[R] Matrix-free fused lasso

[Antony Lee]

Hi,
I am looking for an implementation of the fused lasso that allows the
predictor matrix to be an "abstract" linear operator, namely the cumulative
sum (that is, (X.b)_i = sum(b_k, k=1..i)) (due to the size of the problem,
forming the entire matrix is unlikely to be a good approach).  Any pointers
would be welcome.
Thanks!
Antony

	[[alternative HTML version deleted]]