nums.core.linalg module

nums.core.linalg.cholesky(app, X)

nums.core.linalg.direct_tsqr(app, X, reshape_output=True)

nums.core.linalg.fast_linear_regression(app, X, y)

nums.core.linalg.indirect_tsqr(app, X, reshape_output=True)

nums.core.linalg.indirect_tsr(app, X, reshape_output=True)

nums.core.linalg.inv(app, X)

nums.core.linalg.inv_uppertri(app, X)

Distributed algorithm for the inversion of an Upper Triangular Matrix. We use the method and notation described in - https://www.cs.utexas.edu/users/flame/pubs/siam_spd.pdf.

The upper-triangular matrix X is partitioned as follows:

_________________ | R_00 ‖ R_01 | R_02 | | R_TL | R_TR | |==========================|

X = R -> |---------------| -> | 0 ‖ R_11 | R_12 |

0 | R_BR | |--------------------------|

⎻⎻⎻⎻⎻⎻⎻⎻⎻⎻⎻⎻⎻⎻⎻⎻⎻ | 0 ‖ 0 | R_22 |

⎻⎻⎻⎻⎻⎻⎻⎻⎻⎻⎻⎻⎻⎻⎻⎻⎻⎻⎻⎻⎻⎻⎻⎻⎻⎻⎻⎻

where the double lines represent the boundaries between R_TL, R_TR, and R_BR.

nums.core.linalg.linear_regression(app, X, y)

nums.core.linalg.pca(app, X)

nums.core.linalg.qr(app, X)

nums.core.linalg.ridge_regression(app, X, y, lamb)

nums.core.linalg.svd(app, X)