Something I learned


AppImage

I will keep this week short since I'm still working on the same problem as last week. The blog post about it is still in the making. 

But throughout the struggle, I got to learn about AppImage, and it's pretty neat. I didn't realize that bundling for Linux-based systems was so tricky that someone had to find a way to fix it. It really saved me from having to compile a library on my own. It stopped me mid-way while going through the rabbit hole of Makefiles and their configuration. It was not fun.

Deep learning and material discovery

Through my usual YouTube consumption, I came across this talk by Jeff Dean. In it, he boasts about Google's role in recent machine learning advancements. 

One of the things he mentioned that caught my attention was their recent work in material science and material discovery. A team at DeepMind trained a neural network that can evaluate permutations of new materials similar to ones we already know. Through this, they discovered 2.2 million new materials. Their work adds about 380,000 new stable materials to the 48,000 we already knew.

This is just amazing. I'm not saying I came up with it, but I've had this idea for a while. Ever since I learned about superconductors and Y123 back in high school. It's basically the idea of computationally iterating through the search space of all possible materials and then evaluating which ones are viable. It's the same idea Stephen Wolfram pushed for in his cellular automata explorations. For example, his iterative search over a two-dimensional space with some rules led to many exciting discoveries. Among them is Rule 30, which is a random number generator that was used in Wolfram Alpha and Mathematica for a while.

This line of work will lead us to many discoveries that we never even thought possible. I wouldn't be surprised if this is how we will discover the first room-temperature superconductor. Apply the same approach to biology and organic materials, and this could be how we find the first cure for cancer. I know it sounds as if I'm trivializing the problem when I state it in such a way. But that is not my intention; I know that a lot more complexity is attached to certain problems. It's just this paper, and their approach gives me hope.

Gratefulness


I finally managed to renew my Wisconsin license, which has been expired for years now. Some back story, I got my Wisconsin license while on a temporary work permit. Because of that, it was a probationary one, which expired when my work permit did. So I went through the entire process of getting a license for it to only last me 7 months.

When I went to the San Francisco DMV to renew it, I was told I needed to ask the Wisconsin DMV for proof that I did a behind-the-wheel test or I'd have to do it again. So, just like anyone who lives in the modern age, I mailed the Wisconsin DMV records department a request for such documentation and a written check.

They sent some documents that didn't specify what SF DMV asked for, but I decided to go back and try my luck. This time, I got lucky with the most competent person in the entire DMV. He asked me about the situation, pondered it for a bit, and then got up to go "read the book." He went into some back room and came out with what looked like a 600-page manual of some sort. He flipped through it for a bit and stayed on one page for about 10 minutes. He then came back to me and asked if I had a work permit before, and I told him I did. It turned out I didn't need anything from the Wisconsin DMV; he threw out those documents. The person from earlier assumed I was under 18 when I took my driving test since my license was probationary. It turns out different states have different rules for issuing probationary licenses. Crazy, who would've thought? 

Anyhow, I will now get my license in the mail in about two weeks. I'm grateful and thankful for the competent people I interact with daily. Just not that lady from my first DMV visit.