CAMB: accuracy settings for omk derivative

[Graeme Addison]

Hi, I'm trying to estimate numerical derivatives of the form dCl/dparam where Cl are the CMB theory power spectra. I'm using a simple finite difference method and while this works fine in most cases I'm running into numerical issues for Omega_k even for 'small' non-zero values (e.g. 0.01). Do you have any recommendations for how to change accuracy settings to help with this? I'm playing around with AccuracyBoost, lSampleBoost and lAccuracyBoost in python camb but the computations are individually time consuming and I thought I'd check if there is more specific advice...