uncurl.scalable package¶
Submodules¶
uncurl.scalable.state_estimation module¶
State estimation using SGD
(how to do it?)
TODO: be able to use sparse (CSC) matrices
Basically, we observe one (cell_id, gene_level) pair at a time, iterating through the data point by point, updating the gradient based on that point.
-
uncurl.scalable.state_estimation.cost_grad(th, Xr, X, n)¶ translated from the matlab
-
uncurl.scalable.state_estimation.m_grad(m, X, w)¶
-
uncurl.scalable.state_estimation.poisson_estimate_state(data, clusters, init_means=None, init_weights=None, max_iters=10, tol=0.0001, eta=0.0001, disp=True)¶ Runs Poisson state estimation on a sparse data matrix...
-
uncurl.scalable.state_estimation.w_grad(w, X, m)¶