Hacker Newsnew | past | comments | ask | show | jobs | submit | whyrt12's commentslogin

I WANT to use pytorch, but no bayesian learning or stochastic nodes like in edward. Any chance there are plans to for a compatibility layer with Edward or roll your own bayesian stuff?

Also, have you looked at Numba to do the jitting? Probably best not to have yet another separately maintained python JIT.


as core-devs, we dont plan to build-in something like Edward. However, folks in the community are brewing something:

https://discuss.pytorch.org/t/bayesian-computation-in-pytorc... https://discuss.pytorch.org/t/distribution-implementations/4...


Once you can compile a julia app- front end, back and probabilistic prog/ML to web assembly and have it run in browser and mobile, it will skyrocket in popularity.


Why? The only current benefit of Julia afaict is the tracing jit. If you run the tracing jit on web assembly then it's giving up most of its performance benefits. And Python could be built on web assembly as well.

But who knows. Weird things seem to become popular despite all the negative points.


It does not have a tracing JIT, and its speed is by far not the only benefit. See link below

It can precompile very fast code before runtime.

Python will require an interpreter and or hefty runtime.

https://discourse.julialang.org/t/julia-motivation-why-weren...


Forgive me, it's not a tracing JIT but just LLVM's JIT.

Precompiling in Julia is extremely not-straight-forward. You would think you just use --compile and it would work; but it doesn't at all.

Also, at ~850kb, Python's runtime is not that hefty. It's intended to be embedded and while it's quite a bit larger than lua's 200kb, but smaller than libjulia's 16mb.


Right, It can currently precompile to some extent, but full source-to-binary-blob-compilation is on the roadmap. See here: http://juliacomputing.com/blog/2016/02/09/static-julia.html

Julia's runtime includes its compiler and full huge standard lib, but of which are eventually going to be split off, IIUC.

The former because of static compilation potential and the latter into modules that can be included piecemeal.


In what way?


Do you do data science?


Yes, mostly for finance.


What python 3 features are painful and why?


Golang is far far inferior in many ways, for scientific programming. For starters, it doesn't have generics.


How close are we to web assembly compilation?


I'm not aware of anyone working on it. There were some attempts with Emscripten which ran in to unsupported instructions; some of those may be supported by newer versions of Emscripten, but I haven't heard of any attempts to get this going recently.


Gotcha. Is it definitely technically feasible, or is that still unknown?


For Julia core it should be feasible, but possibly requiring patches to Emscripten, so the level of effort is still unknown. The other issue is the C and Fortran numerical libraries -- some of which use inline-assembly, others (Fortran) don't have a working LLVM frontend as far as I know.


Got it


Julia has tons of general purpose development going, including an mvc web framework (genie.jl) and a reactive web app framework called (escher.jl)


First, sorry for being a bit heavy on the hyperbole and saying "0%"; that is almost guaranteed to be wrong as a statement and also let me say I don't know Julia very well at all.

I want to clarify that when I say "0% web development going on", I don't mean that scientific programmers don't do web development (they do! a lot!), I meant that people don't pick it those languages in general if they are only doing web development without a numerical/scientific/statistical aspect to it.

What would be interesting is, do you know of any teams or companies using Julia in anger in a non-scientific setting, with programmers from a non-scientific background?

The scientific Python community certainly makes good use of all the Python web tools!, and being a "scientific stack" in no way precludes the need for general purpose frameworks that can also be used by others. It's at least as much about people and community and habits as about the tooling...


I think its just a matter of time before this happens (its mature enough or an important app is released).

Julia is designed for general purpose computing from day one, but this community is not dealing with the same painpoints that scientists have been.


> do you know of any teams or companies using Julia in anger in a non-scientific setting, with programmers from a non-scientific background?

It's catching on among early-adopters in finance. What do you consider that?


Is it? Where are they? And how many are they?


All over, but with decent sized groups in NYC and London. Hard to quantify exactly, but see the sponsors list of the juliacon events for some examples.


Even APL has a couple of web frameworks...

Having a few libraries of each kind (of varying quality and with very small adoption) != tons.


What is your opinion on both languages?


My take is from scientific computing perspective. But as my opinion isn't that interesting, I take this opportunity to cite some julia-python-language-relatedtopics on HN.

I'm still betting on Julia https://news.ycombinator.com/item?id=11516374 and, since I do not use it for "production", I'm not giving up https://news.ycombinator.com/item?id=11692155

That said, if it's all about speed, then Python can be, well, faster: https://news.ycombinator.com/item?id=10735840 (especially so if you need BigInt, these still suck in Julia.)

Recently, discovered Go(language). Clean design rules! https://news.ycombinator.com/item?id=4158865, (but you shouldn't take it for what it is not https://news.ycombinator.com/item?id=12330573)


Can tou explain why qm and bell's inequality do not categorically undermine determinism and prove deep randomness ?

Is it because on a macro scale amd normal energy levels effects are negligible? Because there are well followed probability distributions?


Bell's inequality refutes theories with local hidden variables (e.g. some unknown property of particles or spacetime). That leaves two possibilities: no hidden variables (the results truly are non-deterministic) or non-local hidden variables (e.g. some property which exists outside the spacetime which we're constrained to)


There's a third possibility, superdeterminism, in which literally everything is determined by previous conditions, so that no communication is needed, even via a property outside of our spacetime, because everything was determined by preconditions.

Think of it this way: you set two clocks to the same time. The next day, they both have the same time on them. This isn't because the clocks are communicating via nonlocal hidden variables, it's because their precondition was the same and the sequence of events which followed was deterministic.


Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: