I compiled cosmomc on a linux box with Xeon processors with ifc. I found that the likelihoods of every point was set to 1.E+30 (LogZero), and so being rejected, because the logical variable 'wmap_likelihood_ok' was set to false, even though there were no errors reported. My solution was to insert a call to 'wmap_likelhood_error_init' at the beginning of the 'MAPLnLike' function in cmbdata.f90. This way the logical variable would be set to true before the likelihood routine was called. This solved the problem.
I don't know if this problem may crop up with other compilers, or if it was just a 'feature' of the ifc compiler I was using.
cosmomc and wmap3 likelihood with ifc
-
- Posts: 23
- Joined: January 19 2005
- Affiliation: KASI
- Contact:
-
- Posts: 1944
- Joined: September 23 2004
- Affiliation: University of Sussex
- Contact:
Re: cosmomc and wmap3 likelihood with ifc
Make sure you have the latest WMAP likelihood code from their site - they updated it after thursday.David Parkinson wrote:I compiled cosmomc on a linux box with Xeon processors with ifc. I found that the likelihoods of every point was set to 1.E+30 (LogZero), and so being rejected, because the logical variable 'wmap_likelihood_ok' was set to false, even though there were no errors reported. My solution was to insert a call to 'wmap_likelhood_error_init' at the beginning of the 'MAPLnLike' function in cmbdata.f90. This way the logical variable would be set to true before the likelihood routine was called. This solved the problem.