Add Div, Grad, Curl, Lap operators (autodiff #7)

[sritchie]

• Add Div, Grad, Curl, Lap operators (autodiff #7) #224 adds new Div, Grad, Curl and Lap operators in
  sicmutils.calculus.derivative and installs them into sicmutils.env. Add Div, Grad, Curl, Lap operators (autodiff #7) #224
  also removes the g/transpose implementation for Operator instances, and
  exposes sicmutils.calculus.derivative/taylor-series to sicmutils.env.

I THINK that Grad and div are related through a one-level-deep transpose only: https://en.wikipedia.org/wiki/Gradient not anything deeper. @littleredcomputer , lmk if you think this impl is correct.

[codecov-io]

Codecov Report

❗ No coverage uploaded for pull request base (sritchie/diff_six@92e829c). Click here to learn what that means.
The diff coverage is n/a.

@@                 Coverage Diff                  @@
##             sritchie/diff_six     #224   +/-   ##
====================================================
  Coverage                     ?   84.33%           
====================================================
  Files                        ?       81           
  Lines                        ?     8336           
  Branches                     ?      449           
====================================================
  Hits                         ?     7030           
  Misses                       ?      857           
  Partials                     ?      449           

---

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update 92e829c...128896d. Read the comment docs.

[littleredcomputer]

I wonder if these definitions are FDG-compliant, in the following sense: you make Grad by taking d and then doing a "static cast" of the variance. But the contravariant components of the gradient are equal to the covariant components only when the ambient metric is the identity matrix. Before FDG, SICMutils has no concept of metric, so there's no way to lift the index except to just patch it in as you do here. I think this loses one of the very important features of SICMutils in that it respects that fact that when f is a scalar function df is a one-form.

[littleredcomputer]

But there's a lot to be said for being able to use this system to do your Calculus 3 homework.