Skip to content

openscm_calibration.minimize#

Minimisation helpers

Functions:

Name Description
to_minimize_full

Calculate cost for given set of parameters

to_minimize_full #

to_minimize_full(
    x: NDArray[number[Any]],
    cost_calculator: SupportsCostCalculation[DataContainer],
    model_runner: SupportsModelRun[DataContainer],
    store: OptResStore[DataContainer] | None = None,
    known_error: type[ValueError] | None = None,
) -> float

Calculate cost for given set of parameters

This is a function that can be minimised with scipy

Before passing to scipy, all the arguments except x should be filled using e.g. :func:functools.partial

Parameters:

Name Type Description Default
x NDArray[number[Any]]

Parameter array

 required
cost_calculator SupportsCostCalculation[DataContainer]

Calculator of the cost function for each set of model results

 required
model_runner SupportsModelRun[DataContainer]

Model runner

 required
store OptResStore[DataContainer] | None

Store of results at each step in the optimisation (useful for plotting)

 None
known_error type[ValueError] | None

Known error that can occur when solving. If any other error is encountered, it is raised rather than being allowed to pass.

 None

Returns:

Type Description
Cost function for array ``x``
Source code in src/openscm_calibration/minimize.py 
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
def to_minimize_full(
    x: np.typing.NDArray[np.number[Any]],
    cost_calculator: SupportsCostCalculation[DataContainer],
    model_runner: SupportsModelRun[DataContainer],
    store: OptResStore[DataContainer] | None = None,
    known_error: type[ValueError] | None = None,
) -> float:
    """
    Calculate cost for given set of parameters

    This is a function that can be minimised with scipy

    Before passing to scipy, all the arguments except ``x`` should be filled
    using e.g. :func:`functools.partial`

    Parameters
    ----------
    x
        Parameter array

    cost_calculator
        Calculator of the cost function for each set of model results

    model_runner
        Model runner

    store
        Store of results at each step in the optimisation (useful for plotting)

    known_error
        Known error that can occur when solving. If any other error is
        encountered, it is raised rather than being allowed to pass.

    Returns
    -------
        Cost function for array ``x``
    """
    if known_error:
        try:
            model_results = model_runner.run_model(x)
        except known_error:
            cost = np.inf
            if store:
                store.note_failed_run(cost, x)
            return cost
    else:
        model_results = model_runner.run_model(x)

    cost = cost_calculator.calculate_cost(model_results)
    if store:
        store.append_result_cost_x(model_results, cost, x)

    return cost