Model

Model(**kwargs)

Abstract model class to calculate a component’s S matrix.

Models must be derived from this class and implement the method start.

Additionally, for model parameterization to work correctly, the __init__ method must call super().__init__ with all parameters as keyword arguments (see example below). By default, models are updated by calling their __init__ method with the new keyword values. If a different behavior is required, the __init__ method must replace parametric_kwargs with the proper values for the user-defined update function and keywords. The update function must update the model in place and also update parametric_kwargs when called.

Finally, the derived class must be registered with register_model_class before it can be used in a phf file.

Important

The following attributes and methods defined in the base class should never be redefined in the derived one: s_matrix, setup_time_stepper, update, time_stepper, parametric_function, parametric_kwargs, and random_variables.

Example

>>> class CustomModel(pf.Model):
...     def __init__(self, *, coeff):
...         super().__init__(coeff=coeff)
...         self.coeff = coeff
...
...     def start(self, component, frequencies, **kwargs):
...         # Do any type of model calculation
...         s_param = numpy.exp(self.coeff * frequencies)
...         s = {("port_in@mode_in", "port_out@mode_out"): s_param}
...         return pf.SMatrix(s)
...
...
>>> pf.register_model_class(CustomModel)
>>> model = CustomModel(coeff=5e-15j)

Methods

Method	Description
autograd_smatrix	Compute an autograd-compatible S matrix for traced parameters.
estimate_cost	Estimate the cost for S matrix computation.
s_matrix	Compute the S matrix for a component using this model.
setup_time_stepper	Obtain a time stepper for a component using this model.
start	Start the S matrix computation for a component using this model.
transform	Apply a transformation to this model.
update	Update this model.

autograd_smatrix

autograd_smatrix(
    self,
    *,
    component,
    traced_params,
    frequencies,
    component_for_params,
    param_names,
    show_progress=True,
    model_kwargs=None,
)

Compute an autograd-compatible S matrix for traced parameters.

Important

This is an abstract method that must be implemented by derived classes that support autograd-based S-matrix computation.

Args

• component (Component): Nominal component associated with this autograd evaluation.
• traced_params (ndarray): Traced parameter vector used by autograd.
• frequencies (Sequence[float]): Frequency values at which to calculate the scattering parameters (in Hz).
• component_for_params (Callable[[ndarray], Component]): Callback that returns the component for a given traced parameter vector.
• param_names (Sequence[str]): Flattened traced parameter names aligned with traced_params.
• show_progress (bool): Progress display hint from Component.s_matrix.
• model_kwargs (dict[str, Any] | None): Model-specific keyword arguments forwarded from Component.s_matrix.

Returns

AutogradSMatrix: Autograd-compatible S matrix.

estimate_cost

estimate_cost(self, *args, **kwargs)

Estimate the cost for S matrix computation.

All arguments are passed to the model’s start method with the addition of the keyword argument cost_estimation=True.

Returns

float: Cost estimation.

s_matrix

s_matrix(self, component, frequencies, show_progress=True, model_kwargs={})

Compute the S matrix for a component using this model.

Args

• component (Component): Component to perform the calculation.
• frequencies (Sequence[float]): Frequency values at which to calculate the scattering parameters (in Hz).
• show_progress (bool): If True, show computation progress.
• model_kwargs (dict[str, object]): Keyword arguments passed to Model.start. Accepted arguments depend on the model type.

Returns

SMatrix: Computed S matrix.

Example

Compute only the responses from mode 0 of P0 with Tidy3D to reduce the number and cost of simulations:

>>> model = pf.Tidy3DModel()
>>> s_matrix = model.s_matrix(  # doctest: +SKIP
...     component,
...     frequencies,
...     model_kwargs={"inputs": ("P0@0",)},
... )

A port name such as ("P0",) selects all modes of that port. All outputs for the selected inputs are returned; responses from unselected inputs are not computed. Use Model.estimate_cost with the same inputs passed directly to estimate the reduced run.

setup_time_stepper

setup_time_stepper(
    self,
    component,
    time_step,
    carrier_frequency=0,
    show_progress=True,
    time_stepper_kwargs={},
)

Obtain a time stepper for a component using this model.

Args

• component (Component): Component for the time stepper.
• time_step (float): The interval between time steps (in seconds).
• carrier_frequency (float): The carrier frequency used to construct the time stepper. The carrier should be omitted from the input signals, as it is handled automatically by the time stepper.
• show_progress (bool): If True, show computation progress.
• time_stepper_kwargs (dict[str, object]): Keyword arguments passed to TimeStepper.setup. Each time stepper type can accept a different set of arguments. Check the documentation for the each time stepper’s setup_state method for more information.

Returns

TimeStepper: Initialized time stepper.

start

start(self, component, frequencies, **kwargs)

Start the S matrix computation for a component using this model.

Important

This is an abstract method that must be implemented by derived classes.

If the S matrix calculation can be performed synchronously, this method should return it directly.

Otherwise, it must return an object with attributes status and s_matrix that will be pooled to check the calculation progress. The status attribute must return a dictionary with 'progress' (a number from 0 to 100 indicating the calculation progress) and 'message' (one of 'running', 'success', or 'error', indicating the current calculation status). The 's_matrix' attribute should return the computed SMatrix once the calculation is successful (or None otherwise).

Important

Once the start method returns, there are no guarantees that the components, technologies or any other objects used as arguments will remain constant. That means no references to the original arguments should be kept after the method returns if the S matrix computation will continue latter, because they might change externally before the computation finishes. If needed, copies of the required objects must be used.

Args

• component (Component): Component to perform the calculation.
• frequencies (Sequence[float]): Frequency values at which to calculate the scattering parameters (in Hz).
• **kwargs (object): Keyword arguments forwarded from the Component.s_matrix call plus others generated by calling models (such as CircuitModel).

Returns

SMatrix | object: Object with status and s_matrix.

transform

transform(
    self, translation=(0, 0), rotation=0, scaling=1, x_reflection=False
)

Apply a transformation to this model.

The base implementation is a no-op and returns self. Derived classes can override this method to transform model-specific geometry settings.

Args

• translation (Sequence[float] | complex): Translation vector.
• rotation (float): Rotation angle in degrees.
• scaling (float): Magnification factor.
• x_reflection (bool): Mirror across the horizontal axis.

Returns

Model: This object with transformations applied.

update

update(self, *args, **kwargs)

Update this model.

All contents of the model are updated with the contents from the updated version. object contents or modifications introduced after the model creation (outside of the original parametric function) are not propagated.

The model’s time stepper does not participate in the update, unless the __init__ method of the derived model class replaces it via set_time_stepper.

Args

• *args (object): Positional arguments to the parametric model.
• **kwargs (object): Keyword arguments to the parametric model.

Returns

Model: This model.

Note

The original keyword arguments used to generate the model are stored and updated with the passed kwargs. This updated version is used in the function call and will be stored for future updates. Positional arguments are not stored.

Attributes

Attribute	Description
parametric_function	str: Function used to update the model.
parametric_kwargs	dict[str, object]: Keyword arguments used to update the model.
properties	Properties: Object properties.
random_variables	list[RandomVariable]: Random variables associated to the model’s parameters.
time_stepper	TimeStepper | None: Time stepper associated with this model.

parametric_function

str: Function used to update the model.

parametric_kwargs

dict[str, object]: Keyword arguments used to update the model.

properties

Properties: Object properties.

random_variables

list[RandomVariable]: Random variables associated to the model’s parameters.

time_stepper

TimeStepper | None: Time stepper associated with this model.