ItPropFit.jl v0.2.0: Multidimensional array margins (#7)

Big overhaul of the API, but maintaining some of the ease-of-use / qol interface of v0.1.0.

Changelog:

* Add vararg constructor for ArrayFactors

* add convenient method for ipf without seed matrix

as per Agresti, sec. 9.7.2

* Start multidimensional

* progress on multidimensional margins

* Create arraydims and arraymargins structs

* update for arraymargins & rename arraydims

- ArrayDims are now DimIndex (should maybe be DimIndices?)

* arrayfactors for multidim margins

* Add multidim sorting to ArrayFactors

* Refactor DimIndex -> DimIndices

* make arraymargins sorted constructor

* Update DimIndices

- add uniqueness check
- add docs
- update issorted method

* Clean up array margins

- Fix reference to DimIndices in docs
- Add size method
- Remove buggy sorting stuff

* update ArrayFactors

- Find failure
- remove weird sorting code

* Fix ArrayFactors Array method

- Now works with arbitrary DimIndices
- Fixed the size method
- Also fixed the size method for ArrayMargins

* small doc fix arrayfactors

* Make DimIndices consistent nested int vector

* Update tests

- add DimIndices tests
- add ArrayMargins tests
- overhaul ArrayFactors tests

* add todo in arrayfactors

* ipf_mult deals with ArrayMargins and DimIndices

- also add getindex method for dimindices

* Small fix for multidimensional arraymargins

* start update for multidimensional ipf

* first working version of multidimensional ipf

- also update to julia 1.8.0

* Add dict methods for AF & AM

preparation for #6

* Remove DimIndices convenience methods

* Merge api for nultidimensional with ipf

- remove original ipf
- add convenience interface for new ipf
- adjust tests
- move consistency tests to ArrayMargins

* update new interface

- Add consistency checks to test
- Add margin transform to export
- Fix small bug with inconsistent margin info message

* Update docs manifest

* Update benchmarks.md

The new algorithm is quite a bit faster!

* Update documentation, bump version

* Add constructor for mixed-eltype ArrayFactors

* change devbranch in documentation to "dev"

* Clean up constructors for types

- Move getdims to default_dimindices in DimIndices file
- Adjust tests for new constructor setup

* Small fixes for constructors in AF & AM docs

* final small doc update

* Disable info logging in tests

For cleaner test output

* Remove logging disable again

* Update README.md

Author: vankesteren

Manifest.toml

@@ -1,6 +1,7 @@
 # This file is machine-generated - editing it directly is not advised
 
-julia_version = "1.7.2"
+julia_version = "1.8.0"
 manifest_format = "2.0"
+project_hash = "f6dd4230b36847a6d40a3eea7d2682365388770e"
 
 [deps]

Project.toml

@@ -1,12 +1,12 @@
 name = "ItPropFit"
 uuid = "7a994e6b-35c0-41b3-a964-b7f62fc12992"
 authors = ["Erik-Jan van Kesteren <[email protected]>"]
-version = "0.1.0"
+version = "0.2.0"
 
 [deps]
 
 [compat]
-julia = "1.7"
+julia = "1.8"
 
 [extras]
 Test = "8dfed614-e22c-5e08-85e1-65c5234f0b40"

README.md

@@ -10,6 +10,8 @@ Multidimensional iterative proportional fitting in Julia.
 See the full documentation and getting started [here](https://vankesteren.github.io/ItPropFit.jl/).
 
 ```julia
+using ItPropFit
+
 # matrix to be adjusted
 X = [40 30 20 10; 35 50 100 75; 30 80 70 120; 20 30 40 50]
 
@@ -21,9 +23,10 @@ v = [200, 300, 400, 100]
 fac = ipf(X, [u, v])
 ```
 ```
-Factors for array of size (4, 4):
-  1: [0.9986403503185242, 0.8833622306385376, 1.1698911437112522, 0.8895042701910321]
-  2: [1.616160156063788, 1.5431801747375655, 1.771623700829941, 0.38299396265192226]
+[ Info: Converged in 8 iterations.
+Factors for 2D array:
+  [1]: [0.9986403503185242, 0.8833622306385376, 1.1698911437112522, 0.8895042701910321]
+  [2]: [1.616160156063788, 1.5431801747375655, 1.771623700829941, 0.38299396265192226]
 ```
 ```julia
 # compute adjusted matrix
@@ -38,10 +41,10 @@ Z = Array(fac) .* X
 ```
 ```julia
 # check that the margins are indeed [u, v]
-margins(Z)
+ArrayMargins(Z)
 ```
 ```
-2-element Vector{Vector{Float64}}:
- [150.0000000009452, 299.99999999962523, 399.99999999949796, 149.99999999993148]
- [200.0, 299.99999999999994, 399.99999999999994, 99.99999999999997]
+Margins of 2D array:
+  [1]: [150.0000000009452, 299.99999999962523, 399.99999999949796, 149.99999999993148]
+  [2]: [200.0, 299.99999999999994, 399.99999999999994, 99.99999999999997]
 ```