Add gmi test; update H5

src/Cuts/Cuts.jl

@@ -10,7 +10,6 @@ include("tableau/structs.jl")
 
 # include("blackbox/cplex.jl")
 include("tableau/numerics.jl")
-include("tableau/collect.jl")
 include("tableau/gmi.jl")
 include("tableau/moi.jl")
 include("tableau/tableau.jl")

src/Cuts/tableau/gmi.jl

@@ -154,7 +154,7 @@ function collect_gmi(
         else
             all_cuts.lhs = [all_cuts.lhs; cuts.lhs[keep, :]]
             all_cuts.lb = [all_cuts.lb; cuts.lb[keep]]
-            all_cuts.lb = [all_cuts.lb; cuts.lb[keep]]
+            all_cuts.ub = [all_cuts.ub; cuts.ub[keep]]
         end
         push!(stats_ncuts, length(all_cuts.lb))
 

test/Project.toml

@@ -17,6 +17,7 @@ MIPLearn = "2b1277c3-b477-4c49-a15e-7ba350325c68"
 PyCall = "438e738f-606a-5dbb-bf0a-cddfbfd45ab0"
 Revise = "295af30f-e4ad-537b-8983-00126c2a3abe"
 SCIP = "82193955-e24f-5292-bf16-6f2c5261a85f"
+SparseArrays = "2f01184e-e22b-5df5-ae63-d93ebab69eaf"
 Test = "8dfed614-e22c-5e08-85e1-65c5234f0b40"
 
 [compat]

test/src/BB/test_bb.jl

@@ -86,11 +86,7 @@ function bb_run(optimizer_name, optimizer; large = true)
                     BB.ReliabilityBranching(aggregation = :min, collect = true),
                 ]
                     h5 = H5File("$FIXTURES/$instance.h5")
-                    mip_lower_bound = h5.get_scalar("mip_lower_bound")
+                    mip_obj_bound = h5.get_scalar("mip_obj_bound")
-                    mip_upper_bound = h5.get_scalar("mip_upper_bound")
-                    mip_sense = h5.get_scalar("mip_sense")
-                    mip_primal_bound =
-                        mip_sense == "min" ? mip_upper_bound : mip_lower_bound
                     h5.file.close()
 
                     mip = BB.init(optimizer)
@@ -98,7 +94,7 @@ function bb_run(optimizer_name, optimizer; large = true)
                     @info optimizer_name, branch_rule, instance
                     @time BB.solve!(
                         mip,
-                        initial_primal_bound = mip_primal_bound,
+                        initial_primal_bound = mip_obj_bound,
                         print_interval = 1,
                         node_limit = 25,
                         branch_rule = branch_rule,

test/src/Cuts/tableau/test_gmi.jl

@@ -0,0 +1,23 @@
+#  MIPLearn: Extensible Framework for Learning-Enhanced Mixed-Integer Optimization
+#  Copyright (C) 2020-2024, UChicago Argonne, LLC. All rights reserved.
+#  Released under the modified BSD license. See COPYING.md for more details.
+
+using HiGHS
+
+function test_cuts_tableau_gmi()
+    mps_filename = "$BASEDIR/../fixtures/bell5.mps.gz"
+    h5_filename = "$BASEDIR/../fixtures/bell5.h5"
+    collect_gmi(mps_filename, optimizer = HiGHS.Optimizer)
+    h5 = H5File(h5_filename, "r")
+    try
+        cuts_lb = h5.get_array("cuts_lb")
+        cuts_ub = h5.get_array("cuts_ub")
+        cuts_lhs = h5.get_sparse("cuts_lhs")
+        n_cuts = length(cuts_lb)
+        @test n_cuts > 0
+        @test n_cuts == length(cuts_ub)
+        @test cuts_lhs.shape[1] == n_cuts
+    finally
+        h5.close()
+    end
+end

test/src/MIPLearnT.jl

@@ -19,6 +19,7 @@ include("BB/test_bb.jl")
 include("components/test_cuts.jl")
 include("components/test_lazy.jl")
 include("Cuts/BlackBox/test_cplex.jl")
+include("Cuts/tableau/test_gmi.jl")
 include("problems/test_setcover.jl")
 include("problems/test_stab.jl")
 include("problems/test_tsp.jl")

test/src/test_io.jl

@@ -4,6 +4,7 @@
 
 using MIPLearn
 using JLD2
+using SparseArrays
 
 struct _TestStruct
     n::Int
@@ -35,6 +36,8 @@ function test_h5()
     _test_roundtrip_array(h5, [1, 2, 3])
     _test_roundtrip_array(h5, [1.0, 2.0, 3.0])
     _test_roundtrip_str_array(h5, ["A", "BB", "CCC"])
+    _test_roundtrip_sparse(h5, sparse([1; 2; 3], [1; 2; 3], [1; 2; 3]))
+    # _test_roundtrip_sparse(h5, sparse([1; 2; 3], [1; 2; 3], [1; 2; 3], 4, 4))
     @test h5.get_array("unknown-key") === nothing
     h5.close()
 end
@@ -79,3 +82,11 @@ function _test_roundtrip_str_array(h5, original)
     @test recovered !== nothing
     @test all(original .== recovered)
 end
+
+function _test_roundtrip_sparse(h5, original)
+    h5.put_sparse("key", original)
+    recovered = MIPLearn.convert(SparseMatrixCSC, h5.get_sparse("key"))
+    @test recovered !== nothing
+    @test size(original) == size(recovered)
+    @test all(original .== recovered)
+end