I did some algebraic transformations on equations with multiple real variables $x_i$, and I'd like to check whether the transformed equation is still valid. The equations are basically only rational functions with rational coefficients.

I found https://en.wikipedia.org/wiki/Lindemann%E2%80%93Weierstrass_theorem and made a guess that using $x_i=\exp{\sqrt{p_i}}$ ($p_i$ is $i$-th prime number) for trial numbers would be a good idea, since plain arithmetic operations are unlikely to introduce spurious equalitities?! So, I plug in the trial numbers and see if the equations still hold (given enough calculator precision).

Is my interpretation correct? Am I guaranteed to detect transformation errors this way, if my equations consist of basic arithmetics $+,-,*,/$ only? Or under what circumstances would the test be reliable?


You cannot always guarantee the correctness by testing finitely many cases or even a countably infinite sequence of cases, as it may be that the error only occurs on some other values. It can be done in some instances. For example, for a broad class of continuous transformations it may be sufficient to check all rational values (based on the fact that if F and G are continuous real functions and F(x)=G(x) for all rational x , then F=G ).It depends on what you mean by a transformation.

  • $\begingroup$ But my function class seems pretty trivial. It's mostly rational functions. And I only do basic high school rearrangements. Do this help to prove anything? $\endgroup$
    – Gerenuk
    Aug 22 '15 at 19:30
  • $\begingroup$ If p,q,r,s are polynomials then p/q=r/s implies pr=qs. If the degree deg(pr)=deg(qs) =n, you can see whether pq and rs agree at n+1 points because if two polynomials of degree n agree at n+1 points they are equal everywhere.. $\endgroup$ Aug 26 '15 at 18:36
  • $\begingroup$ The question is still whether a single test is enough. Can you give an example equation which includes $\exp\sqrt{2}$ and $\exp\sqrt{3}$, has only basic arithmetics and has conflicting truth values after transformations? $\endgroup$
    – Gerenuk
    Aug 27 '15 at 5:23

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.