cr.sparse.cvx.focuss.matrix_solve_noiseless

cr.sparse.cvx.focuss.matrix_solve_noiseless(Phi, y, p=1.0, max_iters=20)[source]

Solves the sparse recovery problem using FOCUSS.

Parameters

  • Phi (jax.numpy.ndarray) – A sensing matrix / dictionary

  • y (jax.numpy.ndarray) – Measurements

  • max_iters (int) – Maximum number of iterations

  • p (float) – norm type [ p <= 1.]

FOCUSS (FOcal Underdetermined System Solver) uses an iterated reweighted least squares method. The algorithm is described in [Ela10] (section 3.2.1).

Examples