Bump Polyhedra and JuMP #1668

mforets · 2019-09-02T11:25:09Z

Closes #1666.

schillic · 2019-09-02T13:59:27Z

This needs to be resolved together with #1666 because the JuMP version is exclusive at the moment.

schillic · 2019-09-03T07:29:17Z

MethodError: Cannot convert an object of type Symbol to an object of type GLPK.MethodEnum

It seems we (= you 😉) need to do some changes in the interface.

mforets · 2019-09-04T20:17:54Z

The problem can be traced to:

julia> convex_hull([[1//1, 3//1, 5//1], [0//1, 3//1, 5//1]])

ERROR: MethodError: Cannot `convert` an object of type Symbol to an object of type GLPK.MethodEnum
Closest candidates are:
  convert(::Type{T}, ::JuMP.GenericAffExpr{T,VarType} where VarType) where T at C:\Users\marce\.julia\packages\JuMP\iGamg\src\aff_expr.jl:298
  convert(::Type{T}, ::JuMP.GenericQuadExpr{T,VarType} where VarType) where T at C:\Users\marce\.julia\packages\JuMP\iGamg\src\quad_expr.jl:323

and then in the remove_redundant_vertices function (https://github.com/JuliaReach/LazySets.jl/blob/master/src/Sets/VPolytope.jl#L422)

julia> vQ
V-representation Polyhedra.Hull{Rational{Int64},Array{Rational{Int64},1},Int64}:
2-element iterator of Array{Rational{Int64},1}:
 Rational{Int64}[1//1, 3//1, 5//1]
 Rational{Int64}[0//1, 3//1, 5//1]

julia> solver
JuMP.OptimizerFactory(GLPK.Optimizer, (), Base.Iterators.Pairs(:method=>:Exact))

julia> Polyhedra.removevredundancy(vQ, solver)
ERROR: MethodError: Cannot `convert` an object of type Symbol to an object of type GLPK.MethodEnum
Closest candidates are:
  convert(::Type{T}, ::JuMP.GenericAffExpr{T,VarType} where VarType) where T at C:\Users\marce\.julia\packages\JuMP\iGamg\src\aff_expr.jl:298
  convert(::Type{T}, ::JuMP.GenericQuadExpr{T,VarType} where VarType) where T at C:\Users\marce\.julia\packages\JuMP\iGamg\src\quad_expr.jl:323
  convert(::Type{T}, ::T) where T at essentials.jl:154

I dont know yet for sure but it seems like this should be reported in Polyhedra.

cc: @blegat

blegat · 2019-09-05T06:56:44Z

Replace :Exact by GLPK.EXACT.

mforets · 2019-09-05T19:27:26Z

thanks for the feedback!

