Mathematics Stack Exchange is a question and answer site for people studying math at any level and professionals in related fields. Join them; it only takes a minute:

Sign up
Here's how it works:
  1. Anybody can ask a question
  2. Anybody can answer
  3. The best answers are voted up and rise to the top

Here is a convex programming problem I encountered while working on an estimation problem for a mixture of multinomial distributions.

We have a matrix $A_{m \times n}$ containing non-negative real numbers.

We seek to maximize

$$ \sum_{i=1}^m \log (\sum_{j=1}^n a_{ij} \theta_j) $$

such that

$$ \sum_{j=1}^n \theta_j = 1, \theta_j\geq 0 $$

I'm not sure if this has been studied extensively before, and I'd like to have some analytic results about this convex programming problem.

share|cite|improve this question
This is probably not news to you, but it helped me visualize the problem: We are given $m$ different affine functions on the $(n-1)$-simplex, and we want to maximize (the log of) their product. – Rahul Jun 15 '12 at 3:45
Isn't this simply the logarithmic barrier? Check section 8.5.3 of Boyd and Vandenberghe. – TenaliRaman Jun 15 '12 at 3:47
Not a logarithmic barrier because hyperplanes $A_i$ are not active because of simplex constraint. As Rahul said, logarithms could be excessive here, because the problem is the same as maximizing polynomial (product of hyperplanes) on simplex. Poster should check polynomial optimization literature - it's vast – mirror2image Jul 11 '12 at 5:59

Your Answer


By posting your answer, you agree to the privacy policy and terms of service.

Browse other questions tagged or ask your own question.