1
$\begingroup$

I am trying to find numerically $\arg\min_{x\in(1, +\infty),y\in(0, 1)}\sum_i\log(xy+\alpha_ix+\beta_iy+\gamma_i)$, where the sum has a large number of terms, and the coefficients are such that the expression is always defined ($\alpha_i, \gamma_i$>0, $\beta_i>-1$).

Of course, one can rephrase this as trying to solve the following system: $\sum_i\frac{y+\alpha_i}{xy+\alpha_ix+\beta_iy+\gamma_i}=\sum_i\frac{x+\beta_iy}{xy+\alpha_ix+\beta_iy+\gamma_i}=0$.

Is there any theory behind this kind of equations? (an actual way to solve this is welcome but I'd like to know about theory as well.)

Thanks in advance.

Edit: more precise wording.

Improve this question
$\endgroup$
3
  • 1
    $\begingroup$ What is the domain of definition? What are the signs of the coefficients? DO you really mean arg min, as opposed to argmax (notice that at the origin the sum equals $-\infty,$ so that's a pretty good candidate for arg min...) $\endgroup$
    – Igor Rivin
    Mar 6, 2012 at 1:21
  • 1
    $\begingroup$ Also posted to m.se, math.stackexchange.com/questions/116851/… $\endgroup$
    – Gerry Myerson
    Mar 6, 2012 at 4:19
  • $\begingroup$ Sorry, I did not intend to be rude. Is this explicitly discouraged somewhere? $\endgroup$
    – antony
    Mar 6, 2012 at 8:26

0

Reset to default

Your Answer

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