ChrisRackauckas

7 ms·

31.
by ChrisRackauckas 1y ago
That got its own whole post: https://juliahub.com/blog/why-dyad-a-perspective-for-modelic...
32.
Why Dyad?: A Perspective for Modelica Users(juliahub.com)
3 points by ChrisRackauckas 1y ago|0 comments
33.
by ChrisRackauckas 1y ago
Julia does not just have bindings to CUDA. Native Julia code can compile to build .ptx kernels https://cuda.juliagpu.org/stable/development/kernel/ . This same code can also generate kernels for AMD GPUs, Inte
34.
Dyad: Making Hardware as Easy as Software(juliahub.com)
3 points by ChrisRackauckas 1y ago|1 comments
35.
HuggingFace deprecates TensorFlow and Jax support(twitter.com)
5 points by ChrisRackauckas 1y ago|0 comments
36.
Machine learning with hard constraints: Neural Differential-Algebraic Equations(stochasticlifestyle.com)
2 points by ChrisRackauckas 1y ago|0 comments
37.
by ChrisRackauckas 1y ago
It's not uncommon to dispatch on 3 or more arguments. Linear algebra specializations are one case where I tend to do this a lot, for example specializing on structured matrix types (block banded matrices) against non-standard vectors (
38.
How chaotic is chaos? How some AI for Science / SciML overstates accuracy claims(stochasticlifestyle.com)
4 points by ChrisRackauckas 1y ago|0 comments
39.
by ChrisRackauckas 1y ago
Yes, should have a (2021) in the title.
40.
by ChrisRackauckas 1y ago
It's at least 5 at this point.
41.
by ChrisRackauckas 1y ago
Indeed, I also like modern Fortran.
42.
by ChrisRackauckas 1y ago
One of the reasons why I started using Julia was because the NumPy syntax was so difficult. Going from MATLAB to NumPy I felt like I suddenly became a mediocre programmer, spending less time on math and more time on "performance engine
43.
by ChrisRackauckas 1y ago
This changed with v1.0 with the introduction of literal_pow and the interprocedural optimizations necessary so that compile-time constants (literal numbers) could optimize this case and determine the alternative type. Yeah this was written
44.
Compilation of Julia code for deployment in model-based engineering(arxiv.org)
4 points by ChrisRackauckas 2y ago|0 comments
45.
by ChrisRackauckas 2y ago
I'm sending this to Michael: looks like he has a fan!!!
46.
by ChrisRackauckas 2y ago
The Julia tools can target embedded devices through LLVM if it has an LLVM compiler. This is one of the big reasons for the juliac ahead-of-time compilation functionalities recently added to Julia v1.12, it's for the JuliaSim/Mode
47.
by ChrisRackauckas 2y ago
The same team for interfaces was first doing the caching of binaries to solve time to first plot for v1.9, and then for v1.12 just finished up the binary trimming. Another big thing landing soon is improvements to escape analysis which shou
48.
by ChrisRackauckas 2y ago
v1.12 already has merged binary trimming, so we have cases with a full ODE solver that's not 10's of MiB (mostly linear algebra BLAS stuff left), and hello world is 900kB. See https://info.juliahub.com/blog/ne
49.
Julia Developer Experience with Tim Holy – Julia Dispatch Podcast [video](youtube.com)
3 points by ChrisRackauckas 2y ago|0 comments
50.
Static Compilation of Julia with Jeff Bezanson – Julia Dispatch Podcast [video](youtube.com)
7 points by ChrisRackauckas 2y ago|0 comments
51.
Add –trim option for generating smaller binaries(jbytecode.github.io)
3 points by ChrisRackauckas 2y ago|0 comments
52.
by ChrisRackauckas 2y ago
Did you link the wrong script? The script you show runs everything to statistical significance using Chairmarks.@b. Also, I don't understand what the issue would be with mixing Python and Julia code in the benchmark script. The Julia s
53.
by ChrisRackauckas 2y ago
There's also the new GC(s) that are almost ready to merge that have different properties that can be tuned.
54.
by ChrisRackauckas 2y ago
IO is thread-safe by default, but that does slow it down. There's a keyword argument to turn that off (if you know you're running it single threaded) and right now it's a rather large overhead. It needs some GC work IIRC to r
55.
by ChrisRackauckas 2y ago
The last updates deleted modules from the standard library like distutils. Any library that used those modules is now broken.
56.
by ChrisRackauckas 3y ago
It's varied, that's why it's measured as work-precision. It would be good to similarly produce a full work-precision line as point estimates do not tell the full story. Note that we are preparing an extension to easily use th
57.
by ChrisRackauckas 3y ago
The full DFN has a lot more going on. Use the benchmark-ified version from the benchmarks: https://docs.sciml.ai/SciMLBenchmarksOutput/dev/NonlinearPro... But as Avik states, this same kind of thing is done as par
58.
by ChrisRackauckas 3y ago
> It's very robust, matrix-free, and doesn't require an excessive amount of code to implement. Further, the method can be preconditioned with a positive definite preconditioner. It is not robust, in fact it's well-known as
59.
by ChrisRackauckas 3y ago
That makes your Newton method require calculating the Hessian of the objective function, which is a second derivative. A normal Newton method for nonlinear systems uses the Jacobian, which is the first derivative (the connection is that the
60.
by ChrisRackauckas 3y ago
Yes, especially Highs is completely unrelated because it's for (mixed integer) linear programming problems and quadratic programming problems, so it's not able to solve the types of problems described in this manuscript. IPOPT is

More ›