Update for JuMP v0.16 #74

yeesian · 2017-03-23T05:01:06Z

Main changes:

drop curly braces in comprehension syntax
JuMP no longer has a mechanism for selecting solvers by default. Not specifying a solver before calling solve() will result in an error.

@ShimShtern let us know if it fixes the problem(s) you've faced in https://discourse.julialang.org/t/issues-with-jumper/2794

codecov-io · 2017-03-29T01:04:21Z

Codecov Report

Merging #74 into master will decrease coverage by 0.19%.
The diff coverage is 100%.

@@            Coverage Diff            @@
##           master      #74     +/-   ##
=========================================
- Coverage   84.33%   84.13%   -0.2%     
=========================================
  Files          15       15             
  Lines        1334     1330      -4     
=========================================
- Hits         1125     1119      -6     
- Misses        209      211      +2

Impacted Files	Coverage Δ
src/uncsets_basic_cut.jl	`98.5% <100%> (-0.07%)`	⬇️
src/solve.jl	`91.02% <100%> (ø)`	⬆️
src/JuMPeR.jl	`83.87% <0%> (-3.23%)`	⬇️
src/print.jl	`86.14% <0%> (-0.38%)`	⬇️
src/uncsets_basic_reform.jl	`100% <0%> (ø)`	⬆️

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 e161a36...8327d36. Read the comment docs.

needs an upstream bugfix in jump-dev/JuMP.jl@2f2aacf555293ff7e47 59e22cefd7b10e6031696

it’s started to be a little dated

relies on the bugfix in JuliaOpt/MathProgBase.jl#160 see discussion in jump-dev/JuMP.jl#999 (comment)

yeesian · 2017-03-29T20:58:30Z

@IainNZ sorry for the noise; this is now ready for review.

IainNZ · 2017-03-30T21:36:06Z

test/uncsets_basic.jl

@@ -191,7 +190,7 @@ print_with_color(:yellow, "  MILP tests...\n")
        @objective(m, Max, 1.1*x[1] + x[2])
        @constraint(m, u1*x[1] + 1*x[2] <= 2)
        @constraint(m, u2*x[1] + 1*x[2] <= 6)
-        @test solve(m, prefer_cuts=cuts) == :Optimal
+        @test solve(m, disable_cuts=cuts) == :Optimal


Whats going on here? Was this test not being run before?

Sorry my bad, it used to fail on travis (error message here: https://travis-ci.org/IainNZ/JuMPeR.jl/builds/216164410#L1304) when it was passing locally, and I was trying to understand the difference between prefer_cuts and disable_cuts, but forgot to switch back.

IainNZ · 2017-03-30T21:37:35Z

test/adp_inventory.jl

        end
    end

    @testset "Affine, manual" begin
-        rm = RobustModel()
+        rm = RobustModel(solver=GLPKSolverLP())


shouldn't this be solver=solver or something like that?

Yeah my bad, it should

IainNZ · 2017-04-05T21:08:11Z

test/adp_inventory.jl

@@ -13,7 +13,7 @@
 #   "Adjustable Robust Solutions of Uncertain Linear Programs"
 #-----------------------------------------------------------------------

-using JuMP, JuMPeR
+using JuMP, JuMPeR, GLPKMathProgInterface


GLPK needed here?

IainNZ · 2017-04-05T21:08:16Z

test/adp_newsvendor.jl

@@ -20,7 +20,7 @@
 #     D ∈ { ‖(D - μ)/σ‖₁ ≤ ⌊√N⌋, ‖(D - μ)/σ‖∞ ≤ 1 }
 #-----------------------------------------------------------------------

-using JuMP, JuMPeR
+using JuMP, JuMPeR, GLPKMathProgInterface


GLPK needed here?

IainNZ · 2017-04-05T21:08:45Z

test/adp_newsvendor.jl

@@ -56,7 +56,7 @@ print_with_color(:yellow, "Adaptive Newsvendor Model...\n")
    end

    @testset "Static, manual" begin
-        m = RobustModel()
+        m = RobustModel(solver=GLPKSolverLP())


Should be =solver?

IainNZ · 2017-04-05T21:09:35Z

test/uncsets.jl

-    @test_throws ErrorException JuMPeR.generate_reform(IncompleteSet(), RobustModel(), Int[])
-    @test_throws ErrorException JuMPeR.generate_cut(IncompleteSet(), RobustModel(), Int[])
-    @test_throws ErrorException JuMPeR.generate_scenario(IncompleteSet(), RobustModel(), Int[])
+    @test_throws ErrorException JuMPeR.setup_set(IncompleteSet(), RobustModel(solver=GLPKSolverLP()), Int[], false, nothing)


Could be lp_solvers[0] then it works with any solver?

IainNZ · 2017-04-05T21:10:56Z

Getting closer
Not sure why tests are failing?

yeesian · 2017-04-05T22:10:31Z

Re the following test:

@test string(new_con) == "5 x[1] + x[2] + 4 x[3] + 11 x[4] $(JuMP.repl[:leq]) 10"

It sometimes fails with

string(new_con) = "4.999999999999986 x[1] + x[2] + 3.9999999999999805 x[3] + 10.999999999999947 x[4] ≤ 9.999999999999975"

How do you feel about replacing it with a test like

@test all(contains(string(new_con), "x[$i]") for i in 1:4)
@test contains(string(new_con), "$(JuMP.repl[:leq])")

?

yeesian · 2017-04-05T22:13:17Z

Also,

isapprox(getvalue(F),44272.82749,atol=TOL)

sometimes fails when

getvalue(F) = 44272.823734690275

when using Ipopt. But passes with ECOS, Gurobi, Clp, and GLPK.

