#coding runnerpost

[somerunner]

Since I lost my laptop last summer, I also lost access to two short coding projects (like, 2 hours tops) that I'd done. One was following along with a 2-d planetary orbit simulator, to learn a bit about OpenGL. The other one was to take a text file and exchange each character for its equivalent on an opposite keyboard mapping (QWERTY <--> DVORAK).

I'll probably want to remake both again; I still even now don't understand Lagrange points (the whole reason I did the tutorial was to get that simulator up and running for myself), and I'd like some basic cryptography on my personal journal entries that I can easily unscramble.

Anyway, I also picked up this new project: trying to get this program (a Verlet Integration particle simulator), whose code is public, to run on my Windows laptop. It was written for MacOS and I've run into issues my first time using Makefiles to try to get it working.

[somerunner]

I’ve always been amazed by 3Blue1Brown’s video quality. And then I watched this video:

Turns out he wrote an entire Python library for it all. I just felt like this watching that whole video:

The amount of work displayed here is probably not truly ridiculous, but it’s definitely beyond what I understand. To write a library capable of all that is incredible.

[somerunner]

Does anyone on the coding, physics, math, or education subsets of Tumblr know how to use Manim?

I might try to learn it this summer. If no one else has started posting Manim liveblogs and/or snippets on here…I’ll take that duty on myself

(I figure at least a few people behind the Manim community are on Tumblr, but searching “Manim” on here only shows a few scattered posts)

[somerunner]

I almost want to cry. In another class we used "plt.pause(0.01)" and "plt.show()" to crudely animate a wave on a string using loops. It created a new figure every time the loop went for another go-around, and the pausing was crucial to actually allow the figures to be drawn.

In this class, I was updating a single figure, 25 subplots together, and I stuck the "plt.pause()" and "plt.show()" instructions in there. And it only displayed a single subplot. I hardcoded another subplot to be filled, and it worked, but no matter what I did it would then only be those two. I didn't have the patience to hardcode the entire 25; I had to figure out why my plots weren't working.

Professor posted the solution. No "pause()" or "show()" anywhere. Clearly, those interrupt or prevent any further updates to an active figure, at least in some sense, and I can't figure out why from the documentation. Does plotting run on a separate thread from the loop? Does the loop keep iterating while plt is paused? Does show prevent all further updates?

Looking more deeply at the documentation does not help. From now on, I'm starting a day early.

[somerunner]

Very surprised by this and how impressive it was. As with a lot of coding/engineering projects on youtube, it looks nearly impossible to get the barest notion of understanding on...and then after a few minutes you think "oh I see that there are some basic principles in this that I do understand."

Doesn't mean I'd have the understanding to build or debug any of them, but like, I can see at least one of the principles it operates under. In this case, that there are reflective spots burned into the crystal at varying depths, so instead of 1080p on a flat screen, it's 1080p scattered through like 50 layers of depth (EDIT: so it has very very bad resolution! But it has depth unlike a normal flat screen). Two "adjacent" pixels on a flat screen (e.g. the projector) would be far apart in the depth sense (in the axis pointing to/from the projector).

[somerunner]

Why does Youtube keep recommending these insanely impressive projects. First it's the Manim library, then it's some marbles thing, then it's this. I've seen plenty of cool "Lego Technic" engineering on the Brick Experiments Channel and Brick Technology, but this one...man. Synchronized robot movement without collisions is wild to me, though it's probably no harder than some of the other things I've seen.

I'd be hard-pressed to even understand a quarter of the code that went into this.

[somerunner]

Looking at some incomprehensible code from a physics simulation Youtube channel. Now, the source code is actually very neatly organized and the variable names make sense, it's just that there are a lot of them. So it's very comprehensible, I'm just barely code literate.

It's only about 1000 lines in total, which is actually a lot, but if I ever tried writing something that did half of what this did, it'd be like 5000 lines.

This is the most popular video on the channel. It's not the one I was looking at the source code for, I think? Unclear to me because I don't even know what kind of sim I'm looking at.

[somerunner]

Why is processing a sorted array faster than processing an unsorted array?

In this C++ code, sorting the data (before the timed region) makes the primary loop ~6x faster: #include &lt;algorithm&gt; #include &lt;ctim

This whole thing was a fascinating read. I've read it before, but I looked at it again today.

Long story short, "branch prediction" is where, if a program needs to have an answer to a logic question before proceeding, it just proceeds anyway, hoping it got the right answer. It checks, and if it was wrong, it backtracks and makes the right decision. But if it was right, then it saved a bunch of time.

And if an array is sorted, then it's going to guess right a ton of the time.