Filtering Models

by: Chris Cave & Meera Machado
(Feyn version 2.0 or newer)

You can filter a list of Models you've sampled from a QLattice, to only include Models that you're interested in. This can be achieved using the builtin Python filter function. Some of the more common use cases require some code, so we've included convenience functions to help.

Here's a code snippet to sample some models we can filter on:

import feyn

# sample a list of models
ql = feyn.connect_qlattice()
models = ql.sample_models(
    input_names = ["Hello", "World"],
    output_name = "Hola!"
)

Here's a list of filters that feyn ships with, but you can also make your own:

Complexity

Contains Inputs

Contains Functions

Exclude Functions

Custom

Complexity

This filters Models of a certain complexity.

from feyn.filters import Complexity

# Get only models of complexity 2
f = Complexity(complexity=2)
f_models = list(filter(f, models))

ContainsInputs

This filters Models that contain the provided input(s).

from feyn.filters import ContainsInputs

# Get models that contain the "World" input
f = ContainsInputs("World")
f_models = list(filter(f, models))

# Get models that contain both "Hello" and "World" inputs
f = ContainsInputs(["Hello", "World"])
f_models = list(filter(f, models))

ContainsFunctions

This filters Models that only contain the provided function(s).

from feyn.filters import ContainsFunctions

# Get models that contains only the "Gaussian" function
f = ContainsFunctions("gaussian")
f_models = list(filter(f, models))

# Get models that contain only the "Gaussian" and "log" function
f = ContainsFunctions(["gaussian", "log"])
f_models = list(filter(f, models))

ExcludeFunctions

This filters Models that exclude the provided function(s).

from feyn.filters import ExcludeFunctions

# Get models that does not contains the "Gaussian" function
f = ExcludeFunctions("gaussian")
f_models = list(filter(f, models))

# Get models that do not contain either a "Gaussian" or a "log" function
f = ExcludeFunctions(["gaussian", "log"])
f_models = list(filter(f, models))

Custom filter

You can supply any function with the following signature to create your own filter:

def my_custom_filter(model):
    return "World" in model.inputs

# Get models that contains the "World" input
f_models = list(filter(my_custom_filter, models))