Programmatic physics

[Philip Thrift]

On Wednesday, June 12, 2019 at 9:44:14 AM UTC-5, Lawrence Crowell wrote:

On Tuesday, June 11, 2019 at 2:09:41 PM UTC-5, Philip Thrift wrote:

The End of Theoretical Physics As We Know It

by Sabine Hossenfelder

https://www.quantamagazine.org/the-end-of-theoretical-physics-as-we-know-it-20180827/

Beyond Math 

by Sophia Magnusdottir (actuailly Sabine Hossenfelder)

https://fqxi.org/data/essay-contest-files/Magnusdottir_fqxi15.pdf

@philipthrift 

I worked out recently a MATHEMATICA solution to a differential equation. The amount of time it would have taken me to do this by analysis would have been 10s of times longer. We are in the age of numerical applications. I suppose I see nothing particularly wrong with that. In the theory of differential equations, since the time of Frobenius et al say a century ago progress has gone into a bit of a crawl. There have been developments with systems of differential forms and Pfaffians with nonlinear DEs, but this is formidable.

LC

Sabine's philosophical perspective expressed in the the two articles above align with my own.

I have looked at Wolfram Language, or "the Wolfram Language" (the language of Mathematica) [he should call the language "Wolfram" and be done with the "the" thing]:

   

Stephen Wolfram Blog: 

The Wolfram Function Repository: Launching an Open Platform for Extending the Wolfram Language

What the Wolfram Language Makes Possible

- https://blog.stephenwolfram.com/2019/06/the-wolfram-function-repository-launching-an-open-platform-for-extending-the-wolfram-language/

I don't know its future in physics. (It has odd syntax, for one thing, and there's its "openness" problem.)

There's lots of alternatives of course:

Programming for Physics and Astronomy

- http://prancer.physics.louisville.edu/astrowiki/index.php/Programming_for_Physics_and_Astronomy

Introducing TensorNetwork, an Open Source Library for Efficient Tensor Calculations

Tuesday, June 4, 2019

- https://ai.googleblog.com/2019/06/introducing-tensornetwork-open-source.html

DiffSharp: Differentiable Functional Programming

- http://diffsharp.github.io/DiffSharp/

...

What Sabine suggests though is more like the use of theorem proving assistants and languages in physics:

Workshop on Formal Reasoning about Physics

- http://hvg.ece.concordia.ca/events/formal-physics.htm

In the last two decades, formal verification and analysis of engineering systems have been widely considered in academia and industry. This results in the availability of more sophisticated proof assistants (e.g., HOL Light, Coq, Isabelle/HOL and Mizar) and large formalization of mathematical theories (e.g., multivariate analysis, linear algebra, etc.). Recently, significant amount of work has been reported describing the prospects of using artificial intelligence (AI) and machine learning techniques to provide better management of large formalized theories and powerful proof automation. In this workshop, we aim at discussing two main research directions: 1) motivation and state-of-the-art related to the formalization of Physics (particularly Optics). 2) Applying already developed AI techniques to the formalization of Physics. 

@philipthrift