Okay to relax the tolerance?

and relax test on constraint expression printing

yeesian · 2017-04-05T23:01:42Z

Not sure why tests are failing?

Yeah, still failing on Travis for the same (as yet unknown) reasons.

blegat · 2017-10-18T06:49:33Z

@yeesian Are tests passing now ? Github says that all checks are passing.
Related: jump-dev/JuMP.jl#1116

yeesian · 2017-10-19T01:58:53Z

Yeah, I believe so. @IainNZ good to merge?

IainNZ · 2017-10-20T12:17:30Z

Yeah if tests passing, go for it.

…

On Thu, Oct 19, 2017 at 2:58 AM, Yeesian Ng ***@***.***> wrote: Yeah, I believe so. @IainNZ <https://github.com/iainnz> good to merge? — You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub <#74 (comment)>, or mute the thread <https://github.com/notifications/unsubscribe-auth/AAqRm6sL8kpKNLm5WHAYk5PcInjkx_pbks5stqzdgaJpZM4MmNSz> .

-- *Iain Dunning, PhD* http://iaindunning.com

it is correct, but the GLPK solver has been problematic on it. see * #74 (comment) and * 8327d3648b0aa59f5355 f9bd56b65e8dda6fa642 for workarounds in the past.

* update for julia 1.0 * update travis script * add LinearAlgebra to REQUIRE * some updates * more changes * add constructors * define constructor for numbers too * don't broadcast over norm expressions * revert to map * switch back to working version of GLPK * comment out flakey test for now it is correct, but the GLPK solver has been problematic on it. see * #74 (comment) and * 8327d3648b0aa59f5355 f9bd56b65e8dda6fa642 for workarounds in the past. * test on different versions of julia

yeesian force-pushed the jump015 branch from 421500f to 7292a54 Compare March 28, 2017 14:35

update to work for current version of JuMP

e250b70

needs an upstream bugfix in jump-dev/JuMP.jl@2f2aacf555293ff7e47 59e22cefd7b10e6031696

yeesian force-pushed the jump015 branch from d1972c1 to e250b70 Compare March 29, 2017 01:10

yeesian added 2 commits March 28, 2017 23:52

remove warning for jumper0.4

7030d0b

it’s started to be a little dated

fix deprecation

5e0bb8d

yeesian mentioned this pull request Mar 29, 2017

don't automatically transform quadratic constraints into conic form jump-dev/JuMP.jl#999

Merged

set requirements for MPB

04d3f59

relies on the bugfix in JuliaOpt/MathProgBase.jl#160 see discussion in jump-dev/JuMP.jl#999 (comment)

IainNZ reviewed Mar 30, 2017

View reviewed changes

yeesian added 2 commits March 30, 2017 21:20

disable_cuts -> prefer_cuts

477b2fb

test all solvers

affb747

IainNZ reviewed Apr 5, 2017

View reviewed changes

yeesian added 3 commits April 5, 2017 18:24

test for all solvers

d05939f

relax test tolerance

fd0d4a7

test for all solvers

6e6f086

and relax test on constraint expression printing

yeesian mentioned this pull request May 4, 2017

Tag GLPK.jl v0.4.2 JuliaLang/METADATA.jl#8419

Merged

exclude latest version of GLPK

8327d36

yeesian changed the title ~~Update for current version of JuMP~~ Update for JuMP v0.6 Oct 20, 2017

yeesian merged commit 090ce1d into master Oct 20, 2017

yeesian changed the title ~~Update for JuMP v0.6~~ Update for JuMP v0.16 Oct 20, 2017

yeesian deleted the jump015 branch September 26, 2018 01:56

yeesian added a commit that referenced this pull request Feb 23, 2019

comment out flakey test for now

1130ad9

it is correct, but the GLPK solver has been problematic on it. see * #74 (comment) and * 8327d3648b0aa59f5355 f9bd56b65e8dda6fa642 for workarounds in the past.

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Update for JuMP v0.16 #74

Update for JuMP v0.16 #74

yeesian commented Mar 23, 2017 •

edited

Loading

codecov-io commented Mar 29, 2017 •

edited

Loading

yeesian commented Mar 29, 2017

IainNZ Mar 30, 2017

yeesian Mar 31, 2017

IainNZ Mar 30, 2017

yeesian Mar 31, 2017

IainNZ Apr 5, 2017

IainNZ Apr 5, 2017

IainNZ Apr 5, 2017

IainNZ Apr 5, 2017

IainNZ commented Apr 5, 2017

yeesian commented Apr 5, 2017 •

edited

Loading

yeesian commented Apr 5, 2017

yeesian commented Apr 5, 2017

blegat commented Oct 18, 2017

yeesian commented Oct 19, 2017

IainNZ commented Oct 20, 2017 via email

Update for JuMP v0.16 #74

Update for JuMP v0.16 #74

Conversation

yeesian commented Mar 23, 2017 • edited Loading

codecov-io commented Mar 29, 2017 • edited Loading

Codecov Report

yeesian commented Mar 29, 2017

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

IainNZ commented Apr 5, 2017

yeesian commented Apr 5, 2017 • edited Loading

yeesian commented Apr 5, 2017

yeesian commented Apr 5, 2017

blegat commented Oct 18, 2017

yeesian commented Oct 19, 2017

IainNZ commented Oct 20, 2017 via email

yeesian commented Mar 23, 2017 •

edited

Loading

codecov-io commented Mar 29, 2017 •

edited

Loading

yeesian commented Apr 5, 2017 •

edited

Loading