`Optimizer`

class nexus.Optimizer(measurements, id='', external_fit_variables=[], inequalities=None)

Constructor for the abstract Optimizer class. The optimizer is used to minimize a certain residual value. The actual implementation has to be provided by the user in python the following way.

# A new python class is derived from the nexus Optimizer class.
class DerivedOptimizerClass(nx.Optimizer):
    def __init__(self, measurements, id):
        super().__init__(measurements, id)

    # implementation of the residual
    def Residual(self):
        # here goes your code which calculates the residual to be minimized
        # it must depend on measurement class(es) object with fittable Var(s)
        ...
        residual = ...
        # the return value must be of type float
        return residual

# Set up a optimization scenario

# There must exist nx.Var object(s) that should be optimized
optimization_parameter = nx.Var(..., fit = True)
...

# There are measurement class object(s) defined which depend on the optimization parameter(s)
measurement_type = nx.MeasurementClassType(...)

# initialize a derived optimizer class object
# pass a list with measurement objects on which the residual implementation Residual() depends
my_optimizer = DerivedOptimizerClass(measurements = [measurement_type, ...], id = "optimizer id")

# calculate the residual value if you want to check your code and print
print(my_optimizer.Residual())

# run the optimizer to optimize all nx.Var object(s) on which the residual depends
my_optimizer() # or my_optimizer.Evaluate()

You can also pass the negative of the value to maximize. For more information have a look to the tutorial.

Parameters:

measurements (list) – List of Measurement objects used in the Optimizer.
id (string) – User identifier.
external_fit_variables (list) –
List of Var objects. These Vars are not directly assigned to a Nexus object but can be related to an equality constraint on a Var. Nexus is not able to detect those automatically, so they have to provided manually.

Added in version 1.1.0.
inequalities (Inequality) –
An Inequality object. All inequalities constraints are handled by this object.

Added in version 1.1.0.

measurements

List of Measurement objects used in the Optimizer.

Type:: list

id

User identifier.

Type:: string

external_fit_variables

List of Var objects. These Vars are not directly assigned to a Nexus object but can be related to an equality constraint on a Var. Nexus is not able to detect those automatically, so they have to provided manually.

Added in version 1.1.0.

Type:: list

inequalities

An Inequality object. All inequalities constraints are handled by this object.

Added in version 1.1.0.

Type:: Inequality

options

Options that specify how the optimization is performed.

Type:: OptimizerOptions

start_residual

The residual value at the beginning of optimization.

Added in version 1.1.0.

Type:: float

residual

The residual value to be optimized.

Type:: float

initial_parameters

List with initial values of the fit parameters.

Type:: list

lower_bounds

List with min values of the fit parameters.

Type:: list

upper_bounds

List with max values of the fit parameters.

Type:: list

ids

List with ids of the fit parameters.

Type:: list

Evaluate(): Evaluates the residual as implemented by the user. Also callable via operator ().

Residual()

Call of the residual implementation function.

Returns:: The residual value.
Return type:: float

Optimizer

`Optimizer`