SVD approximation via random projections.
linalg_rsvd(seed, k, q, x, s, u = NULL, vt = NULL)
| seed | Seed for the random generator. |
|---|---|
| k | The number of components to estimate. Integer > 0. |
| q | Exponent (see paper if you really care). Values of 1 or 2 are good. |
| x | Input data. The input values are overwritten. |
| s | Singular values. |
| u, vt | The left/right singular vectors. Should both be |
Halko, Nathan, Per-Gunnar Martinsson, and Joel A. Tropp. "Finding structure with randomness: Probabilistic algorithms for constructing approximate matrix decompositions." SIAM review 53, no. 2 (2011): 217-288.