/sci/ - Science & Math » Thread #14293941

36KiB, 648x361, 1_DcL9vLM9m-OPxXobLG-0Fg.jpg

View Same Google iqdb SauceNAO

Anonymous Thu 10 Mar 21:11:15 2022 No.14293941 View Reply Original Report

Quoted By: >>14294014 >>14294026 >>14294144 >>14295039 >>14295056 >>14295218 >>14295227 >>14295311 >>14295318 >>14295852 >>14295865 >>14296982 >>14300153 >>14300502

Julia is the language of /sci/entists. It will soon fully replace MATLAB (already does partially) and be a viable alternative to all Python usecases.

Anonymous

Anonymous Thu 10 Mar 2022 21:15:43 No.14293952 Report

Quoted By:

>It will soon fully replace MATLAB
Good.
Fuck MATLAB.
I'll definitely start learning this. I've noticed it's been receiving a lot more attention.

Anonymous

Anonymous Thu 10 Mar 2022 21:21:30 No.14293962 Report

Quoted By: >>14293979 >>14293986

Based always wanted to learn this OP. Any resources?
>MidwitLAB

Anonymous

Anonymous Thu 10 Mar 2022 21:28:23 No.14293979 Report

Quoted By: >>14293982

>>14293962
Not OP but I've never learned programming by reading books. Julia's syntax is easy if you know MATLAB so just start using it and learn as you go.

Anonymous

Anonymous Thu 10 Mar 2022 21:30:08 No.14293982 Report

Quoted By:

>>14293979
Yeah same but I find videos and that sort of thing helpful to get used to the syntax and bullshittery of each language, especially set-up for stuff like C.

Anonymous

Anonymous Thu 10 Mar 2022 21:32:57 No.14293986 Report

Quoted By: >>14294009

>>14293962
>find the official docs page
>decide on some project you're gonna implement
>do it while googling shit you don't know, figure it out as you get a feel for it
you can start pushing out working code in any language in hours to days if you're not a literal retard, anon. so long as you have a good understanding of the basics of programming and know what to expect of a language and its toolset.

Anonymous

Anonymous Thu 10 Mar 2022 21:45:36 No.14294009 Report

Quoted By: >>14294152

>>14293986
Is It not a bitch to set up like python and C with packages and compilers?

Anonymous

Anonymous Thu 10 Mar 2022 21:47:42 No.14294014 Report

Quoted By: >>14294018

>>14293941
why use Julia instead of Python? If I use Python there's millions of troons making opensource software for me to use for every possible application. How many do that for Julia?

Anonymous

Anonymous Thu 10 Mar 2022 21:51:06 No.14294018 Report

Quoted By: >>14294040

>>14294014
The package ecosystem certainly isn't as large or mature as Python, but it's growing. If you need a Python library, you can use PyCall to call it from your Julia program.

Anonymous

Anonymous Thu 10 Mar 2022 21:54:12 No.14294023 Report

Quoted By: >>14294032 >>14300553

fortran is the language of scientists. you zoomers will never win.

Anonymous

Anonymous Thu 10 Mar 2022 21:56:20 No.14294026 Report

Quoted By: >>14295079

>>14293941
I give it another 10 years to see widespread Julia adoption in the scientific community. I'm a data scientist and most of my colleagues haven't even heard of it. Python isn't going away anytime soon and many people would rather stick with that unless forced to change. Even though Julia is fucking awesome for numerical computing, it's still in an early adopter phase.

Anonymous

Anonymous Thu 10 Mar 2022 21:58:28 No.14294032 Report

Quoted By: >>14295638 >>14296939

>>14294023
>muh fortran you whippersnappers
found the grandboomer

Anonymous

Anonymous Thu 10 Mar 2022 22:01:34 No.14294040 Report

Quoted By:

>>14294018
>>What's the point of using Julia over Python?
>Well with Julia you can call Python!

Anonymous

Anonymous Thu 10 Mar 2022 22:28:23 No.14294104 Report

Quoted By:

It great but it is still very far from maturity. For example the debbugers it has are all buggy piece of garbage. They are so bad that I unironically debug exclusively with printing().

Anonymous

Anonymous Thu 10 Mar 2022 22:43:21 No.14294144 Report

Quoted By: >>14294314

>>14293941

Does it is have a version of Simulink, or all the tool boxes of Matlab?

Anonymous

Anonymous Thu 10 Mar 2022 22:46:32 No.14294152 Report

Quoted By:

>>14294009
Not at all, just `sudo apt install julia` (or equivalent with whichever package manager you use) and you're done. From there you can open the REPL and install packages using the inbuilt package manager (Pkg). Alternatively, it runs on jupyter if you prefer.

