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

I thought of writing this question Minimum for this function in a different way, if it helps.

I want to minimize $$\sum_{i=1}^n a_ix_i + \nu \sum_{i=1}^n b_i 2^{x_i} ,$$
where $a_i \in [0,1]$, $b_i \in (0,\infty)$, and $x_i \in [x_{\min},0)$, and
$$ \sum_{i=1}^n 2^{x_i} = 1 .$$

$x_{\min}$, $\nu$, $a_i$ and $b_i$ are constants.

I guess the tricky part is to minimize the function while ensuring all $2^{x_i}$ sum up to $1$ for these $n$ variables.

Thank you for your patience and help.

share|cite|improve this question
up vote 2 down vote accepted

First let $x = (x_1, \dots, x_n)$ and write $f(x) = \sum_i^n a_i x_i + \nu\sum_i^n b_i 2^{x_i}$ and $g(x) = \sum_i^n 2^{x_i} - 1$. Now the method of Lagrange's multipliers tells you that you have a solution $x'$ iff it is an extremal point of the function $f(x) + \lambda g(x)$ with $\lambda$ a parameter to be determined. So you get equations $a_i + (\nu b_i + \lambda) 2^{x'_i} \log{2} = 0$ for all $1 \leq i \leq n$. It's now easy to express $x'_i$ as a function of lambda and plugging this into the remaining constraint $g(x') = 0$ to find the lambda and complete the solution. Hope this helps.

share|cite|improve this answer
I agree but is this (…) not going to happen here. I will get a polynomial equation for $\lambda$, which will be not solvable if degree is greater than 5. – SkypeMeSM Nov 7 '10 at 16:54
It will be solvable numerically. – Yuval Filmus Nov 7 '10 at 16:54
It will not be solvable in radicals (in general). Is this a problem? Another thing is that x' need not be a minimum but only a stationary point, so you will need to further inspect the stationary points. The minimum can also be attained on the boundary of the domain D(f), g=0 which means that some x_k might be equal to x_{min} but that only reduces the problem to having less variables. – Marek Nov 7 '10 at 17:26
Ok. I understand what you are saying. I guess I got my answer here. :) Thank you all. – SkypeMeSM Nov 7 '10 at 17:29

I don't think rewriting this way helps. In the first place, how would you show that if $y$ is the optimal solution of $\min_y f(y)$, then $x = 2^y$ is the optimal solution to $\min_x f(\log x)$?

share|cite|improve this answer
This is not a good answer. – The Chaz 2.0 Dec 13 '11 at 5:23

Your Answer


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

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