modules.applications.qml.generative_modeling.training.qcbm.QCBM

class QCBM

Bases: TrainingGenerative

This module optimizes the parameters of quantum circuit using CMA-ES. This training method is referred to as quantum circuit born machine (QCBM).

__init__()

Constructor method.

Methods

__init__()	Constructor method.
data_visualization(loss_epoch, pmfs_model, ...)	Visualizes the data and metrics for training.
get_available_submodule_options()	Gets the list of available options.
get_available_submodules(option)	If the module has submodules depending on certain options, this method should adjust the submodule_options accordingly.
get_default_submodule(option)	Raises ValueError as this module has no submodules.
get_depending_parameters(option, config)	If the module has parameters depending on certain options, this method should return the parameters for the given option.
get_parameter_options()	This function returns the configurable settings for the quantum circuit born machine.
get_requirements()	Returns requirements of this module.
get_submodule(option)	Submodule is instantiated according to the information given in self.sub_options.
kl_divergence(pmf_model, pmf_target)	This function calculates the Kullback-Leibler divergence, that is used as a loss function.
mmd(pmf_model, pmf_target)	This function calculates the maximum mean discrepancy, that is used as a loss function.
nll(pmf_model, pmf_target)	This function calculates th negative log likelihood, that is used as a loss function.
postprocess(input_data, config, **kwargs)	Perform the actual training of the machine learning model.
preprocess(input_data, config, **kwargs)	Essential method for the benchmarking process.
sample_from_pmf(pmf, n_shots)	This function samples from the probability mass function generated by the quantum circuit.
setup_training(input_data, config)	Method to configure the training setup including CMA-ES and tensorboard.
start_training(input_data, config, **kwargs)	This function finds the best parameters of the circuit on a transformed problem instance and returns a solution.

class Config

Bases: TypedDict

Attributes of a valid config.

population_size: int
max_evaluations: int
sigma: float
pretrained: str
loss: str

clear() → None.  Remove all items from D.

copy() → a shallow copy of D

fromkeys(value=None, /)

Create a new dictionary with keys from iterable and values set to value.

get(key, default=None, /)

Return the value for key if key is in the dictionary, else default.

items() → a set-like object providing a view on D's items

keys() → a set-like object providing a view on D's keys

pop(k[, d]) → v, remove specified key and return the corresponding value.

If the key is not found, return the default if given; otherwise, raise a KeyError.

popitem()

Remove and return a (key, value) pair as a 2-tuple.

Pairs are returned in LIFO (last-in, first-out) order. Raises KeyError if the dict is empty.

setdefault(key, default=None, /)

Insert key with a value of default if key is not in the dictionary.

Return the value for key if key is in the dictionary, else default.

update([E, ]**F) → None.  Update D from mapping/iterable E and F.

If E is present and has a .keys() method, then does: for k in E.keys(): D[k] = E[k] If E is present and lacks a .keys() method, then does: for k, v in E: D[k] = v In either case, this is followed by: for k in F: D[k] = F[k]

values() → an object providing a view on D's values

class Timing

Bases: object

This module is an abstraction of time measurement for both CPU and GPU processes.

start_recording_cpu() → None

This is a function to start time measurement on the CPU.

start_recording_gpu() → None

This is a function to start time measurement on the GPU.

stop_recording_cpu() → float

This is a function to stop time measurement on the CPU.

.return: Elapsed time in milliseconds

stop_recording_gpu() → float

This is a function to stop time measurement on the GPU.

Returns:

Elapsed time in milliseconds

data_visualization(loss_epoch: ndarray, pmfs_model: ndarray, samples: any, epoch: int) → ndarray

Visualizes the data and metrics for training.

Parameters:

• loss_epoch -- Loss for the current epoch

• pmfs_model -- The probability mass functions from the model

• samples -- The samples from the model

• epoch -- The current epoch number

Returns:

Best probability mass function for visualization

get_available_submodule_options() → list

Gets the list of available options.

Returns:

List of module options

get_available_submodules(option: list) → list

If the module has submodules depending on certain options, this method should adjust the submodule_options accordingly.

