distrib¶

The formulas of the distributions of the fitted parameters, without a simulator.

A fitted parameter carries the distribution of its draws as a formula of the SBML package distrib (pkpdsim.uncertainty.ParameterUncertainty.distribution): normal(mean, sd[, lower, upper]), lognormal(mu, sigma[, lower, upper]) (mu and sigma of the natural logarithm) and uniform(lower, upper), with numbers as arguments, truncated to the bounds of the fit when they are given. This module is the one grammar of these formulas: the draws of pkpdsim.uncertainty parse them into the distributions of sbmlsim, and the PBPK library reads their support to bound a fitted fraction (it must not import a simulator).

ARGUMENTS `module-attribute` ¶

ARGUMENTS = {
    "normal": (2, 4),
    "lognormal": (2, 4),
    "uniform": (2,),
}

The distributions and their numbers of arguments; the last two of four are the bounds of a truncation.

DistribFormula `dataclass` ¶

DistribFormula(name, args)

A formula of a distribution of distrib.

Attributes:

Name	Type	Description
`name`	`str`	the distribution, a key of `ARGUMENTS`.
`args`	`tuple[float, ...]`	the arguments.

bounds `property` ¶

bounds

The bounds of a truncated or uniform distribution, None without.

parse ¶

parse(formula)

Parse a formula of a distribution.

Parameters:

Name	Type	Description	Default
`formula`	`str`	e.g. `normal(0.3, 0.05, 0.01, 0.6)`.	required

Returns:

Type	Description
`DistribFormula`	The distribution and its arguments.

Raises:

Type	Description
`ValueError`	if the formula is no function of numbers, or no distribution of `ARGUMENTS` with a supported number of arguments.

support ¶

support(formula)

Get the support of a distribution: the values its draws can take.