lasso regression non-positive constraints

Hello, I read about the l1_ls package http://stanford.edu/~boyd/l1_ls/ to solve large sparse problems. It has another version to solve lasso with non-negative constraints. I went through the code of both l1_ls and l1_ls_nonneg. But I am not sure what changes to make in the code to implement lasso with non-positive constraints. Can someone please help me?

Thanks so much!

[Sandipan]

Could you solve the problem??? I have a solution using Bayesian method...

Answer 1

[John D'Errico]

Suppose you were to just negate your dependent variable?

Solve now, using non-negative solver. You already know how to use that.

Finally, negate the coefficients.