Parameters:

option -- List of chosen options

Returns:

List of available submodules

get_default_submodule(option: str) → Core

Raises ValueError as this module has no submodules.

Parameters:

option -- Option name

Raises:

ValueError -- If called, since this module has no submodules

get_depending_parameters(option: str, config: dict) → dict

If the module has parameters depending on certain options, this method should return the parameters for the given option.

Parameters:

• option -- The chosen option

• config -- Current config dictionary

Returns:

The parameters for the given option

get_parameter_options() → dict

This function returns the configurable settings for the quantum circuit born machine.

Returns:

Configuration settings for QCBM

return {

    "population_size": {
        "values": [5, 10, 100, 200, 10000],
        "description": "What population size do you want?"
    },

    "max_evaluations": {
        "values": [100, 1000, 20000, 100000],
        "description": "What should be the maximum number of evaluations?"
    },

    "sigma": {
        "values": [0.01, 0.5, 1, 2],
        "description": "Which sigma would you like to use?"
    },

    "pretrained": {
        "values": [False],
        "custom_input": True,
        "postproc": str,
        "description": "Please provide the parameters of a pretrained model."
    },

    "loss": {
        "values": ["KL", "NLL"],
        "description": "Which loss function do you want to use?"
    }
}

static get_requirements() → list[dict]

Returns requirements of this module.

Returns:

List of dict with requirements of this module

get_submodule(option: str) → Core

Submodule is instantiated according to the information given in self.sub_options. If self.sub_options is None, get_default_submodule is called as a fallback.

Parameters:

option -- String with the options

Returns:

Instance of a module

kl_divergence(pmf_model: ndarray, pmf_target: ndarray) → ndarray

This function calculates the Kullback-Leibler divergence, that is used as a loss function.

Parameters:

• pmf_model -- Probability mass function generated by the quantum circuit

• pmf_target -- Probability mass function of the target distribution

Returns:

Kullback-Leibler divergence

mmd(pmf_model: ndarray, pmf_target: ndarray) → ndarray

This function calculates the maximum mean discrepancy, that is used as a loss function.

Parameters:

• pmf_model -- Probability mass function generated by the quantum circuit

• pmf_target -- Probability mass function of the target distribution

Returns:

Maximum mean discrepancy

nll(pmf_model: ndarray, pmf_target: ndarray) → ndarray

This function calculates th negative log likelihood, that is used as a loss function.

Parameters:

• pmf_model -- Probability mass function generated by the quantum circuit

• pmf_target -- Probability mass function of the target distribution

Returns:

Negative log likelihood

postprocess(input_data: dict, config: dict, **kwargs) → tuple[dict, float]

Perform the actual training of the machine learning model.

Parameters:

• input_data -- Collected information of the benchmarking process

• config -- Training settings

• kwargs -- Optional additional arguments

Returns:

Training results and the postprocessing time

preprocess(input_data: any, config: dict, **kwargs) → tuple[any, float]

Essential method for the benchmarking process. This is always executed before traversing down to the next module, passing the data returned by this function.

Parameters:

• input_data -- Data for the module, comes from the parent module if that exists

• config -- Config for the module

• kwargs -- Optional keyword arguments

Returns:

The output of the preprocessing and the time it took to preprocess

sample_from_pmf(pmf: ndarray, n_shots: int) → ndarray

This function samples from the probability mass function generated by the quantum circuit.

Parameters:

• pmf -- Probability mass function generated by the quantum circuit

• n_shots -- Number of shots

Returns:

Number of counts in the 2**n_qubits bins

setup_training(input_data: dict, config: Config) → tuple[float, dict]

Method to configure the training setup including CMA-ES and tensorboard.

Parameters:

• input_data -- A representation of the quantum machine learning model that will be trained

• config -- Config specifying the parameters of the training

Returns:

Random initial parameter and options for CMA-ES

start_training(input_data: dict, config: dict, **kwargs: dict) → dict

This function finds the best parameters of the circuit on a transformed problem instance and returns a solution.

Parameters:

• input_data -- A representation of the quantum machine learning model that will be trained

• config -- Config specifying the parameters of the training

• kwargs -- Optional additional settings

Returns:

Dictionary including the information of previous modules as well as of the training