How would I go about solving a system of nonlinear equations where the highest degree is two?
For example: $$f_1(x) = f_1(x_1, x_2,\dots, x_n) = 0,$$ $$f_2(x) = f_2(x_1, x_2,\dots, x_n) = 0,$$ $$\vdots$$ $$f_n(x) = f_n(x_1, x_2,\dots, x_n) = 0$$

If you could direct me to a pdf or anything it'd be greatly appreciated.

  • $\begingroup$ Look at convex optimization if the $f_i$ are convex (or concave, traverse to $-f_i$ in this case). If they're not, you're gonna have a bad time. $\endgroup$ – AlexR Dec 1 '14 at 9:40
  • $\begingroup$ It doesn't matter if they are convex or concave. Nonlinear equations are non-convex. $\endgroup$ – Michael Grant Dec 1 '14 at 14:06
  • $\begingroup$ @MichaelGrant Thank you, I'm not familiar w/ LateX and couldn't have presented the question as well as you did, thanks dude. $\endgroup$ – Manny P Dec 2 '14 at 10:09

As already said in comments and answers, optimization is the way to go. You can formulate it as the minimization of $$\Phi(x_1, x_2,\dots, x_n) = \sum_{i=1}^nf_i^2(x_1, x_2,\dots, x_n)$$ hoping that you arrive to something close to zero.

In any manner, except if you use global optimization, this will require a "good" starting point. If you have such a point, you could linearize each of the $f_i(x_1, x_2,\dots, x_n)$ and solve the problem as a linear problem; each iteration would hopefully improve the guess. This is Newton-Raphson method.

You could be interested by http://fourier.eng.hmc.edu/e161/lectures/ica/node13.html, http://tx.technion.ac.il/~dlewin/054374/Day_5.pdf

  • $\begingroup$ Leibovic, Thank you, much appreciated. $\endgroup$ – Manny P Dec 2 '14 at 10:10
  • $\begingroup$ You are very welcome ! If fact, this applies to any system of nonlinear equations. The fact that the highest degree is two does not make anything simpler. If you need more, just post. $\endgroup$ – Claude Leibovici Dec 2 '14 at 10:15

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.