Julia's flagship feature is that it loads one of either of the C libs Intel's OpenMP or mpi as normal part of execution to facilitate thread/heterogeneous parallelism, isn't it? On Mon, Feb 3, 2020 at 1:41 PM Lawrence D'Oliveiro <ldo(a)geek-central.gen.nz> wrote:
Came across this <https://github.com/mitmath/julia-mit> repo, which is an intro to the Julia language. Apparently this is increasingly being used in place of MATLAB in numerical-computation courses at MIT.
Why not Python? Because it is fine as long as you can use existing computational kernels, but as soon as you need custom ones, it gets slow if you do them in Python, or requires special skills to write them in C or Fortran, and make them callable from Python. Julia incorporates a “just-in-time” compiler using LLVM to achieve speed close to that of C in this case.
Why not MATLAB? Because it is proprietary:
Unlike Matlab, [Julia] is free/open-source software, which eliminates licensing headaches and allows you to look inside the Julia implementation to see how it works (since Julia is mostly written in Julia, its code is much more readable than a language like Python that is largely implemented in low-level C).
They do like using Julia in Jupyter notebooks. I see a few .ipynb files in the repos for other courses (along with a few .pptx ones as well, unfortunately). Also several “.jmd” files, which look like some variety of markdown. _______________________________________________ wlug mailing list -- wlug(a)list.waikato.ac.nz | To unsubscribe send an email to wlug-leave(a)list.waikato.ac.nz Unsubscribe: https://list.waikato.ac.nz/postorius/lists/wlug.list.waikato.ac.nz