Julia 1.0

This supports Julia 1.0 (and 0.7) and drops support for 0.6.

Fix 0.7 Deprecations

0.7 Deprecations (#13)

* Update Julia version

* Complex128 -> ComplexF64

* Fix deprecation warnings

* Make 0.7 changes compatible to 0.6

* Don't allow travis failure on nightly

* Require only 0.6

* Fix using Test

* use Compat in tests

* Compat.LinearAlgebra

* Fix

v0.3.1: Merge pull request #12 from JuliaAlgebra/fix-allocations

• Fixes accidental allocations in some cases.

v0.3.0

New

• This adds a new type System for polynomial systems.
• JacobianConfig and GradientConfig can now be constructed by config(F, x) where x is representative with which F should be evaluated.
• A new function evaluate_and_jacobian[!] to evaluate F and its Jacobian at the same time.

Breaking

• JacobianConfig, evaluate, evaluate!, jacobian, jacobian! now expects a System instead of Vector{<:Polynomial}

Performance Improvement

I removed bound checks in some internal methods. This resulted in a 20% performance improvement in my benchmarks. Enjoy :)

v0.2.1

9d0809f

New API

I refactored the API to make it similar to ForwardDiff.jl. It's probably best to simply check out the documentation for all the changes.

Some comments:
The most noticeable change is that now there is a Config object (GradientConfig and JacobianConfig) which holds a bunch of values which where previously held in the Polynomialstruct.
Furthermore, you can now evaluate gradients and jacobians directly (without the need of computing the derivatives). This also enables an optimisation where computations for evaluate and gradient (resp. jacobian) are shared. Especially for compute intensive types (like BigFloat or DoubleFloat64 from HigherPrecision.jl) this results in a significant performance boost.

v0.1.1

This release includes the following:

• A new type PolynomialEvalulationArray for the fast evaluation of systems of polynomials #4
• Eliminated a type instability in evaluate

v0.1.0

📦