Omega_k and COSMOMC

 Posts: 23
 Joined: January 19 2005
 Affiliation: KASI
 Contact:
Omega_k and COSMOMC
In the routine in cosmomc that calculates the likelihood, there is a prior on the curvature, such that Omega_k(min) =0.3 and Omega_k(max) = +0.3. I am interested in running over a much larger prior range (0.9 up to +1.0). Does anyone know if this prior can be extended safely, or if major modifications need to be made to the routine?

 Posts: 183
 Joined: September 24 2004
 Affiliation: Brookhaven National Laboratory
 Contact:
Omega_k and COSMOMC
I think that for a very closed universe, camb will barf:
chi >= pi in closed model not supported
and give up.
chi >= pi in closed model not supported
and give up.

 Posts: 1696
 Joined: September 23 2004
 Affiliation: University of Sussex
 Contact:
Re: Omega_k and COSMOMC
This is true. On the other hand \chi=\pi corresponds to the acoustic peak being at l=0, so with any data it is unlikely to go anywhere near there. So it will probably work for a significantly wider range of \Omega_K if you want to just edit that prior. (of course the curvature may then be implicitly bounded by the prior on H_0 and the age, so you may need to change them too).
On the other hand I haven't actually tested it on large \Omega_K models for years.
On the other hand I haven't actually tested it on large \Omega_K models for years.

 Posts: 183
 Joined: September 24 2004
 Affiliation: Brookhaven National Laboratory
 Contact:
Re: Omega_k and COSMOMC
Yes, but it is enough to propose to such place for camb to stop, even though the sample would have been rejected. I remember that with preWMAP data this was stopping my chains even though it was clear at the time that there is no peak at l=0... So with big enough proposal width it might happen.Antony Lewis wrote:This is true. On the other hand \chi=\pi corresponds to the acoustic peak being at l=0, so with any data it is unlikely to go anywhere near there.