AutoDiff.NET is a pure .NET library that allows a developer to easily compose functions symbolically and then automatically calculates the function's value and gradient at any given point. It can be very useful in conjunction with a gradients-based optimization library. It has been tested to work on Mono 2.10 on Linux and on .NET4 on Windows.
|Tags||Math Optimization .NET Framework 4.0|
|Operating Systems||Linux Windows (32 and 64bit)|
|Implementation||.Net 4 C#|
Release Notes: This release adds a substantial performance improvement.
Release Notes: The IntPower class has been renamed from ConstPower, breaking the API. A TermPower class was added. A division operator was added. User defined nary functions are now provided in addition to unary and binary functions. Evaluation of gradients using a subset of the variables was implemented. This is used for parameterized functions having "constant" parameters in addition to the actual function argument. Power terms can now have terms in the exponent, in addition to constants.
Release Notes: Differentiation bugfixes. Fixes a bug with the TermUtils.Differentiate method's contract failure. Fixes differentiation for functions that consist of simply a variable.
Release Notes: User-defined unary and binary functions.
Release Notes: Reverse-mode AD, samples, and bugfixes.