LAGRANGIAN.M
Performs RPCA decomposition of input matrix.
i.e., solves
argmin_{L,S} ||L||_* + lambda.||shrink(epsilon,S)||_1
s.t. M-L-S=0
[L,S,iter]=lagrangian(M, epsilon, tol, maxIter, lambda-Scale, lambda, rho, mu)
M is the only mandatory argument.

GETSPARSE.M
Produces a sparse matrix of given sparsity
S = getSparse(sp,N,[M,typ])
typ 1 : fixed (default)
typ 2 : uniform
typ 3 : gaussian
First two arguments are mandatory.

GETLOWRANK.M
Constructs a low rank matrix
L = getLowRank(n,r,typ,U1,Sig1h,U2,Sig2h)
typ 1: 1st ord Gauss
typ 2: 2nd ord Gauss
typ 3: Wishart
typ 4: 1st ord Vandermonde
typ 5: 2nd ord Vandermonde
All arguments are mandatory.

GENG.M GENW.M GENZ.M
Produce row and columns spaces for Gaussian, Wishart and Vandermonde matrices.

CODE01.M -- CODE10.M
Data generation codes used for experiments in "Empirical Recovery Regions of RPCA" paper.
