Direct Automatic Differentiation of (Differential Equation) Solvers vs Analytical Adjoints: Which is Better?
October 11 2022 in Differential Equations, Julia, Mathematics, Science, Scientific ML | Tags: automatic differentiation, differentiable programming, sciml | Author: Christopher Rackauckas
Automatic differentiation of a “solver” is a subject with many details for doing it in the most effective form. For this reason, there are a lot of talks and courses that go into lots of depth on the topic. I recently gave a talk on some of the latest stuff in differentiable simulation with the American Statistical Association, and have some detailed notes on such adjoint derivations as part of the 18.337 Parallel Computing and Scientific Machine Learning graduate course at MIT. And there are entire organizations like my SciML Open Source Software Organization which work day-in and day-out on the development of new differentiable solvers.
I’ll give a brief summary of all my materials here below.
Continuous vs Discrete Differentiation of Solvers
AD of a solver can be done in essentially two different ways: either directly performing automatic … READ MORE
Engineering Trade-Offs in Automatic Differentiation: from TensorFlow and PyTorch to Jax and Julia
December 25 2021 in Julia, Programming, Science, Scientific ML | Tags: automatic differentiation, compilers, differentiable programming, jax, julia, machine learning, pytorch, tensorflow, XLA | Author: Christopher Rackauckas
To understand the differences between automatic differentiation libraries, let’s talk about the engineering trade-offs that were made. I would personally say that none of these libraries are “better” than another, they simply all make engineering trade-offs based on the domains and use cases they were aiming to satisfy. The easiest way to describe these trade-offs is to follow the evolution and see how each new library tweaked the trade-offs made of the previous.
Early TensorFlow used a graph building system, i.e. it required users to essentially define variables in a specific graph language separate from the host language. You had to define “TensorFlow variables” and “TensorFlow ops”, and the AD would then be performed on this static graph. Control flow constructs were limited to the constructs that could be represented statically. For example, an `ifelse` function statement is very different from … READ MORE
Useful Algorithms That Are Not Optimized By Jax, PyTorch, or Tensorflow
July 21 2021 in Julia, Programming, Science, Scientific ML | Tags: automatic differentiation, differentiable programming, julia, machine learning | Author: Christopher Rackauckas
In some previous blog posts we described in details how one can generalize automatic differentiation to give automatically stability enhancements and all sorts of other niceties by incorporating graph transformations into code generation. However, one of the things which we didn’t go into too much is the limitation of these types of algorithms. This limitation is what we have termed “quasi-static” which is the property that an algorithm can be reinterpreted as some static algorithm. It turns out that for very fundamental reasons, this is the same limitation that some major machine learning frameworks impose on the code that they can fully optimize, such as Jax or Tensorflow. This led us to the question: are there algorithms which are not optimizable within this mindset, and why? The answer is now published at ICML 2021, so lets dig into … READ MORE
Generalizing Automatic Differentiation to Automatic Sparsity, Uncertainty, Stability, and Parallelism
March 10 2021 in Differential Equations, Julia, Mathematics, Programming, Science, Scientific ML | Tags: abstract interpretation, automatic differentiation, non-standard interpretation, Pantelides algorithm | Author: Christopher Rackauckas
Automatic differentiation is a “compiler trick” whereby a code that calculates f(x) is transformed into a code that calculates f'(x). This trick and its two forms, forward and reverse mode automatic differentiation, have become the pervasive backbone behind all of the machine learning libraries. If you ask what PyTorch or Flux.jl is doing that’s special, the answer is really that it’s doing automatic differentiation over some functions.
What I want to dig into in this blog post is a simple question: what is the trick behind automatic differentiation, why is it always differentiation, and are there other mathematical problems we can be focusing this trick towards? While very technical discussions on this can be found in our recent paper titled “ModelingToolkit: A Composable Graph Transformation System For Equation-Based Modeling” and descriptions of methods like intrusive uncertainty quantification, I want … READ MORE