His version at least returns complex answers

Yeah, complex numbers is something I should know more about. Sadly, I don't (yet). Not being able to solve x²+x=0 is just weird behavior by a quadratic solver. For anyone still caring, here is my current code:


Code:


I'd love suggestions on making the program smaller.

A real number and a complex number are two different entities on the calculator, and if it finds that it's about to do a real→complex conversion during its evaluation, it checks the mode first to see whether you're okay with that. "Real" says you're not. If, after simplification of its argument, the square root finds that it must operate on a complex number rather than a real, then no conversion or check is required.

Using that to our advantage, we can add a +0i term to the inside of the square root to force the computation regardless of the mode. In our case, we can even turn √(B²-4AC) into √(B²+4ACi²), which is two bytes more. But, that's fewer than having a+bi at the top of the program, which is one byte for the newline and two bytes for the mode. Plus, you don't have to mess with the user's settings.

And speaking of your routines, Weregoose, are you planning to put them back online someday? Your unitedti.org page was amazing.

Definitely agreed! I would be more than happy to host them here in some capacity, of course.

I believe the only time you will get an imaginary number is when the discriminate ends up being negative, otherwise you shouldn't have to worry about it. just set you calc to a+bi before using the program

EDIT: nevermind, looks like you guys got this one figured out

Yes, the discriminant is indeed how you know if you're going to end up with imaginary roots, since the radical of the discriminant is the only place that an imaginary number could come from in the quadratic formula. A nice thorough quadratic solver will check for imaginary roots and a double root, report either case, and then display one or two roots, as necessary.