Page 1 sur 1

possible regression in solve

Publié : jeu. déc. 14, 2017 9:56 pm
par lukamar
Hi,

When solving z^2*conj(z)=z for complex z (set in configuration) using solve, one obtains:

Code : Tout sélectionner

list[i*` y`+sqrt(-` y`^2+1),i*` y`-(sqrt(-` y`^2+1))]
This is only part of solution, as an obvious solution z=0 is missing.
I believe this had worked in version 1.2.3, where solve returned -1, 0, 1 alongside with the above solutions (but I'm not 100% sure). There might be a regression in solve...

Re: possible regression in solve

Publié : ven. déc. 15, 2017 10:53 am
par frederic han
indeed, I still have a working 1.1.0:
Capture du 2017-12-15 11-46-11.png
Capture du 2017-12-15 11-46-11.png (31.65 Kio) Consulté 2842 fois
while the current one:
Capture du 2017-12-15 11-48-08.png
Capture du 2017-12-15 11-48-08.png (47.27 Kio) Consulté 2842 fois

Re: possible regression in solve

Publié : ven. déc. 15, 2017 5:42 pm
par parisse
Indeed, I have improved solving for 2x2 polynomial systems with the resultant, but I forgot the case where the resultant is 0. Will be fixed, but it's unfortunately too late for 1.4.9-41...