Getting started¶

To illustrate the algorithm in this package, simulated data will be used. Functionality to create simulated matrix layouts is provided.

In [ ]:

from numpy.random import default_rng
from sklearn import set_config

import solrcmf

set_config(display="text")  # show text representation of sklearn estimators

# Control randomness
rng = default_rng(42)

# The function below generates sparse orthogonal matrices V_i and uses the
# supplied D_ij to form signal matrices V_i D_ij V_j^T. Noise with
# residual variance controlled by the signal-to-noise ratio is added.
xs_sim = solrcmf.simulate(
    viewdims={0: 100, 1: 50, 2: 50},
    factor_scales={
        (0, 1): [7.0, 5.1, 4.6, 0.0, 0.0],
        (0, 2): [8.3, 0.0, 0.0, 5.5, 0.0],
        (1, 2, 0): [6.3, 0.0, 4.7, 0.0, 5.1],
        (1, 2, 1): [0.0, 8.6, 4.9, 0.0, 0.0],
    },
    factor_sparsity={0: 0.25, 1: 0.25, 2: 0.25},
    snr=0.5,
    rng=rng,
)


# `xs_sim` is a dictionary containing
# - "xs_truth", the true signal matrices
# - "xs", the noisy data
# - "vs", the simulated orthogonal factors

from numpy.random import default_rng from sklearn import set_config import solrcmf set_config(display="text") # show text representation of sklearn estimators # Control randomness rng = default_rng(42) # The function below generates sparse orthogonal matrices V_i and uses the # supplied D_ij to form signal matrices V_i D_ij V_j^T. Noise with # residual variance controlled by the signal-to-noise ratio is added. xs_sim = solrcmf.simulate( viewdims={0: 100, 1: 50, 2: 50}, factor_scales={ (0, 1): [7.0, 5.1, 4.6, 0.0, 0.0], (0, 2): [8.3, 0.0, 0.0, 5.5, 0.0], (1, 2, 0): [6.3, 0.0, 4.7, 0.0, 5.1], (1, 2, 1): [0.0, 8.6, 4.9, 0.0, 0.0], }, factor_sparsity={0: 0.25, 1: 0.25, 2: 0.25}, snr=0.5, rng=rng, ) # `xs_sim` is a dictionary containing # - "xs_truth", the true signal matrices # - "xs", the noisy data # - "vs", the simulated orthogonal factors

In [ ]: