Implementing Scientific Paper in C

[@sooth6487]

implementing a scientific paper in C, but also an in depth strawberry pasta discussion at 20:10. Where else would I go to find such diverse content

[@AHuMycK]

Hi Tsoding, not sure if you read the paper or not but your sum operation is not correctly defined. According to the paper """ All functions appearing in the RandomArt algorithm are scaled so that they map the interval [-1; 1] to the interval [-1; 1]. This condition ensures that all randomly generated expression trees are valid. For example, the scaling for the add function is achieved by defining add(x; y)=(x + y)/2 """ Because your add(x;y) is defined as x+y, it means that your sum node generates invalid expression tree and that's why you were able to generate all those beautifully looking non-linear patterns. I'm not sure it's possible to generate images that you did (4:26:19, 4:26:47 and so on) by using any combination of x,y, add and mult (linear) operands

[@weekipi5813]

it's not that he generated an invalid expression tree. The expression tree generated was a valid one, rather the evaluation as you're correctly pointing out was not aligned with the paper.

[@froop2393]

Creating a terminating rule is easy if you do it in reverse. Start with some terminals and then join them with add, mult etc. until you have three left and return them as triple. It's just like building a mahjong stack is super easy, but solving is hard and may fail. I really enjoyed the full 4h. This was one of your best streams ever!

[@FaZekiller-qe3uf]

tsodinPenger

[@krux02]

You missed the opportunity to use Lisp in the domain, where it is at it's best: Generating code at runtime and executing it.

[@salim444]

lisp supremacy

[@adrien335]

Could you provide a code snippet/gist to showcase where you would use macros? Just curious.

[@krux02]

@@adrien335 Yes I could.

[@ismbks]

​@@adrien335asking for a code in yt comments that's a bit wild 😂

[@paper_cut9457]

in the end, programming is all about printf-ing a triangle of stars and feeling good about it: but the level here is quite higher ! thanks

[@theevilcottonball]

Errr, no. Tsogramming is much more than that. Printf to a terminal, are you from the 1970s?

[@itsmenewbie03]

functions mentioned

[@Italya3343]

Thank you so much for sharing your AMAZING video!! 🔥🔥🔥💐 Plz, continue making more and more paper 📃 to C videos 🙏

[@lifelover69]

this is the kind of generative art i can get behind, very cool

[@asdfmonstrosity]

Don't know if you resolve it later, but the paper says to always pick branches->items[0] when depth

[@nel_tu_]

This. Also, there is no min. depth for a grammar; you just follow the first branch assuming it terminates. He also missed the part in the paper where there is only 50% chance to go deeper, bro just needs to read. He should have titled the video "Implementing a scientific paper but changing everything because I don't feel like reading it" smh

[@kaviisuri9997]

Is there a video for when you implemented the bnfuzzer?

[@geertdepuydt2683]

Best way to learn C when coming from another language 100%. Extremely educational.

[@ecosta]

Implementing a scientific paper is always a bet. Sometimes it just works. Sometimes you find so many holes in the solution that I feel like I'm rewriting the article from scratch.

[@coath]

Epic programming moment

[@darktale6716]

wow Tsoding knows about the valencian flooding! it was pretty rough

[@edbrito-swdev]

You have the coolest KZbin developer channel. Do you work in the IT industry or do you only do KZbin and these projects (plus patreon and stuff) as a means to make money?

[@edbrito-swdev]

Ok, I just saw you saying "Why pasta with strawberry isn't fine" and now I've lost all my respect for you. Just kidding (but I was shook).

[@blackt0wer]

He writes code for banks in the .ru

[@H3llsHero]

He doesn't make money from Twitch/KZbin/Patreon because he is in Russia. He used to work as a software developer for some American company, presumably he saved enough and lives frugally enough to not have a day job anymore. But who knows?

[@thebirdhasbeencharged]

he's a cia bot

[@zaitsev5990]

@@thebirdhasbeencharged KGB*

[@Hamsters_Rage]

you dont need all this stuff with bools and ifs. if you make all funcs binary, like if_gt, if_less etc - you can store it in just an array-packed binary tree and execute all genetic ops on this array

[@John-yg1cq]

In reference to the terminal nodes: I have to ask, is a terminal node not simply a node of type A? As you're calling E, you get a triplet of C C C which then can be a op which then can be self-referential to As, and As can only be random number within range or x or y. As you traverse the grammar, any A node is a terminal node, as described in figure 3 (a)? Am I misunderstanding?

[@markahlenganpanjang]

The resulting image is insane

[@samuraijosh1595]

what tool do you use for dark mode on pdfs?

[@TsodingDaily]

mupdf

[@anon_y_mousse]

You can also use okular and change its settings to invert colors. Seems to have the most features yet still runs fast, even runs on my Pi.

[@theevilcottonball]

Maybe I will implement genetic art as referenced as an inspiration in the paper

[@xspager]

Not a lisper myself but this paper sounded like the ideal case for Lisp

[@thistent]

You know, you can apply an invert filter to the CSS in your browser's PDF reader.

[@Lucrecious]

that is so fucking cool man, nice man

[@IvanNovitskii]

Omg just provide happy path by marking grammar rules that eventually lead to termination with ones, and when you exceed depth take that path. It will no be exactly the right depth but within 2 lvls. You can even have multiple options to exit and scale probabilities. The struggle was painful to watch. But I guess you should have figured it out already. But cool result though nevertheless On a side note, you can precompute how much it takes to exit (in min steps) via each rule in production to make the border strict and do not take path that cannot terminate in time. Its just a graph problem in the end.

[@tens0r884]

I needa find a girl who's as loyal to me as zozin is to C

[@multicoloredwiz]

Damn.. got a little crazy once he got to the nn/grammar stuff.. uff

[@monad_tcp]

BNF is almost basically a CFG for an automaton.

