# Julia

In Computer.

tags: julia

I had planned to write a blog post about how great Julia is. I spent many years using Matlab, both in grad school and in finance. Since then, I've been using NumPy for mathy things, but it feels really awkward, since it's just sitting on top of another language, with it's own special data structures, rather than people able to do serious computation on regular Python data structures. And the NumPy structures feel weird and non-intuitive to me, especially when dealing with vector calculations -- all the operators default to element-wise!

So, I was pretty excited about Julia. The data structures and operators are similar to Matlab, but it's open-source, and much faster! Which isn't a minor thing, since prototyping in Matlab was easy, but it was unuseably slow for any big calculations.

However, every time I've tried to actually use Julia for anything other than the simplest task, I've run into problems. Libraries that just won't install. Inexplicable errors that eventually all trace back to some massive change to the language (no matter how recently the tutorial, or whatnot, was written). And, so, I just can't do it. I'm spending way too much time trying to translate code from Julia-of-two-months-ago to Julia-of-today. And even if I succeed, will it still run two months from now?

I'm still excited about the posibilities of Julia, but it's jut too unstable to use for anything real right now. I'll be sticking with NumPy for now, but I'll probably try Julia again once it stabilizes.