littlemcmc.sample

littlemcmc.sample(logp_dlogp_func: Callable[[numpy.ndarray], Tuple[numpy.ndarray, numpy.ndarray]], model_ndim: int, draws: int = 1000, tune: int = 1000, step: Union[littlemcmc.nuts.NUTS, littlemcmc.hmc.HamiltonianMC] = None, init: str = 'auto', chains: Optional[int] = None, cores: Optional[int] = None, start: Optional[numpy.ndarray] = None, progressbar: Union[bool, str] = True, random_seed: Union[int, List[int], None] = None, discard_tuned_samples: bool = True, chain_idx: int = 0, callback=None, mp_ctx=None, pickle_backend: str = 'pickle', **kwargs)

Draw samples from the posterior using the given step methods.

Parameters:
logp_dlogp_func: Python callable

Python callable that returns a tuple of the model joint log probability and its derivative, in that order.

model_ndim: int

The number of parameters of the model.

draws: int

The number of samples to draw. Defaults to 1000. The number of tuned samples are discarded by default. See discard_tuned_samples.

tune: int

Number of iterations to tune, defaults to 1000. Samplers adjust the step sizes, scalings or similar during tuning. Tuning samples will be drawn in addition to the number specified in the draws argument, and will be discarded unless discard_tuned_samples is set to False.

step: function

A step function. By default the NUTS step method will be used.

init: str
Initialization method to use for auto-assigned NUTS samplers.
  • auto: Choose a default initialization method automatically. Currently, this is jitter+adapt_diag, but this can change in the future. If you depend on the exact behaviour, choose an initialization method explicitly.
  • adapt_diag: Start with a identity mass matrix and then adapt a diagonal based on the variance of the tuning samples.
  • jitter+adapt_diag: Same as adapt_diag, but add uniform jitter in [-1, 1] to the starting point in each chain.
  • adapt_full: Same as ‘adapt_diag’, but adapt a dense mass matrix using the sample covariances.
chains: int

The number of chains to sample. Running independent chains is important for some convergence statistics and can also reveal multiple modes in the posterior. If None, then set to either cores or 2, whichever is larger.

cores: int

The number of chains to run in parallel. If None, set to the number of CPUs in the system, but at most 4.

start: dict, or array of dict

Starting point in parameter space. Initialization methods for NUTS (see init keyword) can overwrite the default.

progressbar: bool, optional default=True

Whether or not to display a progress bar in the command line. The bar shows the percentage of completion, the sampling speed in samples per second (SPS), and the estimated remaining time until completion (“expected time of arrival”; ETA).

random_seed: int or list of ints

A list is accepted if cores is greater than one.

discard_tuned_samples: bool

Whether to discard posterior samples of the tune interval.
Returns:
trace: np.array

An array that contains the samples.

stats: dict

A dictionary that contains sampler statistics.

Notes

Optional keyword arguments can be passed to sample to be delivered to the ``step_method``s used during sampling. In particular, the NUTS step method accepts a number of arguments. You can find a full list of arguments in the docstring of the step methods. Common options are:

  • target_accept: float in [0, 1]. The step size is tuned such that we approximate this acceptance rate. Higher values like 0.9 or 0.95 often work better for problematic posteriors.
  • max_treedepth: The maximum depth of the trajectory tree.
  • step_scale: float, default 0.25. The initial guess for the step size scaled

down by \(1/n**(1/4)\).