v0.15.4

Convex v0.15.4

Diff since v0.15.3

Merged pull requests:

• Less piracy of hcat and vcat (#515) (@ericphanson)

Closed issues:

• Add excludes MOI and Convex tests with presolve (#503)
• Fixing type piracy is becoming more urgent (#512)
• Constraints Are Not Satisfied for a Simple Quad Form (#516)

v0.15.3

Convex v0.15.3

Diff since v0.15.2

Closed issues:

• Precompilation failure (#495)

Merged pull requests:

• CompatHelper: bump compat for LDLFactorizations to 0.10, (keep existing compat) (#496) (@github-actions[bot])
• replace randn(m,1) with randn(m) to be more Julia like (#498) (@alanedelman)
• add cartesian indexing (#500) (@tjdiamandis)
• Prep for v0.15.3 (#501) (@odow)

v0.15.2

Convex v0.15.2

Diff since v0.15.1

Closed issues:

• Differentiable convex optimisation problem (#430)
• Positive semidefinite constraint still gives small negative eigenvalues (#490)
• SDP problem 100x slower than matlab CVX (#492)

Merged pull requests:

• CompatHelper: bump compat for LDLFactorizations to 0.9, (keep existing compat) (#493) (@github-actions[bot])
• remove deprecated AbstractTrees calls (#494) (@ExpandingMan)

v0.15.1

Convex v0.15.1

Diff since v0.15.0

Closed issues:

• LASSO/Elastic Net example (#485)
• When I run EF optimisatios I get this error: Quadratic form only defined for Hermitian matrices (#487)
• Different result each time problem is solved (#488)

Merged pull requests:

• fix lasso_regression (example) (#486) (@PaulSoderlind)
• Change Dict -> OrderedDict (#489) (@baggepinnen)

v0.15.0

Convex v0.15.0

Diff since v0.14.18

Closed issues:

• Test with all capable solvers (#54)
• Unacceptable slow speed in model generation still exists (#137)
• Can't obtain dual variable (#158)
• Various operations don't support broadcast (#218)
• About positive definite matrix in variables? (#226)
• Warm Start example does not work as expected (#229)
• parameter change not detected (#234)
• Strange performance issue (#246)
• Cleanup code formatting (#251)
• Bug using maximum (#275)
• Too tight tolerance in one of the tests? (#295)
• v0.13 now as fast as R/CVXR, also scales linearly (v0.12 was quadratic) (#353)
• Element wise addition after multiplication (#355)
• Ideas for scaling up to large(r) subset selection problems? (#356)
• Not supporting normal distribution CDF function from Distributions.jl (#389)
• Too much memory consumption right before trying to solve a problem. (#390)
• Broadcasting issue with element-wise multiplication (Hadamard product) (#410)
• Example on warmstarting in the manual fails (#451)
• First-run after include(my_file.jl) is faster than later (#471)
• Quesitons about how Convex.jl works (#472)
• Fix the docs (#483)

Merged pull requests:

• move NEWS to docs (#470) (@ericphanson)
• [docs] Remove extra space in explanation (#473) (@ericphanson)
• doc: norm(x, p) (#474) (@PaulSoderlind)
• Update to MOI v1.0 (#476) (@odow)
• [docs] fix warmstart example (#478) (@odow)
• Add JuliaFormatter (#480) (@odow)
• Prep for v0.15 (#481) (@odow)
• Add release notes for v0.15 (#482) (@odow)
• Variety of fixes to the docs (#484) (@odow)

v0.14.18

Convex v0.14.18

Diff since v0.14.17

Closed issues:

• typo in logisticloss definition for scalar variables (#458)

Merged pull requests:

• fix #458 (#469) (@ericphanson)

v0.14.17

Convex v0.14.17

Diff since v0.14.16

Closed issues:

• Dev plan (#320)
• problem with Convex and new SCS (0.8.1) (#466)

Merged pull requests:

• Update README for maintenance mode (#465) (@odow)
• Update to MOI v0.10 (#467) (@ericphanson)
• Support both MOI v0.9 and MOI v0.10 (#468) (@ericphanson)

v0.14.16

Convex v0.14.16

Diff since v0.14.15

• Improve numerical stability when evaluating logsumexp (#457). Thanks @JinraeKim!

Closed issues:

• Consider (probably) numerically stable way of calculating logsumexp (#461)

Merged pull requests:

• Improve numerical stability of logsumexp (#462) (@JinraeKim)
• bump version and news (#463) (@ericphanson)

v0.14.15

Convex v0.14.15

Diff since v0.14.14

• Use sparse factorization for checking for positive semi-definiteness in quadform when possible (#457). Thanks @mtanneau!
• Add assume_psd=false argument to skip checking for positive semi-definiteness in quadform (#456).

Merged pull requests:

• add assume_psd option to quadform (#456) (@ericphanson)
• Use sparse symmetric factorization when possible (#457) (@mtanneau)
• bump version, NEWS (#459) (@ericphanson)

v0.14.14

Convex v0.14.14

Diff since v0.14.13

Closed issues:

• Value of eps() used in new implementation of quadform is too small (#452)

Merged pull requests:

• Increase value added to input matrix in quadform check (#453) (@numbermaniac)
• bump version + NEWS for v0.14.4 release (#455) (@ericphanson)