v0.15.0

LsqFit v0.15.0

Diff since v0.14.0

Merged pull requests:

• Use StatsAPI instead of StatsBase, define confint and vcov (#231) (@ararslan)
• Fix documentation deployment (#247) (@ararslan)

v0.14.0

LsqFit v0.14.0

Diff since v0.13.0

Merged pull requests:

• Clean up the whole project. (#221) (@pkofod)
• Fix the fact that dx is not copied in the trace logic. (#222) (@pkofod)
• Test that constant changes when bounding it. (#223) (@pkofod)
• Project in docs. (#224) (@pkofod)
• Convert g_norm to the type of the sum of squared residuals. (#225) (@pkofod)
• fix documenter (#226) (@pkofod)
• store_trace without show_trace; maxTime (#229) (@viktmar)
• Standardize GHA test setup, replace Travis badge with GHA (#232) (@ararslan)
• CompatHelper: bump compat for "StatsBase" to "0.34" (#237) (@github-actions[bot])
• LsqFit 0.13.1 (#244) (@IanButterworth)
• Update curve_fit.jl to not send big floats to beta_inc_inv (#245) (@pkofod)
• Update Project.toml to reflect what is tested (#246) (@pkofod)

Closed issues:

• Test for levenberg_marquardt.jl fails via DiffEqDiffTools not defined (#144)
• possible goodness of fit (#148)
• likelihood fitting (possible super-useful feature) ? (#153)
• How to use weights parameter (#158)
• lmfit returns optimum worse than starting point (#164)
• Taking too long to be imported (#169)
• Vector valued y (#171)
• Documentation issues (#184)
• Using curve_fit in bigfloat mode and kwarg g_tol (#191)
• trace not stored correctly in levenberg_marquardt.jl (#195)
• Fitting of 2D data is broken (#209)
• Release (#211)
• n_buffer has wrong dimensions (#212)
• stderror throws LinearAlgebra.LAPACKException(1) with weighted least squares fitting. (#234)
• Reqeust to register release (#243)

Release 0.13.0

What's Changed

• Added method example with wt by @KronosTheLate in #188
• Changed standard_error to stderror by @mo8it in #192
• Update README.md by @mo8it in #193
• Fix typo in curve_fit docstring by @johnomotani in #197
• ensure monotonous decay by @JakobAsslaender in #182
• Fix ambiguity errors for matrix-valued weights by @fzierler in #199
• Allow AbstractVector{T} for lower and upper instead of Vector{T} by @RafaelArutjunjan in #177

New Contributors

• @KronosTheLate made their first contribution in #188
• @mo8it made their first contribution in #192
• @johnomotani made their first contribution in #197
• @JakobAsslaender made their first contribution in #182
• @fzierler made their first contribution in #199

Full Changelog: v0.12.1...v0.13.0

v0.12.1

LsqFit v0.12.1

Diff since v0.12.0

Closed issues:

• How can the standard deviation of a parameter be NaN...? (#176)

Merged pull requests:

• Allow Vector{BigFloat} for parameters (#168) (@RafaelArutjunjan)
• CompatHelper: bump compat for "Distributions" to "0.25" (#185) (@github-actions[bot])

v0.12.0

LsqFit v0.12.0

Diff since v0.10.0

Closed issues:

• Interpreting the fit parameter (#146)
• LAPACKException(2) for stderror(fit) (#172)

Merged pull requests:

• CompatHelper: bump compat for "Distributions" to "0.23" (#150) (@github-actions[bot])
• Update Project.toml (#151) (@pkofod)
• CompatHelper: bump compat for "StatsBase" to "0.33" (#152) (@github-actions[bot])
• Fix argument order in example (#157) (@RafaelArutjunjan)
• CompatHelper: bump compat for "Distributions" to "0.24" (#165) (@github-actions[bot])
• Update TagBot.yml (#173) (@jebej)

v0.10.0

v0.10.0 (2020-01-09)

Diff since v0.9.0

Closed issues:

• LAPACKException with standard_errors (#131)
• Passing optional (constant) parameters (#130)

Merged pull requests:

• CompatHelper: bump compat for "Distributions" to "0.22" (#140) (github-actions[bot])
• Don't use Calculus.jl (#138) (pkofod)
• CompatHelper: add new compat entry for "OptimBase" at version "2.0" (#136) (github-actions[bot])
• CompatHelper: add new compat entry for "StatsBase" at version "0.32" (#135) (github-actions[bot])
• CompatHelper: add new compat entry for "NLSolversBase" at version "7.5" (#134) (github-actions[bot])
• CompatHelper: add new compat entry for "ForwardDiff" at version "0.10" (#133) (github-actions[bot])
• added basic logic to check for missing values (#132) (affans)

v0.9.0

v0.9.0 (2019-12-23)

Diff since v0.8.1

Closed issues:

• Optimizing multiple models in a loop returns wrong results (#125)
• Issue with :finiteforward (#124)
• Accessing Levenberg - Marquardt directly (#121)
• Move to project.toml (#119)
• Problem precompliling (#118)
• Error for arrays containing Inf or NaN (#109)
• Delayed gratification and geodesic acceleration in Levenburg-Marquardt (#88)
• Heuristics for initial choice of lambda (#34)

Merged pull requests:

• Add the latest docs badge to README.md (#128) (singularitti)
• add Distributions 0.21 to Project.toml (#126) (bjarthur)
• typo fix (#123) (tlienart)
• Implement suggested start dampening parameter as per IMM-REP-1999-05,… (#122) (pkofod)
• Add Project.toml (#120) (asinghvi17)
• Geomerged (#117) (pkofod)

Add input checks for entries that are not finite

v0.8.1

Check data for inf and nan (#116)

Add inplace residual and Jacobian interfaces

Add support for inplace jacobian and model for unidimensional input (…

…#102)

* add support for inplace jacobian

* make range calls compatible with nightly

* Added entry in readme about inplacejac. Removed superfluous option in one lmfit

* correction to f!_from_f

* update

* added support for inplacef

* update readme

-introduce the `inplacef` option
-generalize use of `@.` for jacobians
-introduce use of `@views` for slices

* correct bug

* update inplace tests

- removed callable type tests (obsolete)
- added more tests to cover all inplace cases

* unified inplacef and inplacejac, adapted tests

* Removed LOAD_PATH modification

* Update curve_fit.jl

* Remove unused f!_from_f

* Remove reference to univariate only in README.md.

* Remove Function annotations.

* Apply AbstractArray annotations

* Update curve_fit.jl

Bugfix for cache variables for value evaluations

Fix it for good this time. (#112)

* Fix it for good this time.

* Remove old functions.

* Remove commented at-shows.