Add an example of column generation #2006

dourouc05 · 2019-07-09T20:16:56Z

Follows #2004

codecov · 2019-07-09T20:44:24Z

Codecov Report

Merging #2006 into master will decrease coverage by 0.03%.
The diff coverage is n/a.

@@            Coverage Diff            @@
##           master   #2006      +/-   ##
=========================================
- Coverage   89.13%   89.1%   -0.04%     
=========================================
  Files          33      33              
  Lines        4324    4312      -12     
=========================================
- Hits         3854    3842      -12     
  Misses        470     470

Impacted Files	Coverage Δ
src/aff_expr.jl	`85.82% <0%> (-0.75%)`	⬇️
src/quad_expr.jl	`89.03% <0%> (-0.65%)`	⬇️
src/objective.jl	`95.23% <0%> (+4.32%)`	⬆️

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update 010fccf...4c19ef8. Read the comment docs.

odow

Please follow the style of the other examples: place everything inside a function, and add an informative docstring.

dourouc05 · 2019-07-09T23:59:39Z

Here is an update example, with a lot more description (the problem to solve, a general description of the technique), but I did not write a column-generation tutorial. However, I did not change the dividing in functions, as there is just a main function and the pricing problem: merging the two would make things less clear, I fear.

matbesancon · 2019-07-10T16:52:56Z

@dourouc05 cool that you're using it. Instead of a repo I will archive both the post and code in a reproducible state (with Manifest and all) to cite something stable

blegat · 2019-07-10T19:46:36Z

examples/cutting_stock_column_generation.jl

+
+using JuMP, GLPK
+using SparseArrays
+const MOI = JuMP.MathOptInterface


This is already exported by JuMP

matbesancon · 2019-07-11T10:08:02Z

@dourouc05 you can use 10.5281/zenodo.3329388 as DOI or:

@misc{mathieu_besancon_2019_3329389,
  author       = {Mathieu Besançon},
  title        = {{A column generation example in Julia and JuMP}},
  month        = jul,
  year         = 2019,
  doi          = {10.5281/zenodo.3329388},
  url          = {https://doi.org/10.5281/zenodo.3329388}
}

dourouc05 · 2019-07-11T12:24:28Z

I did not add the full BibTeX entry, as it is easy enough to find from Zenodo. Is that good enough for you?

matbesancon · 2019-07-11T13:34:26Z

Yup all good thanks

examples/cutting_stock_column_generation.jl

mlubin · 2019-07-11T13:52:28Z

examples/cutting_stock_column_generation.jl

+        new_pattern = try
+            solve_pricing(dual.(demand_satisfaction), maxwidth, widths, rollcost, demand, prices)
+        catch
+            # At the final iteration, GLPK has dual values, but at least one of them is 0.0, and thus GLPK crashes.


Why does a dual value of 0.0 cause GLPK to crash? That shouldn't happen.

Actually, it's more GPLK.jl that gives an error (Cbc.jl too):

ERROR: LoadError: AssertionError: dual >= 0.0 Stacktrace: [1] __assert_dual_sense__ at C:\Users\…\.julia\packages\LinQuadOptInterface\ZMx9f\src\solve.jl:210 [inlined] [2] get at C:\Users\…\.julia\packages\LinQuadOptInterface\ZMx9f\src\solve.jl:217 [inlined] [3] get(::MathOptInterface.Bridges.LazyBridgeOptimizer{GLPK.Optimizer,MathOptInterface.Utilities.UniversalFallback{MathOptInterface.Bridges.AllBridgedConstraints{Float64}}}, ::MathOptInterface.ConstraintDual, ::MathOptInterface.ConstraintIndex{MathOptInterface.ScalarAffineFunction{Float64},MathOptInterface.GreaterThan{Float64}}) at C:\Users\…\.julia\packages\MathOptInterface\C3lip\src\Bridges\bridgeoptimizer.jl:236 [4] get(::MathOptInterface.Utilities.CachingOptimizer{MathOptInterface.AbstractOptimizer,MathOptInterface.Utilities.UniversalFallback{JuMP._MOIModel{Float64}}}, ::MathOptInterface.ConstraintDual, ::MathOptInterface.ConstraintIndex{MathOptInterface.ScalarAffineFunction{Float64},MathOptInterface.GreaterThan{Float64}}) at C:\Users\…\.julia\packages\MathOptInterface\C3lip\src\Utilities\cachingoptimizer.jl:453 [5] _moi_get_result(::MathOptInterface.Utilities.CachingOptimizer{MathOptInterface.AbstractOptimizer,MathOptInterface.Utilities.UniversalFallback{JuMP._MOIModel{Float64}}}, ::MathOptInterface.ConstraintDual, ::Vararg{Any,N} where N) at C:\Users\…\.julia\dev\JuMP\src\JuMP.jl:618 [6] get(::Model, ::MathOptInterface.ConstraintDual, ::ConstraintRef{Model,MathOptInterface.ConstraintIndex{MathOptInterface.ScalarAffineFunction{Float64},MathOptInterface.GreaterThan{Float64}},ScalarShape}) at C:\Users\…\.julia\dev\JuMP\src\JuMP.jl:649 [7] _constraint_dual(::ConstraintRef{Model,MathOptInterface.ConstraintIndex{MathOptInterface.ScalarAffineFunction{Float64},MathOptInterface.GreaterThan{Float64}},ScalarShape}) at C:\Users\…\.julia\dev\JuMP\src\constraints.jl:545 [8] dual(::ConstraintRef{Model,MathOptInterface.ConstraintIndex{MathOptInterface.ScalarAffineFunction{Float64},MathOptInterface.GreaterThan{Float64}},ScalarShape}) at C:\Users\…\.julia\dev\JuMP\src\constraints.jl:538 [9] _broadcast_getindex_evalf at .\broadcast.jl:625 [inlined] [10] _broadcast_getindex at .\broadcast.jl:598 [inlined] [11] getindex at .\broadcast.jl:558 [inlined] [12] copyto_nonleaf!(::Array{Float64,1}, ::Base.Broadcast.Broadcasted{Base.Broadcast.DefaultArrayStyle{1},Tuple{Base.OneTo{Int64}},typeof(dual),Tuple{Base.Broadcast.Extruded{Array{ConstraintRef{Model,C,Shape} where Shape<:AbstractShape where C,1},Tuple{Bool},Tuple{Int64}}}}, ::Base.OneTo{Int64}, ::Int64, ::Int64) at .\broadcast.jl:982 [13] copy at .\broadcast.jl:836 [inlined] [14] materialize at .\broadcast.jl:798 [inlined] [15] #example_cutting_stock#3(::Int64, ::typeof(example_cutting_stock)) at C:\Users\…\Desktop\a.jl:126 [16] example_cutting_stock() at C:\Users\…\Desktop\a.jl:88 [17] top-level scope at C:\Users\…\Desktop\a.jl:175

