Differentiable Programming from Scratch

Historical fact: Differentiable programming was a little known secret back in the 90's, used mainly by engineers simulating numerically stiff systems like nukes and chemicals in FORTRAN 95. It then disappeared for nearly 30 yrs before rediscovery by the ML/AI researchers!

My PhD dissertation included a chapter (originally from a 2006 paper) on generic programming in CAS' on the algorithmic differentiable ring (and operator). By the 1990s, algorithmic differentiation was easily 30–40 years old. Griewank & Monagan both knew guys who had built early electromagnetic naval targeting computers that used the methodology back by at least the early 60s "by hand". (Very literally.)

I watched the ML/AI bros actively ignore previous research — even when they were requested to properly cite sources they were plagiarizing — in real time. The race to publish (even for big journals) was so important that it was easier to ignore the rank dishonesty than it was to correct their misbehavior. I'm 1000x happier to not have stayed around for all that crap.

> on the algorithmic differentiable ring (and operator)

That sounds like an interesting read. Do you have the chapter or the reference to the paper that you can share ?

Regarding th crop of deep neural network research their self-serving and willful blindness has a reputation that's well deserved.

A grad student from Hinton's lab mentioned one researcher who would misspell a citation on purpose so that the citation count of the cited paper does not go up.

My PhD, and its chapters are junk. Just read Griewank; I don't have anything more to say than his monograph. (We could never generalize the AD operator to the hybrid mode, so it's useless for real workloads.)

> My PhD, and its chapters are junk.

I think most people think so about their own dissertation. I have Griewank.