Anonymous

Anonymous Thu 10 Mar 2022 23:07:17 No.14294199 Report

Quoted By:

It looks nice but I'm just gonna use python for now

Anonymous

View Same Google iqdb SauceNAO Bronze_full_helm_detail.png, 26KiB, 924x1495

Anonymous Fri 11 Mar 2022 00:30:22 No.14294314 Report

Quoted By:

>>14294144
Let's be honest, the MATLAB language/ecosystem is a monolithic nightmare. The only reason to use MATLAB are the toolboxes that come with it, specifically Simulink and some domain specific ones. I cannot say if Julia will ever come close to Simulink, there is some development going on (MTK.jl) but given that Python is more famous, exists for a long time already and still has no alternative, might suggest that a replacement will only come a long way. However, I am fairly certain that in the long run, Julia will become the standard for scientific computing, competing with Python as a viable alternative. Julias Type system makes it really easy to write "pseudocode" that is close to mathematical notation.

Anonymous

Anonymous Fri 11 Mar 2022 08:33:57 No.14295016 Report

Quoted By:

maybe. i was very impressed when i looked at examples.

Anonymous

Anonymous Fri 11 Mar 2022 08:46:04 No.14295039 Report

Quoted By: >>14295336

>>14293941
>1-based indices
D R O P P E D

Anonymous

Anonymous Fri 11 Mar 2022 08:55:41 No.14295056 Report

Quoted By:

>>14293941
I have a lecture where they're making us use this. The prof is super into the language and develops libraries for it. Additionally, a dude from my masters wrote the notebook thing for Julia (it's called Pluto).
The language seems to be growing in popularity and honestly I like it a lot, it helps specially that it's a language from mathematicians for mathematicians so everything in it just makes sense to me.

Anonymous

Anonymous Fri 11 Mar 2022 09:11:33 No.14295079 Report

Quoted By: >>14295919

>>14294026
Python is shit but it's easy

Anonymous

Anonymous Fri 11 Mar 2022 11:09:32 No.14295218 Report

Quoted By: >>14295752

>>14293941
tried v1.7.2 it works much better now but still needs a full minute to load/use a normal package and installing something like the ODE package took 7min. Still better than whats up on ubuntu repos where it took half a fucking hour.
I like how subplots works at least I can understand it as opposed to matplotlib.
The docs are horrible too many words they should keep it short and show examples first.
Any suggestions on libraries for solving PDE's with julia or something like taichi?

Anonymous

Anonymous Fri 11 Mar 2022 11:19:53 No.14295227 Report

Quoted By: >>14295361 >>14297111

>>14293941
Why not just use Python?
>Python is slow
If I'm doing numerical work, I'm using numpy. And if I'm using numpy correctly then all the heavy lifting is passed to C/Fortran.

Anonymous

Anonymous Fri 11 Mar 2022 12:47:32 No.14295311 Report

Quoted By: >>14297116

>>14293941
How do you do symbolic solving and stuff though?

Anonymous

Anonymous Fri 11 Mar 2022 12:53:33 No.14295318 Report

Quoted By: >>14297135

>>14293941
so... essentially python and mathematica had a retardbaby together?

Anonymous

Anonymous Fri 11 Mar 2022 13:05:54 No.14295336 Report

Quoted By:

>>14295039
oh dear

Anonymous

Anonymous Fri 11 Mar 2022 13:17:24 No.14295361 Report

Quoted By: >>14295416

>>14295227
Numpy isn’t fast, brainlet

Anonymous

Anonymous Fri 11 Mar 2022 13:30:32 No.14295388 Report

Quoted By: >>14297146

>simple hello world program is 200mb
non starter

>compile time latency in the seconds (or minutes)
non starter

Anonymous

Anonymous Fri 11 Mar 2022 13:58:15 No.14295416 Report

Quoted By: >>14295867

>>14295361
Wrong.

Anonymous

Anonymous Fri 11 Mar 2022 16:09:41 No.14295597 Report

Quoted By: >>14295759

Does anyone here ever play around with Scilab or Octave?

Anonymous

Anonymous Fri 11 Mar 2022 16:25:28 No.14295638 Report

Quoted By: >>14297156 >>14299322

>>14294032
>found the grandboomer
fuck you, retard. Fortran still holds the bench mark for scientific computing. If you are solving PDEs or large matrices without it you are retarded.

Anonymous

Anonymous Fri 11 Mar 2022 17:17:50 No.14295752 Report

Quoted By:

>>14295218
>Any suggestions on libraries for solving PDE's with julia or something like taichi?
https://juliapackages.com/p/voronoifvm

Anonymous

Anonymous Fri 11 Mar 2022 17:21:35 No.14295759 Report

Quoted By:

>>14295597
Yep
>Scilab
Way shitier version of simulink, nice that you can add latex shit under Blocks in order to explain, otherwise just insanely bad, specially the clock shit is super annoying.

>Octave
Matlab but unusable, everything is a fucking hassle with octave, even loading a package can make you rage.

Anonymous

Anonymous Fri 11 Mar 2022 18:07:04 No.14295852 Report

Quoted By: >>14295953 >>14295977

>>14293941
I like R and RStudio far too much. It's too simple that even my boomer colleagues 'get it'. I've heard good things about Julia and I think it is the future and will wreck out or incorporate all the alternatives.

Does Julia have a good IDE?

