by Yubo "Paul" Yang

Automatic Differentiation

Automatic Differentiation exploits the fact that any algebraic function implemented on a computer can be compiled into a long list of elementary operations and elementary functions. Using this observation, exact differentiation can be carried out efficiently by exploiting the chain rule.

Presentation Summary

link to pdf

Examples

compute graph generation example
source code transformation (SCT) example

Active Researcher

Sebastian F. Walter

Package	Description	Last Commit
algopy	native python ad, work with numpy	Jun 30, 2015
pyadolc	wrapper for ADOL-C (C++)	Mar 30, 2016
pycppad	wrapper for CppAD	July 10, 2014

References

Relation to Apollo
Tutorial by Prof. Tamas Terlaky from McMaster Univeristy
Demonstration by Boyana Norris at Argonne with many applications
Dedicated topic website established by Argonne staff

All Numerical Methods

by Kiel Williams · Relaxation Methods in the Solution of Partial Differential Equations
by Brian Busemeyer · Simulated annealing
by Kiel Williams · Numerical ODE Solutions
by Dima Kochkov · Fast Fourier Transform
by Will Wheeler · Finding Eigenvalues: Arnoldi and QR
by Brian Busemeyer · Compressed sensing
by Yubo "Paul" Yang · Automatic Differentiation

Yubo "Paul" Yang 2016-06-01 ALGORITHM
automatic differentiation numerical method