Agathe is a user on mathstodon.xyz. You can follow them or interact with them if you have an account anywhere in the fediverse. If you don't, you can sign up here.

Agathe @agathe@mathstodon.xyz

Légèrement mauvais esprit Show more

Agathe boosted

You asked for it: github.com/halcy/Mastodon-LaTe Here's a mastodon LaTeX client! Screenshots of example output attached!

Unrelated: I could use a "previsualize before tooting" function on Mathstodon, because it's not always easy to spot LaTeX mistakes before posting, and then re-typing the LaTeX code of your posts is a pain.

What mathematical indicator could I use to measure the temporal regularity of an event?
Like, I've got an recurring event that I measured happening at times \((t_n)_n\), and I suspect that \(\forall n, t_{n+1} - t_n = T+\epsilon_n\), where \(epsilon_n\) is "small": \(t_n\) is almost periodic, but with some error.

How could I experimentally quantify the fact that \((t_n)_n\) has this property?

Worn-out joke Show more

Agathe boosted

I love Wikipedia's math gifs!
Here, an illustration of Bresenham's algorithm (en.wikipedia.org/wiki/Bresenha)

So, I'd like another way to compute integrals over segments, and I'd like to do it in a way that is coherent with the rest of my computations, and right now I'm kind of stuck with no ideas.

Has anyone ever done similar stuff, or has ideas as to how I could do this? Thanks!

For the 2-dimensional integral, it's fairly easy: you just have to sum the values of the pixels that make up the polygon you have to integrate over, and then divide by some normalization term (here, it's the total integral over the domain).

However, if you apply this method to, let's say, segments, you end up with an almost-zero value, since what you're doing is computing their 2D Lebesgue measure, which is 0.


Also, in another life I studied \(\lambda\)-calculus and all things rewriting, but I don't think I will be dealing with them academically anymore :/

Sooo, now that I have introduced myself, I can start putting my math problems all over the federated timeline 😀

I've got this program where I need to integrate over a continuous domain that is represented by a bitmap image (so, basically, a m×n grid).
I'd like to compute both 2-dimensional and 1-dimensional integrals, in such a way that both \(\int_{x\in[0;1]}dx\) and \(\int_{x\in[0;1]\times[0;1]}\) have value 1. (so, respectively 1-dim and 2-dim Lebesgue measure)


Hello everyone!
I'm Agathe, a french CS/applied mathematics pre-doctoral student. I'm currently doing an internship on semi-discrete optimal transport, and I hope to continue studying this topic during a PhD!