Parameters

Handling parameters requires some syntax. Below some parameter specification.

params:
  .fixed: 'al*'  # Fix all parameters matching al* pattern
  .varied: ['al2_2']  # Allow parameter al2_2 to vary
  qpar:
    value: 1.  # This is the value the parameter will take as default
    prior:  # Prior for this parameter is Gaussian, centered around 1, with standard deviation of 0.1
      dist: norm
      loc: 1.
      scale: 0.1
    ref:  # To start sampling/profiling, generate first parameter values in this distribution (defaults to prior)
      dist: norm
      loc: 1.
      scale: 0.01
    proposal: 0.02  # Proposal standard deviation for this parameter, used when setting initial covariance matrices (MCCMSampler)
    latex: '\alpha_\parallel'  # Latex
  qpar:
    prior:
      limits: [0.9, 1.1]  # uniform prior
 qiso:
    # A derived parameter, expressed as a function of the others
   derived: 'qper**(1./3.) * qper**(2./3.)'
   # One can specify a prior for this derived parameter:
  prior:
    limits: [0.95, 1.05]
  # If this parameter is not used by any calculator, drop it!
  drop: True
al*:
  derived: '.auto'  # parameter set by analytic marginalization for sampling, or best fit for profiling
power:
  # A derived parameter, as an attribute (or given by __getstate__() method) of this calculator
  derived: True