1
$\begingroup$

I have a system of non-linear equations of the form

$$A x_1^B \exp \bigg(\frac{- C}{x_1} \bigg) = k_1$$ $$A x_2^B \exp \bigg(\frac{- C}{x_2} \bigg) = k_2$$ $$A x_3^B \exp \bigg(\frac{- C}{x_3} \bigg) = k_3$$

where [x1, x2, x3] and [k1, k2, k3] are known. The couple of constants [A, B, C] is the unknown. The solution of this non-linear system of equations is given here: Solving a system of non linear equations

We must now ensure that B < 0 at all times. How would you find one couple [A', B', C'] that best approach the solution of the system, with B' < 0.

Thanks a lot,

$\endgroup$
  • $\begingroup$ I highly suggest you edit this to use TeX/MathJax, or people will probably ignore. $\endgroup$ – The Count Jul 5 '16 at 13:35
0
$\begingroup$

If you take logs, you get linear equations. The first becomes $$\log A +B \log x_1-\frac C{x_1}=\log k_1$$ Your variables are now $\log A, B,$ and $ C$ and you should have a unique solution. Whether $B \lt 0$ depends on the constants.

$\endgroup$
  • $\begingroup$ Yes but when B >0 I want the best approximation [A',B',C'] of the solution [A,B,C] with B' < 0 $\endgroup$ – user1482030 Jul 5 '16 at 15:10
  • $\begingroup$ What do you mean best approximation? Should you leave $A'=A, C'=C, $ and set $B'=-\epsilon?$ That is about the closest point in $A,B,C$ space. You could sum the squares of the errors in the equations (maybe with some weights that suit you) and minimize that subject to $B \lt 0$ $\endgroup$ – Ross Millikan Jul 5 '16 at 15:12

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

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