Pull out a submatrix from a COO SciPy sparse matrix.

gistfile1.py

import scipy as S

def coo_submatrix_pull(matr, rows, cols):
    """
    Pulls out an arbitrary i.e. non-contiguous submatrix out of
    a sparse.coo_matrix. 
    """
    if type(matr) != S.sparse.coo_matrix:
        raise TypeError('Matrix must be sparse COOrdinate format')
    
    gr = -1 * N.ones(matr.shape[0])
    gc = -1 * N.ones(matr.shape[1])
    
    lr = len(rows)
    lc = len(cols)
    
    ar = N.arange(0, lr)
    ac = N.arange(0, lc)
    gr[rows[ar]] = ar
    gc[cols[ac]] = ac
    mrow = matr.row
    mcol = matr.col
    newelem = (gr[mrow] > -1) & (gc[mcol] > -1)
    newrows = mrow[newelem]
    newcols = mcol[newelem]
    return S.sparse.coo_matrix((matr.data[newelem], N.array([gr[newrows],
        gc[newcols]])),(lr, lc))

Very helpful :)
you forgot to import numpy, though ;)

That I most certainly did!

Thanks!, super useful