Anonymous

Anonymous Fri 11 Mar 2022 18:16:35 No.14295865 Report

Quoted By:

>>14293941
>1-indexed
The absolute state.

Anonymous

Anonymous Fri 11 Mar 2022 18:18:22 No.14295867 Report

Quoted By: >>14295869 >>14297332

>>14295416
It objectively is not fast, brainlet.

Anonymous

Anonymous Fri 11 Mar 2022 18:19:10 No.14295869 Report

Quoted By:

>>14295867
Wrong.

Anonymous

Anonymous Fri 11 Mar 2022 18:44:46 No.14295919 Report

Quoted By: >>14295925 >>14297103

>>14295079
Why is Python shit? If I have to make an arbitrary piece of software to do something it's always gonna be my first choice because of how easy it is to get something that just werks out the door.

Anonymous

Anonymous Fri 11 Mar 2022 18:46:20 No.14295925 Report

Quoted By:

>>14295919
which is to say I'll use other languages and tools when I need to due to some constraints - I've built something in most of the popular imperative style ones - but the default choice is pretty much always gonna be python for me.

Anonymous

Anonymous Fri 11 Mar 2022 18:54:59 No.14295953 Report

Quoted By: >>14296005

>>14295852
>Does Julia have a good IDE?
The current kinda go-to is VScode plus the julia plugin. Alternatively you can also use jupyter or pluto notebooks if you like that kind of style.

Anonymous

Anonymous Fri 11 Mar 2022 19:04:25 No.14295977 Report

Quoted By: >>14296005

>>14295852
R is full of redundant and horribly structured packages. It's faster if you're doing some course assignment in R but for just about any real application you're gonna be writing yourself Python is more practical.
>does X have a good IDE
In 1987+35 you can just install VS code and download an extension in 3 clicks.

Anonymous

Anonymous Fri 11 Mar 2022 19:12:28 No.14296005 Report

Quoted By: >>14296016 >>14296020 >>14299128

>>14295953
>>14295977
I've never heard of vscode. Is it just a general purpose IDE?

>>14295977
I've used Python before, but it just seemed like there were a lot of packages that needed to be installed. I haven't had any real issues with R, but then again, I don't really port anything into it. The base packages of R are pretty good for most statistical problems. I have noticed when I do go looking for R packages I find a lot of bloat that do things that base R already does pretty effectively. It's very weird imo. Like all of the 'tinyverse' memes. That whole ecosystem is weird to me since I've never needed it even for large project work.

Still I like it and find it very simple to use.

Anonymous

Anonymous Fri 11 Mar 2022 19:16:34 No.14296016 Report

Quoted By: >>14296020 >>14296032

>>14296005
>I've never heard of vscode. Is it just a general purpose IDE?
In its base form it's a text editor, but what it really is, is a mod of chromium that supports a fuckton a different plugins for just about anything you can imagine.

So for pretty much any language out there you can download a plugin that makes it an IDE - with a debugger, linter, build toolchain, various convenience functions, etc. But it's not limited to just programming languages. I've set it up before to do shit like interact and inspect SQL database content, serve as an editor for latex files (that automatically compiles when you save, does spellcheck, and has a separate sub-window with the PDF output open), etc.

Think vim or emacs but from the 2010s.

Anonymous

Anonymous Fri 11 Mar 2022 19:18:18 No.14296020 Report

Quoted By:

>>14296016
>>14296005
Also, it runs without any problems on linux. Got it both on my ubuntu "workstation" and arch desktopless gaming laptop.

Anonymous

Anonymous Fri 11 Mar 2022 19:23:32 No.14296032 Report

Quoted By:

>>14296016
Oh okay! That does sound nice. I've used Vim before, but the only general purpose text editor I regularly use is notepad++ and that's because it prints code nicely to a printer.

I have to jump between classified and unclassified systems a lot and you can't bring data down so you have to print and hand carry your code.

Anonymous

Anonymous Fri 11 Mar 2022 20:29:43 No.14296201 Report

Quoted By: >>14296204

all syntaxes are shit
that's why I never learned to program

Anonymous

Anonymous Fri 11 Mar 2022 20:30:59 No.14296204 Report

Quoted By: >>14296990 >>14297503

>>14296201
If you learn to program you realize that all syntaxes are piss easy. The hard part is figuring out how to solve your problem and how to build your system in a way that it's not a pain in the ass to debug and maintain.

Anonymous

View Same Google iqdb SauceNAO julia.png, 281KiB, 306x454

Anonymous Sat 12 Mar 2022 01:53:37 No.14296909 Report

Quoted By:

Bump based Julia

Anonymous

Anonymous Sat 12 Mar 2022 02:08:56 No.14296939 Report

Quoted By:

>>14294032
He's got a point though

Anonymous

Anonymous Sat 12 Mar 2022 02:09:57 No.14296941 Report

Quoted By: >>14297000

Redpill me on C,C++ for scientific computing / numerical analysis

Anonymous

Anonymous Sat 12 Mar 2022 02:26:42 No.14296982 Report

Quoted By: >>14297164

>>14293941
will it look good on a resume for a data analyst position? i'm currently learning R but will this also give me an edge?