i get another error now,

julia>  convex_hull([[1//1, 3//1, 5//1], [0//1, 3//1, 5//1]])
ERROR: GLPK.GLPKError("no bf solution found (use GLPK.factorize)")
Stacktrace:
 [1] _bf_exists at C:\Users\marce\.julia\packages\GLPK\nTwL7\src\GLPK_checks.jl:271 [inlined]
 [2] macro expansion at C:\Users\marce\.julia\packages\GLPK\nTwL7\src\GLPK_checks.jl:53 [inlined]
 [3] get_bhead(::GLPK.Prob, ::Int64) at C:\Users\marce\.julia\packages\GLPK\nTwL7\src\GLPK.jl:1619
 [4] get_infeasibility_ray(::GLPK.Optimizer, ::Array{Float64,1}) at C:\Users\marce\.julia\packages\GLPK\nTwL7\src\infeasibility_certificates.jl:49
 [5] optimize!(::GLPK.Optimizer) at C:\Users\marce\.julia\packages\GLPK\nTwL7\src\MOI_wrapper.jl:1328
 [6] optimize!(::MathOptInterface.Bridges.LazyBridgeOptimizer{GLPK.Optimizer}) at C:\Users\marce\.julia\packages\MathOptInterface\016Ol\src\Bridges\bridge_optimizer.jl:139
 [7] optimize!(::MathOptInterface.Utilities.CachingOptimizer{MathOptInterface.AbstractOptimizer,MathOptInterface.Utilities.UniversalFallback{MathOptInterface.Utilities.Model{Float64}}}) at C:\Users\marce\.julia\packages\MathOptInterface\016Ol\src\Utilities\cachingoptimizer.jl:178
 [8] #optimize!#78(::Bool, ::Bool, ::Base.Iterators.Pairs{Union{},Union{},Tuple{},NamedTuple{(),Tuple{}}}, ::Function, ::JuMP.Model, ::Nothing) at C:\Users\marce\.julia\packages\JuMP\iGamg\src\optimizer_interface.jl:141
 [9] #optimize! at .\none:0 [inlined] (repeats 2 times)
 [10] #parameter_optimizehook#2(::Base.Iterators.Pairs{Union{},Union{},Tuple{},NamedTuple{(),Tuple{}}}, ::Function, ::JuMP.Model) at C:\Users\marce\.julia\packages\ParameterJuMP\NjgYh\src\ParameterJuMP.jl:318
 [11] parameter_optimizehook(::JuMP.Model) at C:\Users\marce\.julia\packages\ParameterJuMP\NjgYh\src\ParameterJuMP.jl:311
 [12] #optimize!#78(::Bool, ::Bool, ::Base.Iterators.Pairs{Union{},Union{},Tuple{},NamedTuple{(),Tuple{}}}, ::Function, ::JuMP.Model, ::Nothing) at C:\Users\marce\.julia\packages\JuMP\iGamg\src\optimizer_interface.jl:131
 [13] optimize! at C:\Users\marce\.julia\packages\JuMP\iGamg\src\optimizer_interface.jl:111 [inlined] (repeats 2 times)
 [14] is_feasible(::JuMP.Model, ::String) at C:\Users\marce\.julia\packages\Polyhedra\2GDrj\src\redundancy.jl:79
 [15] nonredundant_elements(::Polyhedra.Hull{Rational{Int64},Array{Rational{Int64},1},Int64}, ::JuMP.Model, ::Array{JuMP.GenericAffExpr{Float64,JuMP.VariableRef},1}, ::Array{ParameterJuMP.ParameterRef,1}, ::Array{Array{Rational{Int64},1},1}, ::Array{Polyhedra.Index{Rational{Int64},Array{Rational{Int64},1}},1}, ::JuMP.Containers.DenseAxisArray{JuMP.VariableRef,1,Tuple{Array{Polyhedra.Index{Rational{Int64},Array{Rational{Int64},1}},1}},Tuple{Dict{Polyhedra.Index{Rational{Int64},Array{Rational{Int64},1}},Int64}}}) at C:\Users\marce\.julia\packages\Polyhedra\2GDrj\src\redundancy.jl:140
 [16] nonredundant_points(::Polyhedra.Hull{Rational{Int64},Array{Rational{Int64},1},Int64}, ::JuMP.Model, ::Array{JuMP.GenericAffExpr{Float64,JuMP.VariableRef},1}, ::Array{ParameterJuMP.ParameterRef,1}) at C:\Users\marce\.julia\packages\Polyhedra\2GDrj\src\redundancy.jl:196
 [17] removevredundancy(::Polyhedra.Hull{Rational{Int64},Array{Rational{Int64},1},Int64}, ::JuMP.OptimizerFactory) at C:\Users\marce\.julia\packages\Polyhedra\2GDrj\src\redundancy.jl:217
 [18] #remove_redundant_vertices#133(::Nothing, ::Nothing, ::Function, ::VPolytope{Rational{Int64}}) at C:\Users\marce\.julia\dev\LazySets\src\Sets\VPolytope.jl:435
 [19] #remove_redundant_vertices at .\none:0 [inlined]
 [20] #_convex_hull_nd!#223(::Nothing, ::Nothing, ::Function, ::Array{Array{Rational{Int64},1},1}) at C:\Users\marce\.julia\dev\LazySets\src\ConcreteOperations\convex_hull.jl:374
 [21] #_convex_hull_nd! at .\array.jl:0 [inlined]
 [22] #convex_hull!#222(::Nothing, ::Nothing, ::Nothing, ::Function, ::Array{Array{Rational{Int64},1},1}) at C:\Users\marce\.julia\dev\LazySets\src\ConcreteOperations\convex_hull.jl:193
 [23] #convex_hull#221 at .\none:0 [inlined]
 [24] convex_hull(::Array{Array{Rational{Int64},1},1}) at C:\Users\marce\.julia\dev\LazySets\src\ConcreteOperations\convex_hull.jl:148
 [25] top-level scope at none:0

