import pandas as pd
import itertools
from itertools import combinations
[docs]
def single_parameter_sweep(
base_design_configuration: dict,
parameter_name: str,
parameter_sweep_values: list,
):
"""
This function takes a base_design_configuration dictionary and sweeps a single parameter over a list of values. It returns a list of dictionaries that correspond to the parameter sweep.
Args:
base_design_configuration(dict): Base design configuration dictionary.
parameter_name(str): Name of parameter to sweep.
parameter_sweep_values(list): List of values to sweep.
Returns:
parameter_sweep_design_dictionary_array(list): List of dictionaries that correspond to the parameter sweep.
"""
parameter_sweep_design_dictionary_array = []
for parameter in parameter_sweep_values:
design = base_design_configuration.copy()
design[parameter_name] = parameter
parameter_sweep_design_dictionary_array.extend([design])
return parameter_sweep_design_dictionary_array
[docs]
def multi_parameter_sweep(
base_design_configuration: dict, parameter_sweep_dictionary: dict
) -> list:
"""
This multiparameter sweep is pretty cool, as it will generate designer list of dictionaries that comprise of all the possible combinations of your parameter sweeps. For example, if you are sweeping `parameter_1 = np.arange(0, 2) = array([0, 1])`, and `parameter_2 = np.arange(2, 4) = array([2, 3])`, then this function will generate list of dictionaries based on the default_design dictionary, but that will comprise of all the potential parameter combinations within this list.
For the example above, there arould be 4 combinations [(0, 2), (0, 3), (1, 2), (1, 3)].
If you were instead sweeping for `parameter_1 = np.arange(0, 5)` and `parameter_2 = np.arange(0, 5)`, the dictionary generated would correspond to these parameter combinations of::
[(0, 0), (0, 1), (0, 2), (0, 3), (0, 4), (1, 0), (1, 1), (1, 2), (1, 3), (1, 4), (2, 0), (2, 1), (2, 2), (2, 3), (2, 4), (3, 0), (3, 1), (3, 2), (3, 3), (3, 4), (4, 0), (4, 1), (4, 2), (4, 3), (4, 4)].
Make sure to use the parameter_names from default_design when writing up the parameter_sweep dictionary key name.
Example project_structure formats::
example_parameter_sweep_dictionary = {
"parameter_1": np.arange(1, -40, 1),
"parameter_2": np.arange(1, -40, 1),
}
example_base_design_configuration = {
"parameter_1": 10.0,
"parameter_2": 40.0,
"parameter_3": 0,
}
Args:
base_design_configuration(dict): Dictionary of the default design configuration.
parameter_sweep_dictionary(dict): Dictionary of the parameter sweep. The keys should be the same as the keys in the base_design_configuration dictionary.
Returns:
parameter_sweep_design_dictionary_array(list): List of dictionaries that comprise of all the possible combinations of your parameter sweeps.
"""
parameter_names_sweep_list = []
parameter_sweep_values_list = []
parameter_sweep_design_dictionary_array = []
for parameter_sweep_name in parameter_sweep_dictionary.keys():
parameter_names_sweep_list.extend([parameter_sweep_name])
parameter_sweep_values_list.extend(
[parameter_sweep_dictionary[parameter_sweep_name].tolist()]
)
sweep_combinations = list(itertools.product(*parameter_sweep_values_list))
for parameter_combination in sweep_combinations:
design = base_design_configuration.copy()
parameter_index = 0
for parameter in parameter_combination:
design[parameter_names_sweep_list[parameter_index]] = parameter
parameter_index += 1
parameter_sweep_design_dictionary_array.extend([design])
return parameter_sweep_design_dictionary_array
[docs]
def get_unique_dataframe_subsets(
dataframe: pd.DataFrame, max_depth: int = 2
) -> dict[str, pd.DataFrame]:
"""
This function takes a pandas DataFrame and returns a dictionary of unique subsets of the DataFrame.
It is useful for identifying unique operating points in a dataset.
The function returns a dictionary where the keys are string identifiers for the subsets, and the values are DataFrames containing
the unique combinations of values for each column in the input DataFrame.
Parameters
----------
dataframe : pd.DataFrame
max_depth : int, optional
The maximum depth of combinations to consider. For example, if max_depth is 2, only combinations of up to 2 columns will be generated.
Returns
-------
dict[str, pd.DataFrame]
A dictionary where the keys are string identifiers for the subsets and the values are DataFrames containing the unique combinations of values.
"""
def create_identifier(columns, elements):
"""
This function takes a list of columns and a list of elements and returns a string identifier for the subset.
"""
return "_".join([f"{col}_{val}" for col, val in zip(columns, elements)])
subset_dictionary = {}
max_depth = max_depth or len(
dataframe.columns
) # Set max_depth to the number of columns if not provided
for r in range(1, min(max_depth, len(dataframe.columns)) + 1):
for column_combination in combinations(dataframe.columns, r):
unique_combinations = (
dataframe[list(column_combination)].drop_duplicates().values
)
for unique_value_combination in unique_combinations:
subset = dataframe[
(
dataframe[list(column_combination)] == unique_value_combination
).all(axis=1)
]
if len(subset) >= 2:
identifier = create_identifier(
column_combination, unique_value_combination
)
subset_dictionary[identifier] = subset.copy()
return subset_dictionary
