Solving 2D equations using color [video] | Hacker News Reader

Solving 2D equations using color [video]

youtube.com

93 pointsadamnemecek7 comments

If you want to play with the winding number proof of the fundamental theorem of algebra, here's a toy I made a while back to help my linear algebra students (hopefully) gain some intuition for complex numbers: https://math.berkeley.edu/~kmill/toys/roots/roots.html

If you want to play with a wider palette of complex functions as well as domain coloring of the Riemann sphere, there is also https://math.berkeley.edu/~kmill/toys/zgraph/zgraph.html

Documentation for each of them is the "Help" link in their respective upper right corners.

One thing I think would be amusing is to be able to change the texture used for the domain coloring, from the contoured rainbow to say a cat.

3Blue1Brown @11:40: "Being wrong is a regular part of doing math. We had a hypothesis and it led us to this algorithm but we were mistaken somewhere. Being good at math is not about being right the first time. It's about the resilience to carefully look back and understand the mistakes and understand how to fix them."

This is essentially how I feel also about Computer Science: resilience is the best quality that an aspiring mathematician can possess!

3Blue1Brown is one of my favorite youtubers. It is a nice format that is inbetween "explain it to me like I'm 5" and an open courseware. Grant tends to give enough information to become familiar enough with a subject that you can do good research on your own.

Their podcast is also great, Ben Ben Blue. Which has Ben Eater, another great Youtuber.

This is nice and pretty, but it still requires evaluating the function in infinitely many points. Is there an easy fix for that?

1. If you can calculate rough bounds for the terms and their derivatives in the winding number integral, you can figure out how densely you need to sample the curve to calculate the integral exactly.

2. If the winding number is zero, there still might be a root in the region because poles and zeros contribute opposite winding number, and they might exactly cancel.

3. Polynomials are great because the bounds are easy to compute and because you don't have to worry about cancelation since the only pole is at infinity.

You'd think so, but check out the 'smooth surprise' in Kahan's great writeup: https://people.eecs.berkeley.edu/~wkahan/Mindless.pdf

We can approximate quite a bit, but ultimately further analysis is required to show that there are no such surprises on any particular root-finding problem.

Reply on news.ycombinator.com