v2.13.0

LinearOperators v2.13.0

Diff since v2.12.0

Merged pull requests:

• Improve inferrability (#383) (@timholy)
• Add ShiftedOperator for shifted linear operators (#394) (@farhadrclass)
• Patch bfgs upper bound (#399) (@MaxenceGollier)
• Fix allocation issue on julia pre CI (#402) (@MaxenceGollier)
• Fix inferrability of size(::AdjTrans) (#403) (@timholy)
• Skip conj.(v) for real-valued vectors (#404) (@timholy)
• Describe LinearOperator{T, S}(args...) in docs (#405) (@timholy)
• Use hasmethod for checking mul! implementation (#406) (@timholy)

Closed issues:

• Efficient get_nargs (#275)

v2.12.0

LinearOperators v2.12.0

Diff since v2.11.0

Merged pull requests:

• CompatHelper: bump compat for JLArrays in [weakdeps] to 0.3, (keep existing compat) (#379) (@github-actions[bot])
• Optimize computations in LBFGS (#381) (@MaxenceGollier)
• CompatHelper: bump compat for JLArrays in [extras] to 0.3, (keep existing compat) (#385) (@github-actions[bot])
• Update FreeBSD image family to 14-3 (#386) (@amontoison)
• Remove allocations in lsr1 push! (#388) (@MaxenceGollier)
• Update package compatibility versions in Project.toml (#389) (@amontoison)
• Update Breakage.yml (#390) (@amontoison)
• Remove @. macro from BFGS and SR1 computations (#392) (@MaxenceGollier)
• Add an upper bound of the operator norm for limited-memory qn (#395) (@MaxenceGollier)

Closed issues:

• Inverse LSR1Operator documentation (#367)
• Empty Tutorial Page (#376)
• Inefficient implementation of LBFGS multiply (#378)
• Allocations in LSR1 push! (#387)

v2.11.0

LinearOperators v2.11.0

Diff since v2.10.0

Merged pull requests:

• fix normest not working on rectangular matrices (#364) (@njericha)
• Add correct storage_type for lazy wrapped GPU arrays (adjoint, transpose, Diagonal) (#365) (@nHackel)
• CompatHelper: bump compat for AMDGPU in [weakdeps] to 2, (keep existing compat) (#374) (@github-actions[bot])

Closed issues:

• normest broken for non-square matrices (#362)

v2.10.0

LinearOperators v2.10.0

Diff since v2.9.0

Merged pull requests:

• Add storage_type for Sparse CuArrays (#333) (@nHackel)
• Eliminate useless allocation for diagonal quasi-Newton operators (#337) (@MohamedLaghdafHABIBOULLAH)
• Breakage: use upload-artifact v4 (#340) (@dpo)
• build docs with julia 1.10 (#345) (@dpo)
• upgrade actions (#347) (@dpo)
• breakage: minor cleanup (#348) (@dpo)
• breakage: compute dependencies (#349) (@dpo)
• Add JLArrays, Metal and AMDGPU extensions and S kwarg tests (rebased) (#351) (@dpo)
• Update with JSOBestieTemplate for LinearOperators (#353) (@tmigot)
• Dbfgs update (#354) (@MohamedLaghdafHABIBOULLAH)
• use LTS 1.10 in CI (#355) (@dpo)
• Disable symtridiagonal s-kwarg test (#356) (@nHackel)
• CompatHelper: bump compat for AMDGPU in [weakdeps] to 1, (keep existing compat) (#359) (@github-actions[bot])
• Update with JSOBestieTemplate for LinearOperators (#360) (@tmigot)
• Use Type instead of DataType (#361) (@amontoison)
• CompatHelper: bump compat for JLArrays in [weakdeps] to 0.2, (keep existing compat) (#363) (@github-actions[bot])

Closed issues:

• Some operations do not work although they (probably) should (#153)
• In-place solve for inverse operators (#184)
• Upgrade to v2.0 (#189)
• Verify the LBFGSOperator update (push!) (#223)
• Test for unbound parameters (#242)

v2.9.0

LinearOperators v2.9.0

Diff since v2.8.0

Merged pull requests:

• preallocate storage for LSR1 operator (#335) (@dpo)
• solve_shifted_system! method for LBFGS solving step in recursive way (#338) (@farhadrclass)

v2.8.0

LinearOperators v2.8.0

Diff since v2.7.0

Merged pull requests:

• Review julia compatibility in readme #313 (#320) (@hpmouton)
• Add LinearOperator constructor for CUDA (#321) (@nHackel)
• Add S kwarg to opRestriction (#324) (@nHackel)
• Change LDLFactorizations to be a weakdep (#326) (@tmigot)
• Add mul! for matrices (#330) (@gdalle)

Closed issues:

• Change LDLFactorizations to be a weakdep, i.e., and extension (#295)
• ERROR: LinearOperatorException("storage types cannot be promoted to a concrete type") (#307)
• Review julia compatibility in readme (#313)
• Version 2.7.0 is breaking (though the version might indicate it is not) – DiagonalQN used in other packages. (#319)
• Missing mul!(::Matrix, ::LinearOperator, ::Matrix) (#322)
• Missing S keyword for opRestriction (#323)

v2.7.0

LinearOperators v2.7.0

Diff since v2.6.0

Merged pull requests:

• Adjoint opEye (#309) (@geoffroyleconte)
• Use CITATION.cff (#310) (@tmigot)
• Fix documentation build (#311) (@tmigot)
• Fix Aqua issue (#312) (@geoffroyleconte)
• Separate diagonal quasi-Newton types (#316) (@dpo)
• update CirrusCI config (#317) (@dpo)
• bump version to 2.7.0 (#318) (@dpo)

Closed issues:

• Separate updates for Diagonal QN operators (#290)
• Warning during the precompilation (#291)
• Declare missing compat entries (#299)
• Bug: adjoint(opEye()) * randn(10) (#308)
• ChainRulesCore error (#315)

v2.6.0

LinearOperators v2.6.0

Diff since v2.5.2

Merged pull requests:

• URL link for CITATION.bib (#274) (@paraynaud)
• Implement chainrules (#279) (@migrosser)
• CompatHelper: add new compat entry for Requires at version 1, (keep existing compat) (#281) (@github-actions[bot])
• Remove Aqua stale deps check (#288) (@geoffroyleconte)
• [documentation] Update Documenter.jl (#292) (@amontoison)
• [CI] Remove the build for ARMv8 (#293) (@amontoison)
• [breaking] Remove shape function (duplicate of size) (#297) (@JakobAsslaender)
• [JSOTemplate.jl] Template compliance update (#300) (@abelsiqueira)
• bump version to 2.6.0 (#301) (@dpo)
• use exclamation mark for increment operations (#303) (@dpo)
• Fix a bug in the derived adjoint (#304) (@JakobAsslaender)

Closed issues:

• Chainrules (#278)
• Remove shape function (duplicate of size) (#298)

v2.5.2

LinearOperators v2.5.2

Diff since v2.5.1

Closed issues:

• Link to tutorial is broken (#269)

Merged pull requests:

• Fix link in tutorial (#270) (@tmigot)
• Improve DiagonalQN stability when s is close to zero (#271) (@geoffroyleconte)
• Release 2.5.2 (#273) (@geoffroyleconte)

v2.5.1

LinearOperators v2.5.1

Diff since v2.5.0

Merged pull requests:

• Add reset function for Diagonal QN operators, fix bug SpectralGradient (#266) (@geoffroyleconte)
• Release 2.5.1 (#268) (@geoffroyleconte)