Add Maros-Meszaros dataset

Author: gdalle

README.md

@@ -17,8 +17,8 @@ Please read and abide by the license of the dataset you plan to use.
 
 ### QP
 
+- [x] [Maros-Meszaros](https://www.doc.ic.ac.uk/~im/#DATA)
 - [ ] [QPLIB](https://qplib.zib.de/)
-- [ ] [Maros-Meszaros](https://www.doc.ic.ac.uk/~im/#DATA)
 
 ## Getting started
 

src/MathOptBenchmarkInstances.jl

@@ -13,7 +13,7 @@ include("lists.jl")
 include("mittelmann.jl")
 include("init.jl")
 
-export Dataset, MIPLIB2017, Netlib, MittelmannLP
+export Dataset, MIPLIB2017, Netlib, MittelmannLP, MarosMeszaros
 export read_instance
 export list_instances
 

src/datasets.jl

@@ -8,8 +8,9 @@ Possible values:
 - [`MIPLIB2017`](@ref)
 - [`Netlib`](@ref)
 - [`MittelmannLP`](@ref)
+- [`MarosMeszaros`](@ref)
 """
-@enum Dataset MIPLIB2017 Netlib MittelmannLP
+@enum Dataset MIPLIB2017 Netlib MittelmannLP MarosMeszaros
 
 """
     MIPLIB2017
@@ -37,3 +38,12 @@ Linear Programs from the Mittelmann benchmark dataset.
 Source: <https://plato.asu.edu/ftp/lptestset/>.
 """
 MittelmannLP
+
+"""
+    MarosMeszaros
+
+Quadratic Programs from the Maros-Meszaros benchmark dataset.
+
+Source: <https://www.doc.ic.ac.uk/~im/#DATA>.
+"""
+MarosMeszaros

src/init.jl

@@ -63,5 +63,22 @@ function __init__()
             ),
         )
     )
+    # Maros-Meszaros
+    register(
+        DataDep(
+            "marosmeszaros",
+            """
+            All instances in the Maros-Meszaros QP benchmark set.
+            Source: https://www.doc.ic.ac.uk/~im/#DATA.
+            Compressed download size: 28.1 MB.
+            Decompressed size: ??""",
+            [
+                "http://www.doc.ic.ac.uk/%7Eim/QPDATA1.ZIP",
+                "http://www.doc.ic.ac.uk/%7Eim/QPDATA2.ZIP",
+                "http://www.doc.ic.ac.uk/%7Eim/QPDATA3.ZIP",
+            ];
+            post_fetch_method = [unpack, unpack, unpack]
+        )
+    )
     return nothing
 end

src/instances.jl

@@ -10,12 +10,17 @@ Return a tuple `(data, path)` where `data isa QPSReader.QPSData` is an object cr
 - [`Dataset`](@ref)
 """
 function read_instance(dataset::Dataset, name::String)
+    if !(name in list_instances(dataset))
+        throw(ArgumentError("Instance $name is not available in dataset $dataset: please use `list_instances` to find valid instance names."))
+    end
     if dataset == MIPLIB2017
         return read_miplib2017_instance(name)
     elseif dataset == Netlib
         return read_netlib_instance(name)
     elseif dataset == MittelmannLP
         return read_mittelmann_lp_instance(name)
+    elseif dataset == MarosMeszaros
+        return read_marosmeszaros_instance(name)
     end
 end
 
@@ -49,14 +54,24 @@ function read_mittelmann_lp_instance(name::String)
     mps_bz2_path2 = joinpath(folder_path, "$name.bz2")
     if ispath(mps_bz2_path1)
         return read_mps(mps_bz2_path1; scratch_subfolder = "mittelman-lp")
-    else
+    elseif ispath(mps_bz2_path2)
         return read_mps(mps_bz2_path2; scratch_subfolder = "mittelman-lp")
     end
 end
 