Anonymous

Anonymous Sat 12 Mar 2022 02:28:51 No.14296990 Report

Quoted By:

>>14296204
+10 post

Anonymous

Anonymous Sat 12 Mar 2022 02:32:23 No.14297000 Report

Quoted By: >>14297004 >>14297081

>>14296941
You don't. Only time those are used is to write the core libraries and functions which are required to be as fast and efficient as possible. But they are then called by higher level languages for rapid development and ease of use.

Anonymous

Anonymous Sat 12 Mar 2022 02:34:01 No.14297004 Report

Quoted By: >>14297006 >>14297011 >>14297524

>>14297000
Right I always use python. So it is never worth me learning C++? Surely a situation where a loop would be unavoidable in a Python script I'd want to somehow run that bit in C for speed?

Anonymous

Anonymous Sat 12 Mar 2022 02:36:19 No.14297006 Report

Quoted By:

>>14297004
Because in principle Python is roughly as fast as a lower level language only if you treat it as a functional language right? As soon as you induce any sort of recurrences etc you're then using a slower loop?

Anonymous

Anonymous Sat 12 Mar 2022 02:42:19 No.14297011 Report

Quoted By: >>14297454

>>14297004
Python itself and its libs are written in C or C++. So unless you are writing new functionality for Python you don't need it.

> Surely a situation where a loop would be unavoidable in a Python script I'd want to somehow run that bit in C for speed?
In general that doesn't happen. You loop and call some function(s). Those functions will already to written in C and that's where all the CPU cycles will be used. The looping done by Python is trivial in comparison.

Saying all that the best programmers you will ever meet will know C and know it well. It teaches you the fundamentals of low level coding and what is going on under the hood when you use a higher level one. But when it comes to academia or enterprise coding it is rarely used.

Anonymous

Anonymous Sat 12 Mar 2022 03:11:22 No.14297081 Report

Quoted By: >>14297088 >>14297101

>>14297000
How about quant types working for Wall St.? Do they not code from scratch in C++?

Anonymous

Anonymous Sat 12 Mar 2022 03:14:03 No.14297088 Report

Quoted By:

>>14297081
Don't think so in general. If you code in C++ you've got to be pretty sure you can outperform the python library. Possibly, but usually a waste of time.

Most of what quants are doing with their programming is minimising FLOP count and that kind of thing by noticing symmetry in Hessians and that kind of thing.

Anonymous

Anonymous Sat 12 Mar 2022 03:20:45 No.14297101 Report

Quoted By:

>>14297081
No idea, not my field. But I do know high frequency traders do some funky shit even down to using machine code to write custom firmware for NIC cards.

In general though spending time writing a better algorithm will also beat out time spent on optimization for minimal gains.

Anonymous

Anonymous Sat 12 Mar 2022 03:20:55 No.14297103 Report

Quoted By:

>>14295919
Python is the 2nd best language for everything.

Anonymous

Anonymous Sat 12 Mar 2022 03:24:27 No.14297111 Report

Quoted By:

>>14295227
>not using numba
NGMI

Anonymous

Anonymous Sat 12 Mar 2022 03:26:11 No.14297116 Report

Quoted By:

>>14295311
https://juliasymbolics.org/

Anonymous

View Same Google iqdb SauceNAO 20170613_14DCGCENw-3.jpg, 54KiB, 436x654

Anonymous Sat 12 Mar 2022 03:34:28 No.14297135 Report

Quoted By:

>>14295318
>retard baby
more like a turbo autist baby that effortlessly outclasses that boomer shit. aka the alex honnold of numerical programming languages.

Anonymous

Anonymous Sat 12 Mar 2022 03:38:52 No.14297146 Report

Quoted By: >>14297223

>>14295388
>thinking 200mb base is big
>worries about compile time quicker than taking a shit
found the brainlet who doesn't process terabytes/petabytes of scientific data

Anonymous

View Same Google iqdb SauceNAO primenumberjumps-1.jpg, 57KiB, 515x469

Anonymous Sat 12 Mar 2022 03:43:33 No.14297156 Report

Quoted By: >>14297177 >>14297422 >>14298375

>>14295638
>uses a language with goto statements
goto my balls

Anonymous

Anonymous Sat 12 Mar 2022 03:45:42 No.14297160 Report

Quoted By: >>14297179 >>14297192 >>14297244 >>14297422

Does anyone still use Maple or Mathematica? I used to type set all my hw stuff with Maple.

Anonymous

Anonymous Sat 12 Mar 2022 03:47:37 No.14297164 Report

Quoted By: >>14297178

>>14296982
No. SQL, R, and Python are what the boomers who hire data analysts want.

Anonymous

Anonymous Sat 12 Mar 2022 03:53:29 No.14297177 Report

Quoted By:

>>14297156
You've been filtered. There is nothing wrong with GOTO when used correctly.

