Consider a Geometric Program (GP), $$ \begin{array}{cl} \operatorname{minimize} & f_{0}(x) \\ \text { subject to } & f_{i}(x) \leq 1, \quad i=1, \ldots, m, \\ & g_{i}(x)=1, \quad i=1, \ldots, p, \end{array} $$ where $f_i$ are posynomial functions, $g_i$ are monomials, and $x$ is the optimization variable.

I have problems including the simple equality constraint $Ax - b = 0$, for some $A$ and $b$, into the GP formulation. For example, when I formulate the problem in CVX the problem is not DGP-compliant since this equality violates the disciplined GP rules. This is because standard GPs only allow monomial equality constraints in its formulation, and $Ax - b$ can be interpreted as a posynomial.

Is there any workaround to this? I tried to relax the constraint as $Ax \leq b$ (since polynomials are allowed in inequality constraints) but strangely CVX still raise a DGP error.

  • 2
    $\begingroup$ Perhaps a trivial question, but are you sure that the coefficients in A (in your Ax<=b constraint) are positive? $\endgroup$ Commented Jun 24, 2022 at 14:51
  • $\begingroup$ Yes, they actually are positive $\endgroup$
    – Apprentice
    Commented Oct 20, 2022 at 5:16

1 Answer 1


You might be able to use the suggestions in


to convert your problem to a conic optimization problem. It might also make it clear to you the cause of nonconvexity.


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.