elisa.models.parameter#

The parameter.

class AssignmentTracker[source]#

Bases: object

Track component assignment of a parameter.

Methods

`append`(cid, pname)	Append a new assignment record.
`get_comp_param`(comp_ids)	Get the earliest component assignment record within comp_ids.
`remove`(cid, pname)	Remove an assignment record.

append(cid: str, pname: str) → None[source]#: Append a new assignment record.

remove(cid: str, pname: str) → None[source]#: Remove an assignment record.

get_comp_param(comp_ids: Iterable[str]) → tuple[str, str] | None[source]#: Get the earliest component assignment record within comp_ids.

class ParamInfo(name: str | Callable[[ParamIDStrMapping], str], latex: str | Callable[[ParamIDStrMapping], str], default: Any, bound: str, prior: str, log: bool, fixed: bool, tracker: AssignmentTracker, id_to_value: Callable[[ParamIDValMapping], JAXFloat], dist: Distribution | None = None, composite: bool = False, integrate: IntegralFactory | bool = False)[source]#

Bases: NamedTuple

Parameter information.

Methods

`count`(value, /)	Return number of occurrences of value.
`index`(value[, start, stop])	Return first index of value.

name: str | Callable[[dict[str, str]], str]#

Plain name of the parameter.

It is a getter function for composite parameter.

latex: str | Callable[[dict[str, str]], str]#

\(\LaTeX\) format of the parameter.

It is a getter function for composite parameter.

default: Any#: Default value of the parameter.

bound: str#: Value bound expression of the parameter.

prior: str#: Prior distribution expression of the parameter.

log: bool#: Whether the parameter is parameterized in a logarithmic scale.

fixed: bool#: Whether the parameter is fixed.

tracker: AssignmentTracker#: Component assignment tracker.

id_to_value: Callable[[dict[str, Array]], Array]#: Mapping function from id to value.

dist: Distribution | None#: NumPyro distribution for the parameter.

composite: bool#: Whether the parameter is composite.

integrate: Callable[[Callable[[Array, dict[str, Array]], Array]], Callable[[Array, dict[str, Array]], Array]] | bool#

Integration factory for the parameter for interval parameter.

For composite parameter composed by interval parameter, this is True.

class Parameter[source]#

Bases: ABC

Parameter base.

Attributes

`default`	Default value of the parameter.
`fixed`	Whether the parameter is fixed.
`latex`	\(\LaTeX\) format of the parameter.
`log`	Whether the parameter is parameterized in a logarithmic scale.
`name`	Plain name of the parameter.

abstract property name: str#: Plain name of the parameter.

abstract property latex: str#: \(\LaTeX\) format of the parameter.

abstract property default: Array#: Default value of the parameter.

abstract property log: bool#: Whether the parameter is parameterized in a logarithmic scale.

abstract property fixed: bool#: Whether the parameter is fixed.

class ParameterHelper(name: str, latex: str | None, default: Any | None = None)[source]#

Bases: Parameter

Handle name, latex, and default value of a parameter.

Attributes

`default`	Default value of the parameter.
`fixed`	Whether the parameter is fixed.
`latex`	\(\LaTeX\) format of the parameter.
`log`	Whether the parameter is parameterized in a logarithmic scale.
`name`	Plain name of the parameter.

property name: str#: Plain name of the parameter.

property latex: str#: \(\LaTeX\) format of the parameter.

abstract property default: Any#: Default value of the parameter.

class DistParameter(name: str, dist: Distribution, default: float, *, min: float | None = None, max: float | None = None, log: bool = False, fixed: bool = False, latex: str | None = None)[source]#

Bases: ParameterHelper

Define a parameter by a distribution.

Parameters:

namestr: Parameter name.
distDistribution: Numpyro distribution from which the parameter is sampled.
defaultfloat: Parameter default value.
minfloat, optional: Parameter minimum value, for display purpose only. The default is None.
maxfloat, optional: Parameter maximum value, for display purpose only. The default is None.
logbool, optional: Whether the parameter is parameterized in a logarithmic scale. The default is False.
fixedbool, optional: Whether the parameter is fixed. The default is False.
latexstr, optional: \(\LaTeX\) format of the parameter. The default is as name.

Attributes

`default`	Default value of the parameter.
`fixed`	Whether the parameter is fixed.
`latex`	\(\LaTeX\) format of the parameter.
`log`	Whether the parameter is parameterized in a logarithmic scale.
`name`	Plain name of the parameter.

property default: Array#: Default value of the parameter.

property log: bool#: Whether the parameter is parameterized in a logarithmic scale.

property fixed: bool#: Whether the parameter is fixed.

class UniformParameter(name: str, default: float, min: float, max: float, *, log: bool = False, fixed: bool = False, latex: str | None = None)[source]#

Bases: DistParameter

Define the parameter by a uniform distribution.

Parameters:

namestr: Parameter name.
defaultfloat: Parameter default value.
minfloat: Parameter minimum value.
maxfloat: Parameter maximum value.
logbool, optional: Whether the parameter is logarithmically uniform. The default is False.
fixedbool, optional: Whether the parameter is fixed. The default is False.
latexstr, optional: \(\LaTeX\) format of the parameter. The default is as name.