Anonymous

Anonymous Sat 12 Mar 2022 03:54:18 No.14297178 Report

Quoted By:

>>14297164
Excel

Anonymous

Anonymous Sat 12 Mar 2022 03:54:49 No.14297179 Report

Quoted By:

>>14297160
I'm currently doing a graduate math module that is using Maple. It's a shit piece of software.

Anonymous

Anonymous Sat 12 Mar 2022 04:02:35 No.14297192 Report

Quoted By:

>>14297160
Literal cancer of a software

Anonymous

Anonymous Sat 12 Mar 2022 04:23:54 No.14297223 Report

Quoted By:

>>14297146
It means the ecosystem will never grow. Nobody writing software in another language will ever call into a julia library if they have to pay those exorbitant penalties to use it. On the flipside, library developers who want to reach the greatest number of users will continue writing/maintaining libraries in C/C++/fortran and not julia, because julia and every other language can call into C/C++/fortran but no other language will call into julia.
No ecosystem, no reason to use it.

Anonymous

Anonymous Sat 12 Mar 2022 04:34:44 No.14297244 Report

Quoted By:

>>14297160
I use mathematica, but that isn't really a julia competitor. Julia specializes in high performance numerical methods, not symbolic manipulation, although there are some libraries for it.

Anonymous

Anonymous Sat 12 Mar 2022 05:24:36 No.14297332 Report

Quoted By:

>>14295867
Numpy is heavily optimized C code that uses processor specific SIMD features. There's nothing faster on a CPU.

Anonymous

Anonymous Sat 12 Mar 2022 06:23:22 No.14297422 Report

Quoted By:

>>14297156
Why people ignore that Fortran is being developed after the 70's?
I've wrote my code in Fortran 2008, it is in lower cases and doesn't use GOTO, even used an arbitrary precision package and it runs like hell.

>>14297160
I use Mathematica. I do a lot of symbolic calculations and even some numerical. It may not be the fastest, but the code itself is very easy to understand and write. For example, I wrote a code that solve a non-linear PDE using Newton-Raphson, in Mathematica it took ~20 lines (out of which the matrix construction is ~4 lines) while in Fortran it took ~600 lines (out of which ~300 for the matrix construction alone)

Anonymous

Anonymous Sat 12 Mar 2022 06:51:27 No.14297454 Report

Quoted By: >>14297485

>>14297011
>C
>low level coding
>what is going on under the hood
lmao. Your computer is not a fast PDP-11.
https://queue.acm.org/detail.cfm?id=3212479

Anonymous

Anonymous Sat 12 Mar 2022 07:13:30 No.14297485 Report

Quoted By:

>>14297454
What a shitty article. tl;dr no one writes hard crafted machine code anymore because of complicated CPU architectures. also optimization is hard.

Anonymous

Anonymous Sat 12 Mar 2022 07:28:44 No.14297503 Report

Quoted By: >>14297529

>>14296204
That's funny because I do all of that for hardware just fine but not for software.
I strongly dislike programming syntaxes and never learned to program because of it. But I did learn analog circuit design extremely easily and I absolutely love solving problems with it.

Anonymous

Anonymous Sat 12 Mar 2022 07:36:23 No.14297524 Report

Quoted By:

>>14297004
>Surely a situation where a loop would be unavoidable in a Python script I'd want to somehow run that bit in C for speed?
That's why you use numpy and other wrappers for C.

Anonymous

Anonymous Sat 12 Mar 2022 07:37:23 No.14297526 Report

Quoted By: >>14297703

The reason why I dislike programming syntaxes is because I feel like language is only something for people, not for machines. I should only use english to talk to other people, NEVER to a machine.
When I make a schematic, there's no words involved other than labels for convenience.

Anonymous

Anonymous Sat 12 Mar 2022 07:38:18 No.14297529 Report

Quoted By: >>14297554

>>14297503
I can also do analog circuit design, digital HDL work, mechanical design. Programming is just more of the same shit. Syntaxes aren't shit, you're just lazy.

Anonymous

Anonymous Sat 12 Mar 2022 07:48:46 No.14297554 Report

Quoted By: >>14297564

>>14297529
I got as far as arduino blinky and I made my own custom rgb fade but yeah I hate programming because syntaxes suck. You guys need to work on making one that's purely visual or something like a DAW.

Anonymous

Anonymous Sat 12 Mar 2022 07:52:23 No.14297564 Report

Quoted By: >>14297593

>>14297554
>I hate programming because syntaxes suck
You hate programming because you are too lazy to learn it. Everyone makes tons of mistakes at the start. Get an IDE with syntax highlighting, learn to use a debugger, and you'll get much better very rapidly.

Anonymous

Anonymous Sat 12 Mar 2022 08:09:24 No.14297593 Report

Quoted By: >>14297626 >>14297634

>>14297564
oh okay sudo apt-get autoremove program program-plugin-registry
yeah that makes sense

Anonymous

Anonymous Sat 12 Mar 2022 08:30:31 No.14297626 Report

Quoted By: >>14297655

>>14297593
That's not even a programming language. Just a command and arbitrarily named arguments for it.

