# Signomial optimization

Hi
My model is in GP form, but I have a negative parameter in the objective function.
I wonder if I can solve it with CVX.
Does CVX solve signomial models?

Thanks

A signomial with negative coefficient is not a posynomiial, is not convex, and will not be accepted by CVX.

However, you might be able to call CVX in an iterative fashion to attempt to find a local solution to a signomial program using methods described in section 9.,1 of â€śA tutorial on geometric programmingâ€ť by Stephen Boyd, Seung-Jean Kim,Â·Lieven Vandenberghe, Arash Hassibi, which is available at https://stanford.edu/~boyd/papers/pdf/gp_tutorial.pdf . I recommend you carefully study that tutorial.

Mark is right; if it has a negative coefficient, it is not in GP form.

Thanks a lot for responding.
I make my model in signomial GP form with the help of this article http://www.sciencedirect.com/science/article/pii/S0377221713008394
Now I use the ggplab software. But I have a question about it. Can I ask it in this forum?

From the abstract
`But some transformation and convexification strategies can be used to convert the original signomial geometric programming problem into a series of standard geometric programming problems that can be solved to reach a global solution.`

Apparently, this is one of several papers over the years which present algorithms which (attempt to) solve the global optimization problem for Signomial GP or other generalized geometric programs by solving a sequence of convex optimization problems.

If these convex optimization problems can be formulated consistent with CVXâ€™s rules, then CVX could be called to solve individual convex problems within the context of a higher level algorithm. I think discussion of how to implement such an algorithm, focusing on the formulation and solution of the convex problems for CVX, could be on-topic for the board.

However, this is not the appropriate forum to ask questions about GGPLAB.