CosmoMC GetDist plots
Posted: May 24 2006
Since many people get this wrong, here is an illustration of what happens when generating plots from a tensor run (with prior r>0), with and without setting GetDist limits parameters correctly:
Incorrect result when limits12 is not set:
Correct result when setting limits12=0 N:
Note that settings limits12=0 1 would give an incorrect result: GetDist would think you had a prior 0<r<1, and hence up-weight the last bin at r=1, hence giving an overestimate of the marginalized probability there.
You need to set limits parameters in the GetDist .ini file for any parameters you use or add that have posterior constraints that do not go to zero at one (or both) of your prior limits.
Also check plots are stable to increasing num_bins, and note that setting prior limits changes whether calculated marginalized limits are one or two tail. Please read the CosmoMC readme for more information.
Incorrect result when limits12 is not set:
Correct result when setting limits12=0 N:
Note that settings limits12=0 1 would give an incorrect result: GetDist would think you had a prior 0<r<1, and hence up-weight the last bin at r=1, hence giving an overestimate of the marginalized probability there.
You need to set limits parameters in the GetDist .ini file for any parameters you use or add that have posterior constraints that do not go to zero at one (or both) of your prior limits.
Also check plots are stable to increasing num_bins, and note that setting prior limits changes whether calculated marginalized limits are one or two tail. Please read the CosmoMC readme for more information.