Anonymous

Anonymous Sat 12 Mar 2022 08:32:52 No.14297634 Report

Quoted By: >>14297655

>>14297593
Also I don't think apt autoremove takes additional arguments. And linux doesn't have a registry.

Anonymous

Anonymous Sat 12 Mar 2022 08:48:18 No.14297655 Report

Quoted By: >>14297663

>>14297626
>>14297634
[
{
{Oh.} {{{{Okay} {then} {what about} {this?}}}
}
]

Anonymous

Anonymous Sat 12 Mar 2022 08:50:18 No.14297659 Report

Quoted By:

>Getting filtered by abstract notation

Even third world code sweatshops can handle that

Anonymous

Anonymous Sat 12 Mar 2022 08:54:02 No.14297663 Report

Quoted By:

>>14297655
What syntax is that supposed to be? Malformed json? Nigger the reason you can't program is that you suck, not that programming languages suck. Literal turd world monkeys can do it, and you can't?

I bet your ego got bruised when you made some syntax errors with your babby tier hello world project and it wouldn't compile on the first try, so instead of reading the debug prompt and fixing the issues iteratively like a normal person you decided programming is shit.

Anonymous

Anonymous Sat 12 Mar 2022 09:16:22 No.14297703 Report

Quoted By: >>14297715 >>14297837

>>14297526
>When I make a schematic, there's no words involved other than labels for convenience.
That's what the source of a program is. It's a text based schematic of symbols arranged in a way that conforms to a specified formal grammar. Stop thinking in terms of "words" and think in terms of symbols.
Write a lexer. It's a good exercise to help understand this.
https://en.wikipedia.org/wiki/Lexical_analysis

Anonymous

Anonymous Sat 12 Mar 2022 09:22:21 No.14297715 Report

Quoted By:

>>14297703
People who write code for a living do think of them as words, just not words in English. They're words in a tightly constrained, domain specific language. But ultimately it's still the language processor in your brain doing the heavy lifting. When you've gotten past the initial hump, it's largely automatic. And picking up a new language for basic usage takes a few hours, it's the surrounding toolset that takes longer to learn and many years to master.

Once you know how to code, writing code feels like speaking the language to express your ideas. That's exactly why programming languages are such an incredibly powerful abstraction, they take something that is very difficult for a human brain - arranging machine code instructions - and turn it into a process that is in line with how our brains are set up to work and think naturally.

Anonymous

Anonymous Sat 12 Mar 2022 11:03:08 No.14297837 Report

Quoted By: >>14298553

>>14297703
I can't understand a single word of that wikipedia article. How come I'm stupid in that regard but I can understand analog stuff easily? Nobody had to teach me how to read schematics I just kinda figured it out after playing with batteries and lightbulbs.

Anonymous

View Same Google iqdb SauceNAO Add_all_numbers_block.png, 37KiB, 428x246

Anonymous Sat 12 Mar 2022 11:14:55 No.14297848 Report

Quoted By:

reading the wikipedia article for syntax I found pic related
now THIS is what programming should look like! Your text based shit is hideous to look at.

Anonymous

Anonymous Sat 12 Mar 2022 17:32:54 No.14298375 Report

Quoted By: >>14298569

>>14297156
>>uses a language with goto statements
boohoo programming snob. You know everything just compiles down to assembly anyway and there all languages are reduced to jumps, right?

The fact remains Fortran does scientific computations faster than anything else, so you can use Fortran or wait.

Anonymous

Anonymous Sat 12 Mar 2022 18:58:44 No.14298553 Report

Quoted By:

>>14297837
Code is the same. You don't need to understand the theory behind compilers to just write working code, the other anon is an idiot.

Anonymous

Anonymous Sat 12 Mar 2022 19:05:04 No.14298569 Report

Quoted By: >>14298655

>>14298375
>You know everything just compiles down to assembly anyway and there all languages are reduced to jumps, right?
Programming languages are high level abstractions designed to get you to quickly and reliably produce very large amounts of machine code. Gotos may be how shit works at the processor instruction level but they're still objectively bad for code organization, readability and maintainability.

The reason you can spout bs like this is that you've never had to work with a large code base.
>fortran is faster than anything else
C, C++, Fortran, Rust, native assemblers and anything else that compiles directly to machine code are all roughly equal in performance, so long as they've got a well-supported compiler front-end. Whatever variance there is due to platform differences and front-end implementations is irrelevant for all practical purposes.

Anonymous

Anonymous Sat 12 Mar 2022 19:42:13 No.14298655 Report

Quoted By: >>14298702

>>14298569
fuck you and your false equivalence

https://www.google.com/search?q=what+is+the+fastest+programming+language+for+scientific+computing&rlz=1C1CHBF_enUS960US960&oq=what+is+the+fastest+programming+language+for+sc&aqs=chrome.1.69i57j33i22i29i30l9.13285j0j7&sourceid=chrome&ie=UTF-8

Anonymous

Anonymous Sat 12 Mar 2022 20:07:33 No.14298702 Report

Quoted By: >>14298766

>>14298655
>muh google answers
kek, not even a word on what asshole it was pulled from.