-function read_mps(path::String; scratch_subfolder::String, mpsformat::Symbol = :free)
+function read_marosmeszaros_instance(name::String)
+    folder_path = datadep"marosmeszaros"
+    qps_path = joinpath(folder_path, "$name.QPS")
+    return read_mps(qps_path)
+end
+
+function read_mps(
+        path::String;
+        scratch_subfolder::Union{String, Nothing} = nothing,
+        mpsformat::Symbol = :free
+    )
     name = splitext(splitext(splitpath(path)[end])[1])[1]
-    if !isdir(joinpath(MPS_SCRATCH, scratch_subfolder))
+    if !isnothing(scratch_subfolder) && !isdir(joinpath(MPS_SCRATCH, scratch_subfolder))
         mkdir(joinpath(MPS_SCRATCH, scratch_subfolder))
     end
     if endswith(path, ".gz") || endswith(path, ".bz2")
@@ -70,15 +85,18 @@ function read_mps(path::String; scratch_subfolder::String, mpsformat::Symbol = :
                 compressed = CodecBzip2.read(path)
                 contents = String(CodecBzip2.transcode(CodecBzip2.Bzip2Decompressor, compressed))
             elseif endswith(path, ".bz2")
-                throw(ArgumentError("File at $path not supported"))
+                throw(ArgumentError("File at $path not supported (not a compressed MPS file).\nIf you know how to handle these files, please open an issue at <https://github.com/JuliaDecisionFocusedLearning/MathOptBenchmarkInstances.jl/issues>."))
             end
             open(mps_path, "w") do f
                 write(f, contents)
             end
         end
-    else
-        @assert endswith(path, ".mps") || endswith(path, ".SIF")
+    elseif endswith(lowercase(path), ".mps") ||
+            endswith(lowercase(path), ".qps") ||
+            endswith(lowercase(path), ".sif")
         mps_path = path
+    else
+        throw(ArgumentError("Invalid file extension for $(splitpath(path)[end])"))
     end
 
     qps_data = with_logger(NullLogger()) do

src/lists.jl

@@ -14,6 +14,8 @@ function list_instances(dataset::Dataset)
         return list_netlib_instances()
     elseif dataset == MittelmannLP
         return list_mittelmann_lp_instances()
+    elseif dataset == MarosMeszaros
+        return list_marosmeszaros_instances()
     end
 end
 
@@ -39,3 +41,7 @@ function list_mittelmann_lp_instances()
     all_names = reduce(vcat, MITTELMANN_LP_INSTANCES)
     return map(n -> string(chopsuffix(split(n, "/")[end], ".mps")), all_names)
 end
+
+function list_marosmeszaros_instances()
+    return map(n -> string(chopsuffix(n, ".QPS")), readdir(datadep"marosmeszaros"))
+end

test/marosmeszaros.jl

@@ -0,0 +1,13 @@
+using MathOptBenchmarkInstances
+using Test
+
+maros_list = list_instances(MarosMeszaros)
+@test length(maros_list) == 138
+
+for name in maros_list
+    if name in ["EXDATA", "QFORPLAN", "QGFRDXPN", "VALUES"]
+        @test_broken read_instance(MarosMeszaros, name) isa Tuple{QPSData, String}
+    else
+        @test read_instance(MarosMeszaros, name) isa Tuple{QPSData, String}
+    end
+end

test/mittelmann.jl

@@ -5,8 +5,47 @@ using Test
 mittelmann_list = list_instances(MittelmannLP)
 @test length(mittelmann_list) == 116
 
-mittelmann_small = [n for n in mittelmann_list if startswith(n, "qap")]
+# only instances under 10M, most from the main list and one or two from each subfolder
+# files that are .bz2 (as opposed to .mps.bz2) don't work for now
+
+mittelmann_small = [
+    # "Linf_520c",
+    # "bdry2",
+    "brazil3",
+    "chromaticindex1024-7",
+    "datt256_lp",
+    "ex10",
+    "bal8x12",
+    "n370a",
+    # "fome11",
+    # "fome13",
+    "graph40-40",
+    "irish-electricity",
+    # "cont1",
+    # "neos",
+    # "ns1687037",
+    # "sgpf5y6",
+    # "watson_1",
+    "neos-5052403-cygnet",
+    "neos-5251015",
+    "16_n14",
+    "lo10",
+    # "nug08-3rd",
+    # "pds-20",
+    "physiciansched3-3",
+    "qap15",
+    # "rail507",
+    "rmine15",
+    "s100",
+    "s250r10",
+    "savsched1",
+    "supportcase10",
+    "woodlands09",
+]
 
 for name in mittelmann_small
+    @test name in mittelmann_list
     @test read_instance(MittelmannLP, name) isa Tuple{QPSData, String}
 end
+
+@test_throws ArgumentError read_instance(MittelmannLP, "bdry2")

test/runtests.jl

@@ -14,7 +14,10 @@ ENV["DATADEPS_ALWAYS_ACCEPT"] = true
     @testset "MIPLIB 2017" begin
         include("miplib2017.jl")
     end
-    @testset "Mittelmann" begin
+    @testset "Mittelmann LP" begin
         include("mittelmann.jl")
     end
+    @testset "Maros-Meszaros" begin
+        include("marosmeszaros.jl")
+    end
 end