Image not available

1024x685

julia-vscode.png

๐Ÿงต Untitled Thread

Anonymous No. 16260986

Any thoughts on this for scientific computing / number crunching? Is it better than Matlab?

Anonymous No. 16261173

>>16260986
Very comfy language. Certainly superior to matlab. More beautiful tjan Python too.

Anonymous No. 16261227

>>16260986
very fast, very comfy, can even be GPU accelerated

Anonymous No. 16261233

>>16260986
It's just Python wannabe inferior cousin due to limited software packages and stupid lack of features in the language itself. just use python, or c++ if you want performance.

Anonymous No. 16261243

>>16260986
Some guy implemented a simple model for synchronized oscillators in Python, which I used quite a lot for my thesis. Unfortunately I had to wait quite a while for each run, but it didn't matter too much.
After my thesis I decided to have a crack at improving it, but using Julia as my peers were jerking it off.
Made something that produces the same results, but was 2 orders of magnitude faster, dropping the processing time from about 2 hours to a little over a minute.

Anonymous No. 16261248

>>16261243
you can achieve the same result in python. by default there is no optimization but you can use libraries for such purposes. there's no need to use a limited language like Julia just for the acceleration. sounds like skill issues my guy.

Anonymous No. 16261295

>>16261233
>>16261248
What is missing from Julia?

Anonymous No. 16261618

>>16261295
Nothing really, maybe some niche libraries because it's a relatively new language.
I'm too much of a retard to learn a new syntax so I cannot ever switch to any other lang other than python. even then I'm finding it hard to start using the retarded type hinting shit they introduced in python.

Anonymous No. 16261625

>>16261618
btw I've tested Julia with kernel on jupyter a while ago. I got back to python because of my long habit with python.

Anonymous No. 16261641

>>16260986
Everything is better than Matlab. Octave is better than Matlab. Matlab needs to be voided.

Anonymous No. 16261646

>>16261641
this, papers should be voided if they publish their code in matlab.

Anonymous No. 16261650

>>16261295
A compiler.
You need to install a JRE (Julia Runtime Env) to run Julia. No compiled programs. The programs dont exit the academic space.....
It's good for
write once
Compile once
run once

If you need to run multiple times, it is suboptimal.
Still better than python

Anonymous No. 16261683

>>16260986
I haven't personally used it myself, but it does suffer from it just being a young language from what I've heard.

Depending on exactly what you need for number crunching, Julia does provide nice functionality in the standard library. I recommend looking at the documentation for what sort of things you might need: https://docs.julialang.org/en/v1/

Furthermore, Julia also supports calling code from C: https://docs.julialang.org/en/v1/manual/calling-c-and-fortran-code/

If you're already acquainted with Python though, I'd rather stick to an LLVM like numba for the convenience of familiarity and better support

Anonymous No. 16261712

>>16260986
I've mostly been avoiding it because there are a ton of subtle bugs in the core language and major packages that make it hard to trust any nontrivial program written in it
https://yuri.is/not-julia/

Anonymous No. 16261735

>>16261712
>open source project has bugs
Many such cases

Image not available

945x563

IMG_7616.png

Anonymous No. 16261755

which framework produces the nicest drawings/animations?

Anonymous No. 16262557

>>16261248
The original implementation used a bunch of differential equation libraries. The saving grace for a speedup in Julia was due to static allocation.
Could that be done in Python? Genuinely curious, I'm no Python wiz.

Anonymous No. 16262897

>>16261755
Perl Data Language