There is no difference in how fortran, c, c++ gets rendered in machine code for the vast majority of applications. There may be some that were written ages ago in fortran and haven't been updated, but that's got fuck all to do with the language. There is always some random variance in how compilers handle various structures, but ultimately both C and fortran work directly with the memory model of the machine so there's no room for a systemic advantage for either. C++ and Rust have largely lossless abstractions, too, so though they might occasionally lose out something it's only a small constant time difference and doesn't fucking matter in practice.

You are a terminal brainlet who has never worked on an actual codebase. You've definitely never had to debug code written by other people.

Anonymous

Anonymous Sat 12 Mar 2022 20:31:25 No.14298766 Report

Quoted By: >>14299105

>>14298702
>You are a terminal brainlet who has never worked on an actual codebase.
I have worked on an actual codebase

>You've definitely never had to debug code written by other people.
I have had to debug code written by other people.

and BTW, Fortran holds the benchmark for scientific computing.

Anonymous

Anonymous Sat 12 Mar 2022 23:06:34 No.14299105 Report

Quoted By: >>14299198

>>14298766
Can you clarify something for me? Because I'm not sure if I understand... Which language holds the current benchmark for scientific computing?!?

Anonymous

Anonymous Sat 12 Mar 2022 23:21:52 No.14299128 Report

Quoted By: >>14299207

>>14296005
Tidyverse is fantastic, at the very least dplyr and tidyr make your life 100% easier for data analysis.

Anonymous

Anonymous Sat 12 Mar 2022 23:59:26 No.14299198 Report

Quoted By: >>14299221

>>14299105
>Which language holds the current benchmark for scientific computing?!?
Fortran. Don't be amused by trendy faggots who only ever talk about python/GIT hub/Linux. I spend alot of time solving PDEs. Fortan is the fastest.

Anonymous

Anonymous Sun 13 Mar 2022 00:05:05 No.14299207 Report

Quoted By: >>14299338

>>14299128
I don't know, I have never seen a need for it. Lists and data frames work more than fine and I don't like adding packages that just rework how things already work in the base language.

Anonymous

Anonymous Sun 13 Mar 2022 00:15:40 No.14299221 Report

Quoted By: >>14299245

>>14299198
Thank you! I never would have gotten that from the thread above. You seem very knowledgeable and kind.

Anonymous

Anonymous Sun 13 Mar 2022 00:31:23 No.14299245 Report

Quoted By: >>14299263 >>14299365

>>14299221
The standard is fortran and C with Python and other high level languages used for initial modeling or things that aren't incredibly Computationally expensive (I.e. if it doesn't need a supercomputer it will probably be written on Python or something high level like that). Fortran is used a lot for older scientific projects and lots of old organizations use it pretty frequently. C/C++ I would say is more the modern standard for efficiency.

Most people, (i.e. most people in STEM) don't need the computational power of supercomputing and don't need the high efficiency gained from C/C++ and fortran. This is why you see Python, R, matlab, etc. All the time. MANY of these languages also have converters to C so you can write things up in a high level language and then convert it down and clean up as needed.

Anonymous

Anonymous Sun 13 Mar 2022 00:41:46 No.14299263 Report

Quoted By: >>14300701

>>14299245
>Most people, (i.e. most people in STEM) don't need the computational power of supercomputing and don't need the high efficiency gained from C/C++ and fortran
yea, you could just guess. why do a finite element simulation? the solution is probably like 5, right? I mean the algorithm will take a few years to terminate in Python, and Fortran is old, so the answer is just 5.

Anonymous

Anonymous Sun 13 Mar 2022 01:21:09 No.14299322 Report

Quoted By: >>14299355

>>14295638
>If you are solving PDEs or large matrices without it you are retarded.
You would have to go out of your way to solve PDEs and matrices without it since it's what powers any reasonable BLAS. If you are seriously proposing that most scientists should implement their simulations in FORTRAN directly you are braindead. Getting the top end of performance is a specialty in itself, and certainly not one that would be feasible to teach to every scientist.

Anonymous

Anonymous Sun 13 Mar 2022 01:35:20 No.14299338 Report

Quoted By:

>>14299207
I just feel in my work that tidyverse functionality has made the analysis process much more standardized, efficient, and replicable. The pipe is such a useful operator to structure multiple complex operations.

Anonymous

Anonymous Sun 13 Mar 2022 01:45:17 No.14299355 Report

Quoted By: >>14299362

>>14299322
>boohooo I can't understand simple programming and need a GUI to do all the work for me.

dont be this guy

Anonymous

Anonymous Sun 13 Mar 2022 01:51:23 No.14299362 Report

Quoted By: >>14299420

>>14299355
I have actually built an HPC library for supercomputing using C w/ OpenMP+MPI, FORTRAN and a little handtuned assembly for good measure. No one in their right mind can expect a scientist, who should be worried about model specification, to deal with hand-optimizing. It's absolutely insane to even suggest it. The people who write the subprograms spend an inordinate amount of time to do so, which is why FORTRAN takes the top spot. You are supposed to piggyback off their libraries, not try (and mostly likely fail) to implement them. This attitude outs you as someone who has never actually done this stuff professionally.

