Cosmomc Instalation and Running

Use of Healpix, camb, CLASS, cosmomc, compilers, etc.
Post Reply
Bruna Santos
Posts: 3
Joined: February 10 2009
Affiliation: Centro Astrofisica UP

Cosmomc Instalation and Running

Post by Bruna Santos » February 10 2009

Dear All

I've installed Cosmomc in my computer. In the final steps of the instalation, I had to do:

1)>./camb params.ini – and results are:
Age of universe/Gyr = 13,592
Doing the above command, I didn't get any error.

But, when I write
.>cosmomc params.ini
.>cosmomc params.ini 1

Random seeds: 23978, 26009 rand_inst: 0

Computing tensors: F

Doing CMB lensing: T

lmax = 2100

Number of C_ls = 3

Varying 7 parameters ( 0 fast)

reading: params_CMB.covmat

starting Monte-Carlo

reading WMAP5 data

Initializing WMAP likelihood, version v3

^Cforrtl: error (69): process interrupted (SIGINT)

Image PC Routine Line Source

. FFFFE424 Unknown Unknown Unknown 4008D16E Unknown Unknown Unknown 400A9315 Unknown Unknown Unknown 40C6BDDE Unknown Unknown Unknown

So, I used the suggestion in, i.e. , ulimit -s unlimited and the program starts run and show the following

Initializing WMAP likelihood, version v3

0 rat: 0.4807692 in 104 (M) best: 1330.946

0 rat: 0.4464286 in 224 (M) best: 1329.457

0 rat: 0.4261364 in 352 (M) best: 1329.457

0 rat: 0.4149378 in 482 (M) best: 1329.457

0 rat: 0.4045307 in 618 (M) best: 1329.457

0 rat: 0.3947369 in 760 (M) best: 1329.441

0 rat: 0.3950339 in 886 (M) best: 1329.441

0 rat: 0.3875969 in 1032 (M) best: 1329.441

Is this normal?
I would like to Know the amount of time required to run the algorithm?

Best Regards


Jussi Valiviita
Posts: 1
Joined: May 30 2008
Affiliation: ICG Portsmouth

Cosmomc Instalation and Running

Post by Jussi Valiviita » February 11 2009

Seems that you have managed to install cosmomc. The output is the normal one. Typically with a good proposal matrix the acceptance ratio is about 0.3 (i.e. the mean multiplicity is about 3). The resulting chain(s) is in cosmomc/chains. You can analyse the results with getdist. Remember to cut out the burn-in period.

How long you need to run depends on your problem. For the adiabatic LambdaCDM with the supplied proposal matrix *.covmat maybe less than one day is enough. For "new models" for which you don't have a good proposal matrix to start with, you may need couple of weeks of initial runs (and playing with your guessed propose widths and the MC temperature) before you obtain a good proposal matrix. Then the final run may take couple of days (or even weeks) on a super computer, and you should create several chains, since the MCMC is a probabilistic thing.

Post Reply