Random Sampler

Random Sampler draws independent uniform samples in unit space (first sample in [0,1], then map to your configured parameter distributions).

Minimal example

Sampling:
  Method: "Random"
  Point number: 1000
  Variables:
    - name: x
      description: "example variable"
      distribution:
        type: Flat
        parameters:
          min: 0.0
          max: 1.0
  LogLikelihood:
    - name: L_total
      expression: "-0.5*(x-0.5)^2"

Field meanings

• Sampling: root configuration block.
• Method: must be Random.
• Point number: number of accepted samples to return. If a selection cut is used, more raw points may be generated internally to reach this accepted count.
• Variables: See Scan Parameters (Sampler Variables Schema)
• LogLikelihood: list of named likelihood terms. See Likelihood
  • name: likelihood term name.
  • expression: expression evaluated per point.

Purpose

Use the Random sampler when you want simple, independent sampling with an optional selection cut.

Random is not the right choice for posterior sampling, correlated exploration, or evidence estimation.

Required configuration

• Sampling.Method: Random
• Sampling.Point number: integer (accepted samples)
• Sampling.Variables: variable definitions
• Sampling.LogLikelihood: named likelihood expressions

Optional configuration

• Sampling.selection: selection cut expression that must evaluate to a boolean; it should remain simple, deterministic, and free of side effects. Rejected points are resampled.

LogLikelihood schema

Each entry in Sampling.LogLikelihood must contain:

• name: string
• expression: string

Keep selection simple and deterministic. Rejected points are resampled until the requested number of accepted points is reached.