Anonymous

Anonymous Sun 13 Mar 2022 01:53:12 No.14299365 Report

Quoted By:

>>14299245
I was being sarcastic, seeing as how it was mentioned 10 times in this thread. But whatever. +1 for your autism.
BTW, some compute-heavy earth science models are now being ported from Fortran to Julia.
https://discourse.julialang.org/t/why-use-julia-to-build-an-earth-system-model/53180

Anonymous

Anonymous Sun 13 Mar 2022 02:36:55 No.14299420 Report

Quoted By: >>14299443

>>14299362
>No one in their right mind can expect a scientist, who should be worried about model specification, to deal with hand-optimizing. It's absolutely insane to even suggest it
I taught myself to program when I was in elementary school, faggot. I just can't fathom how dumb "scientists" are I guess.

Anonymous

Anonymous Sun 13 Mar 2022 02:51:24 No.14299443 Report

Quoted By:

>>14299420
>I just can't fathom how dumb "scientists" are I guess.
No, the issue is that getting good performance requires more than just making a naive but functional implementation. Speedups are found from understanding the hardware (e.g. avoiding cache misses) and the mathematics of the model (e.g. treating sparse vs. dense matrices, arrowhead vs. triangular vs. Toeplitz, etc.). This kind of programming is its own specialty in numerical analysis. The vast majority of scientists and teams will get better results using off-the-shelf libraries rather than rolling their own.

Anonymous

Anonymous Sun 13 Mar 2022 12:01:03 No.14300153 Report

Quoted By:

>>14293941
slow as fuck startup time but looks cool
packagecompiler.jl takes like 5 minutes to make a hello world exe as well and the executable is 200mb

I would love if they had went the java route and used bytecode JIT, instead for some stupid reason they JIT compile the source code

Anonymous

View Same Google iqdb SauceNAO 1622524355541.jpg, 15KiB, 474x402

Anonymous Sun 13 Mar 2022 16:43:51 No.14300502 Report

Quoted By:

>>14293941
I don't use languages named after women.

Anonymous

Anonymous Sun 13 Mar 2022 17:05:23 No.14300553 Report

Quoted By: >>14300770

>>14294023
Only two kinds of people use fortran:
1) Boomer professors
2) Lazy grads who only want to learn fortran so they can use their prof's existing code instead of having to make anything new

Anonymous

Anonymous Sun 13 Mar 2022 18:05:56 No.14300701 Report

Quoted By: >>14300792

>>14299263
See people like this you shouldn't listen to unless you want to be unemployable. There is no one language to rule over them all and you will never be expected to know just one unless you get stuck in a position for the rest of your life.

Most engineers don't even touch C let alone fortran. They almost all use fucking matlab for Christ sake.

Anonymous

Anonymous Sun 13 Mar 2022 18:28:36 No.14300770 Report

Quoted By:

>>14300553
you are an idiot if you use anything but the fastest language to do scientific computing.

Anonymous

Anonymous Sun 13 Mar 2022 18:34:58 No.14300792 Report

Quoted By:

>>14300701
>See people like this you shouldn't listen to unless you want to be unemployable
who ever said anything about employment, retard? OP is "Julia is the language of /sci/entists" which mean scientific computing. If you want to make money just learn2resume

>>14300701
>There is no one language to rule over them all
yeah, that why I keep specifying SCIENTIFIC COMPUTING

>>14300701
>unless you get stuck in a position for the rest of your life.
since when is this /wagie/, again?

>>14300701
>Most engineers don't even touch C let alone fortran.
like previous anon already said BLAS is Fortran native and Octave, Matlab, Julia, Phython ... etc are just wrappers for Fortran libraries, retard.

>>14300701
>They almost all use fucking matlab for Christ sake
so they use a wrapper for Fortran libraries, but they don't use Fortran? Fuck you retard. You have any idea how much faster Fortran is without Matlab sitting ontop of it? My estimate is thousands of times.

Anonymous

View Same Google iqdb SauceNAO 81500379-2e8cfa80-92d2-11ea-884a (...).gif, 3MiB, 430x475

Anonymous Sun 13 Mar 2022 23:55:27 No.14301451 Report

Quoted By:

you can call C and Fortran from Julia.

Capcode	All Only User Posts Only Moderator Posts Only Admin Posts Only Developer Posts
Show Posts	All Only With Images Only Without Images
Deleted Posts	All Only Deleted Posts Only Non-Deleted Posts
Ghost Posts	All Only Ghost Posts Only Non-Ghost Posts
Post Type	All Only Sticky Threads Only Opening Posts Only Reply Posts
Results	All Grouped By Threads
Order	Latest Posts First Oldest Posts First

Your latest searches