EmceeSampler (taurex_emcee.emcee_optimizer.EmceeSampler)
Emcee sampler for TauREx3.1.
- param observed
Sets the observation to optimize the model to
- type observed
BaseSpectrum, optional- param model
The forward model we wish to optimize
- type model
ForwardModel, optional- param sigma_fraction
Fraction of weights to use in computing the error. (Default: 0.1)
- type sigma_fraction
float, optional
- param num_global_samples
Number of samples to initially draw from the prior. Default is 10000
- type num_global_samples
int
- param num_chains
Number of independent ensembles to run. Default is 4
- type num_chains
int
- param num_walkers
Ensemble size. Default is max(100, 4 * dim)
- type num_walkers
int
- param max_ncalls
Maximum number of likelihood function evaluations. Default is 1000000
- type max_ncalls
int
- param growth_factor
Factor by which to increase the number of steps. Default is 10
- type growth_factor
int
- param max_improvement_loops
Number of times MCMC should be re-attempted. Default is 4
- type max_improvement_loops
int
- param num_initial_steps
Number of sampler steps to take in first iteration. Default is 100
- type num_initial_steps
int
- param min_autocorr_times
If positive, sets autocorelation as an additional convergence criterion. Default is 0
- type min_autocorr_times
int
- param rhat_max
Sets Gelman-Rubin diagnostic to converge. Default is 1.01
- type rhat_max
float
- param geweke_max
Sets Gelman-Rubin diagnostic to converge. Default is 2.0
- type geweke_max
float
- param progress
If True, show progress bars. Default is True
- type progress
bool