* Add optional kwargs to optimize! for optimize_hook * Add test for unexpected kwarg error * Add documentation on kwargs to optimize! * Formatting tweak

Fixes jump-dev#1996

…variables.

…enough…

dourouc05 · 2019-07-18T16:38:08Z

I'm messing up with the rebase, I'll open a new PR.

jump-dev#2006

* Restore the old PR. #2006 * Use the new function set_objective_coefficient. * Add the standard license header. * Remove the workaround for current GLPK.jl. * Update to JuMP 0.20.

odow reviewed Jul 9, 2019

View reviewed changes

blegat reviewed Jul 10, 2019

View reviewed changes

mlubin reviewed Jul 11, 2019

View reviewed changes

blegat and others added 15 commits July 18, 2019 18:24

[ci skip] Add CDCS

9042390

[ci skip] Update link

502d5a9

Remove blank

6b841a2

Add optional kwargs to optimize! for optimize_hook (jump-dev#1987)

6f4db09

* Add optional kwargs to optimize! for optimize_hook * Add test for unexpected kwarg error * Add documentation on kwargs to optimize! * Formatting tweak

Refactor matmul code with mul!

d9e8681

adds -> sets

3d9c53b

Fix variable names

698ad38

TrAdj -> TransposeOrAdjoint

851a3e2

a -> are

14e2463

Add tests for jump-dev#1990

2000cc1

Better error message for solvers that don't support NLP (jump-dev#1997)

9e81e32

Fixes jump-dev#1996

Implement algebra with UniformScaling

7891141

Fixes

f3cea0f

Improve NLP documentation about ForwardDiff (jump-dev#2000)

ed8444c

Rename m -> model in test_expression (jump-dev#2001)

299375a

dourouc05 force-pushed the dourouc05/example-colgen branch from 9b23c69 to 790da29 Compare July 18, 2019 16:24

dourouc05 added 5 commits July 18, 2019 18:32

Basic port of @matbesancon code. Not tested yet.

77b008c

Rewrite the example to have a master model that gets iteratively new …

1ccdf68

…variables.

Improve documentation, as asked by @odow.

8efd92e

This is what happens when you don't restart your Julia session often …

bc1eb51

…enough…

Removed MOI import, as asked by @blegat; rewrote the example as a test.

4554d00

Replace references to @matbesancon with DOI and Zenodo.

4c19ef8

dourouc05 force-pushed the dourouc05/example-colgen branch from 790da29 to 4c19ef8 Compare July 18, 2019 16:33

dourouc05 closed this Jul 18, 2019

dourouc05 added a commit to dourouc05/JuMP.jl that referenced this pull request Jul 18, 2019

Restore the old PR.

57a022b

jump-dev#2006

dourouc05 mentioned this pull request Jul 18, 2019

Add an example of column generation #2010

Merged

dourouc05 added a commit to dourouc05/JuMP.jl that referenced this pull request Sep 7, 2019

Restore the old PR.

ea2d733

jump-dev#2006

dourouc05 added a commit to dourouc05/JuMP.jl that referenced this pull request Sep 7, 2019

Restore the old PR.

a96cfb0

jump-dev#2006

dourouc05 deleted the dourouc05/example-colgen branch September 7, 2019 01:58

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Add an example of column generation #2006

Add an example of column generation #2006

dourouc05 commented Jul 9, 2019

codecov bot commented Jul 9, 2019 •

edited

Loading

odow left a comment

dourouc05 commented Jul 9, 2019

matbesancon commented Jul 10, 2019

blegat Jul 10, 2019

matbesancon commented Jul 11, 2019

dourouc05 commented Jul 11, 2019

matbesancon commented Jul 11, 2019

mlubin Jul 11, 2019

dourouc05 Jul 11, 2019

dourouc05 commented Jul 18, 2019

Add an example of column generation #2006

Add an example of column generation #2006

Conversation

dourouc05 commented Jul 9, 2019

codecov bot commented Jul 9, 2019 • edited Loading

Codecov Report

odow left a comment

Choose a reason for hiding this comment

dourouc05 commented Jul 9, 2019

matbesancon commented Jul 10, 2019

blegat Jul 10, 2019

Choose a reason for hiding this comment

matbesancon commented Jul 11, 2019

dourouc05 commented Jul 11, 2019

matbesancon commented Jul 11, 2019

mlubin Jul 11, 2019

Choose a reason for hiding this comment

dourouc05 Jul 11, 2019

Choose a reason for hiding this comment

dourouc05 commented Jul 18, 2019

codecov bot commented Jul 9, 2019 •

edited

Loading