Is anyone in scientific/engineering computing actually using Julia in production? What is the state of industry uptake? Is it worth my time to learn if I can already Fortran, Matlab, and Python may way through anything I need?
>>59216154
During last year's ECMI modelling week there was one project that used it. It could replace Matlab some day if it grows the same quality of software suite.
>>59216177
Jupyter Zeppelin Beaker etc (one or multiple of them) and all their more common languages already do this. There is no opening.
>>59216207
The Julia debts make some pretty bold performance statements on their website. Are they blowing smoke or is Julia really as fast as C and Fortran?
>>59216428
Julia is fast sure, I don't see where they pretend it's as fast as C or Fortran though.
>>59216460
"the languageās design allow it to approach and often match the performance of C" http://julialang.org/
>>59216154
why is it beter then java?
>>59216495
Yes, often, not "most of the time" or "usually". Benchmarks at the bottom of the homepage show it's not quite there for everything. And it's true for simple tasks I often get the same performance with C than I do with Julia. The main issue is it's JIT'd so you have some overhead for big programs. Since I can't use Matlab anymore I clearly prefer Julia as Octave is just plain slow.
>>59216154
Gimme the quick rundown, what does it do that nobody else does?
>>59216495
In context of JIT compilers for dynamic languages, "often faster than C" means "in particular in numeric stuff".
Though I'd wager Julia is in a niche where it makes sense.
>>59216428
I'm sure it's fairly fast? A lot of languages are when you code the right way (which isn't always how you'd actually bother to code and compilers that deal with THAT conversion are not easily finished even on the 20y+ much used languages, so that will be a while yet...)
I'll also point out that the people that need fast are actually using Apache Spark, Akka, Cassandra, ... and the supported languages (which are more than just Scala & Java this is written in) because ultimately they're not interested in immediately tweaking every last CPU cycle out of an individual machine for a 5% improvement, but in throwing more machines at the problem for a 95% or so improvement each machine.
Jupyter etc. also got Spark backends REALLY damn quick.
>>59216154
Gonna need source on that pic
>>59216606
>59216606
You can maybe fap to the reflection
>>59216738
I considered it but that's just not gonna do it
>>59216495
yes that's the pretending that the other anon was referring to, well done
>>59216154
Post source on image
>>59217264
>>59216606
no need to thank
>>59217264
>>59216606
Found origimal pic on tinyeye. Her name is Yara Eggimann.
>>59217390
post the link please.
http://www.eroticbeauties.net/model/yara-eggimann/
>>59217460
https://tineye.com/search/e9f2e5bc38de81a9a347a05a8a660ddc61d731c5/
https://www.google.se/search?q=yara+eggimann+nude+beach
>>59216580
For numeric stuff, interpreted languages just call specialized C or Fortran code anyway, so they are all basically going to run at the same speed, so what is the point?
>>59216606
I think it's Julia something
Is this what they use for the bubble nudes?
>>59216154
Used to do computational models in grad school. I briefly looked into Julia but we were already all in on Fortran and Matlab, and newer shit was Python, so why bother? It really doesn't offer anything compelling and at this point is still too beta and glitchy to be worthwhile anyway.
>>59216177
python's SciPy stack already replaced MATSHIT
>>59218987
Call me when Python has a Simulink alternative.
>>59216207
Jupyter stands for JuliaPythonR. It's a notebook that uses one of those languages (or others), and not a language.
>>59218987
>SciPy stack already replaced MATSHIT
lol
>>59216154
>>he thinks there is such a thing as production in scientific computing
there is no production, you just gotta get it barely working
I shall use this opportunity to say something to the mods while i have their attention:
MODS YOU FAGGOTS
DO A BETTER JOB DELETING THREADS ABOUT LINUSKEKTIPS AND SUCH SHIT BECAUSE THOSE ARE NOT TECH CHANNELS, THEY ARE MARKETING CHANNELS
INFORMATION COMING FROM THOSE CHANNELS NEVER EDUCATES, BUT RATHER MISINFORMS
AND ADVERTISING MARKETING CHANNELS IS AGAINST THE RULES
MAKE A LIST OF LEGIT TECH CHANNELS AND STICK TO IT, EVEN IF IT MEANS ONLY 5% OF THE YOUTUBE CHANNELS FALL INTO IT
>>59219525
https://github.com/masfaraud/BMSpy/wiki
>>59219960
Sauce?
>>59219960
Thanks for the mammaries, anon.
>>59219960
Seey
>>59220069
It's in the thread you lazy
>>59217898
The point is not to write C and Fortran because they suck as does their deployment and - more important - creating bindings also sucks and - also important - linking stuff creates some bottlenecks, if you call small code fragments very often.
The last reason is btw the reason why they JIT compile C libraries in Truffle/Graal in some JRuby9000 thing and why you'd want to wait for WebASM to have direct DOM access, before creating a DOM heavy WebASM application.
>>59219927
This is true.
/thread
>>59216154
Hey, who has the original background pic??
>>59216154
Lisp like macros in a language designed for numerics is really nice. It leads to crazy awesome things like this: https://github.com/Jutho/TensorOperations.jl
>>59220220
>>59219960
Just scroll down the damn thread for your fap material, just how lazy can you be?