Attributes

`default`	Default value of the parameter.
`fixed`	Whether the parameter is fixed.
`latex`	\(\LaTeX\) format of the parameter.
`log`	Whether the parameter is logarithmically uniform.
`max`	Parameter maximum value.
`min`	Parameter minimum value.
`name`	Plain name of the parameter.

property name: str#: Plain name of the parameter.

property latex: str#: \(\LaTeX\) format of the parameter.

property default: Array#: Default value of the parameter.

property min: Array#: Parameter minimum value.

property max: Array#: Parameter maximum value.

property log: bool#: Whether the parameter is logarithmically uniform.

property fixed: bool#: Whether the parameter is fixed.

class ConstantParameter(name: str, value: Any, latex: str | None = None)[source]#

Bases: ParameterHelper

Constant parameter.

Parameters:

namestr: Parameter name.
valuearray_like: The constant value of parameter.
latexstr, optional: \(\LaTeX\) format of the parameter. The default is as name.

Attributes

`default`	Default value of the parameter.
`fixed`	Constant parameter is fixed.
`latex`	\(\LaTeX\) format of the parameter.
`log`	Constant parameter is not logarithmically parameterized.
`name`	Plain name of the parameter.

property log: bool#: Constant parameter is not logarithmically parameterized.

property fixed: bool#: Constant parameter is fixed.

class ConstantValue(name: str, value: float, latex: str | None = None)[source]#

Bases: ConstantParameter

Parameter with a fixed value.

Parameters:

name: str: Parameter name.
value: float: Parameter value.
latexstr, optional: \(\LaTeX\) format of the parameter. The default is as name.

Attributes

`default`	Default value of the parameter.
`fixed`	Constant parameter is fixed.
`latex`	\(\LaTeX\) format of the parameter.
`log`	Constant parameter is not logarithmically parameterized.
`name`	Plain name of the parameter.

property default: Array#: Default value of the parameter.

class ConstantInterval(name: str, interval: Sequence[float], method: Literal['quadgk', 'quadcc', 'quadts', 'romberg', 'rombergts'] = 'quadgk', latex: str | None = None, **kwargs)[source]#

Bases: ConstantParameter

Constant parameter to be integrated over a given interval.

When assigning ConstantInterval parameters to a model component, the model will be evaluated according to the following formula:

\[\frac{1}{\prod_i (b_i - a_i)} \int f(E, \vec{\theta}(\vec{p}, \vec{q})) \, \mathrm{d} \vec{p}\]

where \(f\) is the model function, \(\vec{\theta}\) is the parameter vector of the model, \(\vec{p}\) is the ConstantInterval parameters, \(\vec{q}\) is the other parameters, and \(a_i\) and \(b_i\) are the intervals given by \(\vec{p}\).

Parameters:

name: str

Parameter name.

interval: array_like

The interval, a 2-element sequence.

method{‘quadgk’, ‘quadcc’, ‘quadts’, ‘romberg’, ‘rombergts’}, optional

Numerical integration method used to integrate over the parameter. Available options are:

'quadgk': global adaptive quadrature by Gauss-Konrod rule

'quadcc': global adaptive quadrature by Clenshaw-Curtis rule

'quadts': global adaptive quadrature by trapz tanh-sinh rule

'romberg': Romberg integration

'rombergts': Romberg integration by tanh-sinh (a.k.a. double exponential) transformation

The default is 'quadgk'.

latexstr, optional

\(\LaTeX\) format of the parameter. The default is as name.

kwargsdict, optional

Extra kwargs passed to integration methods. See [1] for details.

Attributes

`default`	Default value of the parameter.
`fixed`	Constant parameter is fixed.
`latex`	\(\LaTeX\) format of the parameter.
`log`	Constant parameter is not logarithmically parameterized.
`method`	Numerical integration method.
`name`	Plain name of the parameter.

References

[1]

quadax docs

property default: Array#: Default value of the parameter.

property method: Literal['quadgk', 'quadcc', 'quadts', 'romberg', 'rombergts']#: Numerical integration method.

class CompositeParameter(params: Parameter | Sequence[Parameter], op: Callable[[...], Array], op_name: str, op_latex: str | None = None, *, op_symbol: Literal['+', '-', '*', '/', '^'] | None = None)[source]#

Bases: Parameter

Combine parameters to create a new parameter.

Parameters:

paramsParameter, or sequence of Parameter: Parameters to be combined.
opcallable(): Function applied to params. The function should take the same number and order of arguments as params and return a single value. The function must be compatible with JAX.
op_namestr: Name of the composition operator op.
op_latexstr, optional: \(\LaTeX\) format of the composition operator op. The default is as op_name.

Attributes

`default`	Default value of the parameter.
`fixed`	If the sub-parameters are all fixed.
`latex`	\(\LaTeX\) format of the parameter.
`log`	If the sub-parameters are all logarithmically parameterized.
`name`	Plain name of the parameter.

property name: str#: Plain name of the parameter.

property latex: str#: \(\LaTeX\) format of the parameter.

property default: Array#: Default value of the parameter.

property log: bool#: If the sub-parameters are all logarithmically parameterized.

property fixed: bool#: If the sub-parameters are all fixed.