doubly_constrained_furness

caf.distribute.furness.doubly_constrained_furness(seed_vals, row_targets, col_targets, tol=1e-09, max_iters=5000, warning=True)

Perform a doubly constrained furness for max_iters or until tol is met.

Controls numpy warnings to warn of any overflow errors encountered

Parameters:

• seed_vals (ndarray) – Initial values for the furness. Must be of shape (len(n_rows), len(n_cols)).

• row_targets (ndarray) – The target values for the sum of each row. i.e np.sum(matrix, axis=1)

• col_targets (ndarray) – The target values for the sum of each column i.e np.sum(matrix, axis=0)

• tol (float) – The maximum difference between the achieved and the target values to tolerate before exiting early. R^2 is used to calculate the difference.

• max_iters (int) – The maximum number of iterations to complete before exiting.

• warning (bool) – Whether to print a warning or not when the tol cannot be met before max_iters.

Returns:

• furnessed_matrix – The final furnessed matrix

• completed_iters – The number of completed iterations before exiting

• achieved_rmse – The Root Mean Squared Error difference achieved before exiting

Return type:

tuple[ndarray, int, float]