MaplePrimes - answers and comments on Question, fsolve outputs different solution when using complex parameter.

fsolve outputs different solution when using complex parameter.

gkokovidis — Sat, 24 Nov 2012 05:21:23 Z

In this particular case, the compelx argument to fsolve causes it to look for an answer that is greater than zero, whereas the version without uses a seed value that is different causing it to find -2.753166165 as the root. Both are real and both are correct. You can try this with a given range that you define, with and without the complex option to see that you get the same answer.

fsolve(EQ1,T,complex,300..400);
                        351.311867751985
fsolve(EQ1,T,complex,-3..3);
                       -2.75316616485215
fsolve(EQ1,T,300..400);
                        351.311867751985
fsolve(EQ1,T,-3..3);
                       -2.75316616485215

From the help file:

For a single polynomial equation of one variable with real coefficients, by default the fsolve command computes all real (non-complex) roots. It may not return all roots for exceptionally ill-conditioned polynomials.

Your equation is very sensitive to small variations in x. A plot will show you this.

Regards,

Georgios Kokovidis

Dräger Medical

Not a polynomial

Preben Alsholm — Sat, 24 Nov 2012 05:49:22 Z

In the help page for fsolve and fsolve,details it says:

"For a general equation or system of equations, the fsolve command computes a single real root."

And under the explanation of the optional 'complex':

"complex
Find one root (or all roots, for polynomials) over the complex floating-point numbers."

You may try

fsolve(EQ1,T=300+100*I,complex);
fsolve(EQ1,T=300+100*I);

inspired by the results from

RootFinding:-Analytic((rhs-lhs)(EQ1),T=-5-5*I..400+400*I);

thanks for correction

gkokovidis — Sat, 24 Nov 2012 06:00:25 Z

Thanks for the correction. In my haste, I did not pay attention to the exp() when looking @ the help file.

Regards,

Georgios Kokovidis

Dräger Medical