[@edbrito-swdev]

A BNF is essentially a DFA (from what I remember from my language processing classes and personal experience. Might be some cases where that breaks, though).

[@MrFram]

⁠​⁠@@edbrito-swdev BNF is a notation for CFG, and DFA is below CFGs on the Chomsky Hierarchy, which makes it a subset of CFG, though BNF is usually not used for it. DFA is equivalent to NFA (up to exponential difference in size), and regular expressions are the common notation for NFAs.

[@edbrito-swdev]

@@MrFram Yes, you are obviously completely right. I should have thought a little more about it before writing. Of course Backus Naur has to be above because it is used for grammars and DFAs are used for things like regexes. This should have been obvious to me 🤦

[@mohammedkaka5558]

I'm doing this bro, I was looking for a C refresher project then this came up. Any suggestions?

[@Siroitin]

Suggestions for what?

[@karoI508]

suggestion is: stop asking, do it

[@SlinkyD]

I use your binviz for that. Will you take a PR thru a comment since I can't find my github password or recovery codes? I'll make it a couple of sed commands since its small.

[@greyshopleskin2315]

You can generate a patch with either git or diff. Learn about it

[@SlinkyD]

@@greyshopleskin2315 Who woulda thunk‽

[@SlinkyD]

no ​@@greyshopleskin2315

[@mattshu]

Do you have any vim tutorials? I wish I was as fast as you!

[@eeriemyxi]

He uses Emacs. You can use any code editor for multiple years and start typing at insane speeds. It's about practice.

[@ratofthecity6351]

i think theprimeagen has some decent vim tutorials that are worth watching as a beginner. id recommend using vim (neovim) almost exclusively while you learn it, and to try to incorporate it into your workflow wherever you can. the best way to learn it is by immersing yourself in it. also you should rebind your capslock key to either ctrl or escape - i prefer ctrl but ive heard many vim users like escape

[@JamesSjaalman]

You could have emitted RPN, which is much easier to process.

[@Hodebrot]

30:39 recreational beatboxer / programmer

[@whatthehelliswrongwithyou]

whats up с карнеги мелона

[@Seedwreck]

Implement Chaotic map for PRNG next

[@LarsDonner]

42:26 Just pretend that your image format uses a 0.8 fixed-point numbers for the color channels. Then even PNG uses the [0..1) range!

[@SphereofTime]

2:33 hash visualization

[@explqicot3293]

Can you do more scientific papers please 🙏

[@re_detach]

we gettin 2 the bottom of this 2 day

[@theevilcottonball]

Use Rust at 2:20:01. Me litterally coding along writing the same thing in Rust. (because using the same language aas Mr. Tsoding is too easy) Actually Rust is quite nice with pattern matching on the Node enum and yes I clone all over the place to have a fun time in Rust.

[@theevilcottonball]

C; single file dependency Cargo: Lets download profiling, deques, bigint, unicode, weezl, adler32, adler2-simd,... they even use a crate "rgb" for a single struct of rgb pixels, an entire crate!

[@theevilcottonball]

Sad, that -Wint-conversion does not catch it.

[@hubstrangers3450]

Thank you....

[@devdeev3675]

this in GLSL!!!

[@theevilcottonball]

Tsoding: What maps two spacial coordinates to a 3D color. Me: A shader? (I know its a usually 4D color) Tsoding: Neural Network.

[@devdeev3675]

@@theevilcottonball Lol, he did it with shaders

[@the_yugandharr]

Your streams are free??

[@krux02]

Streams are free, but you can't comment for free there. That only works here in the non life version.

[@the_yugandharr]

@krux02 free to watch. that's already enough for me :D

[@bzboii]

bran cheese

[@darktale6716]

Hello @laur0276, I am a subtitle watcher :)

[@niculaelaurentiu1201]

Omg hello xD

[@dancinglazer1628]

Oh

[@John-yg1cq]

21:00 I want to disagree that bread == pasta so much. Thanks, I hate it.

[@shrddr]

no f way you get stack overflow on first try when there's only 2/3 chance to add a depth level

[@GesteromTV]

They litarly sad that going to moon was easier then faking it. Assuming generating random gramar and parsing it was a moon landing.

[@imya_mujik_familiya_tajik]

> 420 meditation lane lmao

[@emoosavimehr]

30:58

[@xspager]

Based

[@azharalibhutto1209]

Great ❤❤❤❤❤❤❤

[@xbelanch]

from 16:9 to 4:5 weird

[@NicolaCodes]

✅✅✅✅✅✅✅✅

[@m4ya_0]

amogus drip

[@naranyala_dev]

try sokol

[@tilakmadichettitheappdeveloper]

neovim > emacs > my boring depressing life j.k

[@frogg03_]

Big neural networks with random weights basically just give you average values. I think that if you tried small neural networks, one for each r, g and b values, it would be much more interesting!

[@rogo7330]

For an hour he tried to go around a depth by still doing the same thing as the depth does not exist. Instead of doing what chat tried to tell him "just return NK_RAND and make it terminal". You should go to water instead of tea because it's clearly makes you so freaking nervous and unfocussed, man. (joking, but still, wtf is happening with you? you haven't being like that year ago)

[@theevilcottonball]

I don't care that chat is sub only, now. Twitch chat is useless most of the time anyway. In sub only mode its less useless but still useless.

[@KybaLioN66]

zozin ))))))

[@Nvimchad]

sir why don't you make video series about the topic you cover, I mean It's Hard to see specific topic don't you think it's unorganised, I also looked at playlists but I didn't find that I was looking for... can you please make new series on golang...?

[@twl148]

cool but disliking for calling ipv6 a useless protocol