| layout | title | comments | category | tags |
|---|---|---|---|---|
post |
Julia - Kyle Barbary |
true |
posts |
meeting julia |
About 25 people.
Kyle is a postdoc in the Berkeley Center for Cosmological Physics and a BIDS fellow. Like many people, he has a website.
Julia is a high-level language (like Python) that emphasizes performance. Slides and Jupyter notebooks from the talk can be found in this Github repository.
If you're living in future and the link is broken, look in https://github.com/kbarbary/talks/
Julia solves the two language problem, where high-level languages are easy to program in, but they use some other low level language on the backend. Julia, however, only uses Julia.
Fundamentally, Julia was created under the idea that dynamic languages don't need to be slow. Julia seeks to be as fast as C, dynamic as ruby, useful as python, etc...
Julia is pretty fast. In some examples, very very fast. One of the cool things it does is to compile a function only on the first time you run it. Later runs are faster than the first.
The syntax seems really similar to python, except:
;takes you to the shell?takes you to helpbackspaceto get out of the shell or help- supports unicode, like python 3
- you can use
ipython notebook --profile juliato start IJulia typeof(var)gives the type of the variable var- string interpolation is neat. You can use
$varin a string and it will be expanded. - Functions and loops use an explicit
end - Functions can be written like
f(x) = 2x^2 + 3x + 1 - arrays can be either homogeneous or heterogeneous. If heterogeneous, the
array type is
Any. - array type can be explicitly defined
- one-based indexing
- ranges are inclusive at both ends
- the
code_nativefunction gives the machine code for any function. Pretty sweet.
Rather than importing packages, the using syntax is used and the macros in
that package are called with @func
No lightning talks.