src package

Submodules

src.qml module

src.qml.kmeans2Qubo(X, k, alpha=None, beta=None)[source]

Converts a given k-means clustering problem into a QUBO problem

Parameters:

  • X (numpy array) – Data matrix with N rows and d columns

  • k (numpy array) – Number of clusters desired

Returns:

QUBO matrix p (numpy array): QUBO vector

Return type:

Q (numpy array)

Raises:

  • TypeError –

    1. If X is not numpy array 2. If k is not an integer

  • ValueError – If shape of X is not [N, d]

src.qml.kmeansClusters(solutions, N, k)[source]

Returns the clustering assignment for a given k-means problem

Parameters:

  • solutions (list) – List of k-means solutions returned by the adiabatic quantum computer

  • N (int) – Number of data points in the original problem

  • k (int) – Number of clusters in the original problem

Returns:

A list of assignments corresponding to each solution returned by the adiabatic quantum computer

Return type:

assignmentsList

src.qml.regression2Qubo(X, Y, P)[source]

Converts a given regression problem into QUBO problem

Parameters:

  • X (numpy array) – Augmented training data having N datapoints and (d+1) features, i.e. d features augmented by unity

  • Y (numpy array) – Training labels for regression, N dimensional vector

  • P (numpy array) – Precision vector, must be sorted, can contain positive and negative powers of 2

Returns:

QUBO matrix p (numpy array): QUBO vector

Return type:

Q (numpy array)

Raises:

  • TypeError – If X, Y or P are not numpy arrays

  • ValueError –

    1. If shape of X is not [N,d] 2. If shape of Y is not [N] 3. If first dimensions of X and Y are not equal

src.qml.regressionWeights(solutions, P)[source]

Finds regression weights from binarized regression weights

Parameters:

  • solutions – List of binarized regression weight vector

  • P – Precision vector

Returns:

List of real valued regression weights

Return type:

w

src.qml.svm2Qubo(X, Y, P)[source]

Converts a given SVM problem into QUBO problem

Parameters:

  • X (numpy array) – Training data having N datapoints and d features

  • Y (numpy array) – Training labels for binary classification

  • P (numpy array) – Precision vector, must be sorted, can contain positive and negative powers of 2

Returns:

QUBO matrix p (numpy array): QUBO vector

Return type:

Q (numpy array)

Raises:

  • TypeError – If X, Y or P are not numpy arrays

  • ValueError –

    1. If shape of X is not [N,d] 2. If shape of Y is not [N] 3. If first dimensions of X and Y are not equal

src.qml.svmWeights(solutions, P, N, d)[source]

Finds SVM weights from binarized SVM weights

Parameters:

  • solutions – List of binarized SVM solutions (weights, biases and Lagrangian multipliers)

  • P – Precision vector

  • N – Size of training dataset

  • d – Number of features in training dataset

Returns:

List of real valued SVM weights biasList: List of real valued SVM bias lambdas: List of real valued SVM Lagrangian multipliers

Return type:

weightsList

src.utils module

src.utils.array2Dict(x)[source]

Converts numpy 1-dimensional array into dictionary.

Parameters:

x (numpy array) – Input array

Returns:

Dictionary with array indices as keys and array values as values.

Return type:

d (dict)

src.utils.embedChimeraIsing(Q, p, M=16, N=16, L=4, Lambda=None, normalize=True)[source]

Generates embedding for an Ising problem characterized by Q and p for D-Wave Chimera hardware.

Parameters:

  • Q (numpy array) –

  • p (numpy array) –

  • M (int) –

  • N (int) –

  • L (int) –

Returns:

h (dict):

Return type:

J (dict)

Raises:

  • TypeError – If M, N or L are not integers

  • ValueError – If Lambda is given and not positive

src.utils.embedChimeraQubo(Q, p, M=16, N=16, L=4, Lambda=None, normalize=True)[source]

Generates embedding for an Ising problem characterized by Q and p for D-Wave Chimera hardware.

Parameters:

  • Q (numpy array) –

  • p (numpy array) –

  • M (int) –

  • N (int) –

  • L (int) –

Returns:

h (dict):

Return type:

J (dict)

Raises:

  • TypeError – If M, N or L are not integers

  • ValueError – If Lambda is given and not positive

src.utils.postProcessing(response, embeddings, A, b=None)[source]

Post-process the response obtained from D-Wave.

Parameters:

response – response object returned by D-Wave Ocean API

Returns:

List of (solution, energy, number of occurrences) tuples

Return type:

results

Module contents