### Cobaya: not generating any data when using polychord

Posted:

**September 22 2022**Dear all,

I'm using cobaya and polychord to verify my modified CAMB code. However, it doesn't generate any data(raw points, live points, etc.) even after running for a few days on a local server using 48 CPUs. It keeps throwing the following message:

TTanhReionization_zreFromOptDepth: Did not converge to optical depth

tau = 0.68200043554722278 optical_depth = 0.69826116316660436

50.000000000000000 49.993896484375000

(If running a chain, have you put a constraint on tau?)

However, I have indeed put a constraint on tau using cobaya-cosmo-generator as follows:

tau:

prior:

min: 0.01

max: 0.8

ref:

dist: norm

loc: 0.055

scale: 0.006

proposal: 0.003

latex: \tau_\mathrm{reio}

I've checked that the modified CAMB code runs with no problem with sensible ranges of physical parameters, and cobaya also runs safely with the standard mcmc(with fast-dragging) sampler. But I want to use polychord since I suspect that my model might have some multi-modal distributions. Is this a normal situation, so do I have to wait more days? Or should I change the constraints of some of the parameters, maybe tau?

Any comments or help would be appreciated. Thank you in advance.

Jiwon Park

I'm using cobaya and polychord to verify my modified CAMB code. However, it doesn't generate any data(raw points, live points, etc.) even after running for a few days on a local server using 48 CPUs. It keeps throwing the following message:

TTanhReionization_zreFromOptDepth: Did not converge to optical depth

tau = 0.68200043554722278 optical_depth = 0.69826116316660436

50.000000000000000 49.993896484375000

(If running a chain, have you put a constraint on tau?)

However, I have indeed put a constraint on tau using cobaya-cosmo-generator as follows:

tau:

prior:

min: 0.01

max: 0.8

ref:

dist: norm

loc: 0.055

scale: 0.006

proposal: 0.003

latex: \tau_\mathrm{reio}

I've checked that the modified CAMB code runs with no problem with sensible ranges of physical parameters, and cobaya also runs safely with the standard mcmc(with fast-dragging) sampler. But I want to use polychord since I suspect that my model might have some multi-modal distributions. Is this a normal situation, so do I have to wait more days? Or should I change the constraints of some of the parameters, maybe tau?

Any comments or help would be appreciated. Thank you in advance.

Jiwon Park