v0.8.0

• Upgraded dependencies
• Fixed corner cases where CG would not work on the GPU
• Modernized package manager and travis related things

v0.7.1

No feature updates -- just some polishing and updates for Julia 1.0.

v0.7.0: Upgrade to Julia 0.7

IterativeSolvers.jl now works on Julia 0.7 thanks to the work of @lostella.

v0.6.0: LOBPCG

IterativeSolvers.jl now ships with the LOBPCG eigensolver, contributed by @mohamed82008.

This release (v0.6.0) is the last release before upgrading to Julia 0.7.

v0.5.0

• BLAS-1 calls like axpy are replaced with broadcasting operations
• Randomized algorithms have moved to a separate package RandomizedLinAlg.jl
• ConvergenceHistory objects show a summary when printed
• Default maximum number of iterations is now the order of the provided matrix
• Plotting functions unrelated to PlotRecipes are removed
• The solve function for preconditioners is removed in favour of A_ldiv_B
• IDR(s) allocates less

v0.4.2

Minor release to improve documentation deployment.

v0.4.1: Fix element types (#163)

* Remove unused methods

* Clean up common.jl

* Ensure r has the type of x, not b. Remove b from the iterable since it is not used anyway

* Make residual type in BiCGStab(l) equal to solution type and remove rhs from iterable

* Chebyshev: residual type should equal x's type and remove b from iterable

* Similar story for GMRES

* Use x element type everywhere

* Use different types for solution, temporary and rhs in iterables of stationary methods

* zerox over zeros

GSoC 2017 update

• Improve speed and memory usage in GMRES, CG, Chebyshev iteration, the power method and stationary methods mentioned below
• Add missing iterative methods MINRES and BiCGStab(l)
• Implement stationary methods for sparse matrices #156
• Upgrade the package to Julia 0.6 and thereby updating the tests
• Improve and restructure the documentation

v0.3.1

Add julia0.6 to CI