schillic · 2019-09-11T12:02:11Z

i get another error now

Here is a minimal example without LazySets:

julia> using Polyhedra, JuMP, GLPK

julia> ch = convexhull([1//1, 3//1], [0//1, 3//1])
V-representation Polyhedra.PointsHull{Rational{Int64},Array{Rational{Int64},1},Int64}:
2-element iterator of Array{Rational{Int64},1}:
 Rational{Int64}[1//1, 3//1]
 Rational{Int64}[0//1, 3//1]

julia> solver = JuMP.with_optimizer(GLPK.Optimizer, method=GLPK.EXACT)
OptimizerFactory(GLPK.Optimizer, (), Base.Iterators.Pairs(:method=>EXACT))

julia> Polyhedra.removevredundancy(ch, solver)
ERROR: GLPKError("no bf solution found (use GLPK.factorize)")
[...]

@blegat: Should we report this in Polyhedra, JuMP, or GLPK?

blegat · 2019-09-15T08:02:42Z

Seems to be an issue with GLPK, could you print the JuMP model just before this line ?
https://github.com/JuliaPolyhedra/Polyhedra.jl/blob/master/src/redundancy.jl#L79

schillic · 2019-09-15T11:52:31Z

I added a print(model) before that line and got this:

julia> Polyhedra.removevredundancy(ch, solver)
Feasibility
Subject to
 noname + noname = 1.0
 -noname = 0.0
 -3 noname - 3 noname = 0.0
 noname = 0.0
 noname ≥ 0.0
ERROR: GLPKError("no bf solution found (use GLPK.factorize)")
[...]

schillic · 2019-09-16T19:12:14Z

With the dev version of GLPK (see jump-dev/GLPK.jl#116), tests pass locally 🎉
So all we need to do is wait for a new GLPK release.

mforets · 2019-09-17T13:14:38Z

I cannot approve my own PR but it looks good to me. Thanks!

schillic changed the title ~~Bump Polyhedra and JuMP~~ WIP Bump Polyhedra and JuMP Sep 3, 2019

mforets force-pushed the mforets-patch-3 branch from 5b373b5 to 473db8e Compare September 5, 2019 19:43

mforets and others added 3 commits September 5, 2019 19:40

Bump Polyhedra and JuMP

4a9c9c8

bump CDDLib

008f159

use GLPK.EXACT

473db8e

Merge branch 'master' into mforets-patch-3

d04b8fb

blegat mentioned this pull request Sep 15, 2019

EXACT: no bf solution found (use GLPK.factorize) jump-dev/GLPK.jl#116

Closed

schillic changed the title ~~WIP Bump Polyhedra and JuMP~~ Bump Polyhedra and JuMP Sep 17, 2019

schillic approved these changes Sep 17, 2019

View reviewed changes

mforets merged commit bf9a6f6 into master Sep 17, 2019

mforets deleted the mforets-patch-3 branch September 17, 2019 14:38

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Bump Polyhedra and JuMP #1668

Bump Polyhedra and JuMP #1668

mforets commented Sep 2, 2019 •

edited by schillic

Loading

schillic commented Sep 2, 2019

schillic commented Sep 3, 2019 •

edited

Loading

mforets commented Sep 4, 2019 •

edited

Loading

blegat commented Sep 5, 2019

mforets commented Sep 5, 2019

schillic commented Sep 11, 2019

blegat commented Sep 15, 2019

schillic commented Sep 15, 2019

schillic commented Sep 16, 2019

mforets commented Sep 17, 2019

Bump Polyhedra and JuMP #1668

Bump Polyhedra and JuMP #1668

Conversation

mforets commented Sep 2, 2019 • edited by schillic Loading

schillic commented Sep 2, 2019

schillic commented Sep 3, 2019 • edited Loading

mforets commented Sep 4, 2019 • edited Loading

blegat commented Sep 5, 2019

mforets commented Sep 5, 2019

schillic commented Sep 11, 2019

blegat commented Sep 15, 2019

schillic commented Sep 15, 2019

schillic commented Sep 16, 2019

mforets commented Sep 17, 2019

mforets commented Sep 2, 2019 •

edited by schillic

Loading

schillic commented Sep 3, 2019 •

edited

Loading

mforets commented Sep 4, 2019 •

edited

Loading