = 0 0 0 0 0
param[cal_spt_220] = 0 0 0 0 0
Cosmomc complains when I delete the SPT nuisance parameters, so instead here I just zero them out so they are not marginalized over.

To create an SPT-only likelihood, I perform similar steps, instead excluding ACT.

Questions:
(1) Will this in fact produce the desired behavior, i.e. only using the likelihood from one experiment, and not the combined likelihood?

(2) Is this the easiest way to accomplish this, or is there a simpler way?

Thanks in advance.

Tristan L Smith
Posts: 25
Joined: November 14 2005
Affiliation: Swarthmore
Contact:

Using ACT or SPT alone

Post by Tristan L Smith » July 23 2014

Hi Will,

I am trying to run with either WMAP9+ACT or WMAP9+SPT and I am running into some problems that you might have perspective on.

I have been able to make the changes you suggest in this post, but when I run with those data I am not able to reproduce the constraints listed in Table I in 1302.1841 .

To test my code I am just constraining the standard 6 parameters... I was wondering if you would be willing to post some constraints you have obtained from your code with a few different combinations of data sets so that I can compare my output to yours. Thanks in advance!

Erminia Calabrese
Posts: 1
Joined: July 24 2014
Affiliation: Oxford University

Using ACT or SPT alone

Post by Erminia Calabrese » July 24 2014

Hi Tristan,

Will implementation is correct and it is the only way to do it within the Planck likelihood.

However, you can't get the results of that paper using the Planck actspt.clik version. The paper is based on full ACT data (1000-10000) in combination with the SPT 600-3000 new release (Story et al.) not used in the highL code for Planck.
Moreover the ACT data are used to marginalise over foregrounds and the cosmology is obtained from a CMB-only likelihood (this should give pretty identical results compared to the full likelihood though).
You should be able to recover the WMAP+ACT parameters (or maybe only slightly different ones given the different foreground modelling) but not the SPT ones.
The chi2 as well will be very different since you have a different number of both data and dof.

Please let me know if you have further questions.

Erminia

Post Reply