Add Mittelmann benchmark

Author: gdalle

Project.toml

@@ -7,12 +7,14 @@ authors = ["Guillaume Dalle"]
 projects = ["test"]
 
 [deps]
+CodecBzip2 = "523fee87-0ab8-5b00-afb7-3ecf72e48cfd"
 DataDeps = "124859b0-ceae-595e-8997-d05f6a7a8dfe"
 GZip = "92fee26a-97fe-5a0c-ad85-20a5f3185b63"
 Logging = "56ddb016-857b-54e1-b83d-db4d58db5568"
 QPSReader = "10f199a5-22af-520b-b891-7ce84a7b1bd0"
 
 [compat]
+CodecBzip2 = "0.8.5"
 DataDeps = "0.7.13"
 GZip = "0.6.2"
 Logging = "1.10.0"

README.md

@@ -8,7 +8,7 @@ Supported datasets:
 
 - [x] [Netlib](https://www.netlib.org/lp/data/index.html)
 - [x] [MIPLIB 2017](https://miplib.zib.de/index.html)
-- [ ] [Mittelman benchmark](https://plato.asu.edu/ftp/lptestset/)
+- [ ] [Mittelmann benchmark](https://plato.asu.edu/ftp/lptestset/)
 
 The source files are downloaded automatically thanks to [DataDeps.jl](https://github.com/oxinabox/DataDeps.jl).
 The returned problem format is `QPSData` from [QPSReader.jl](https://github.com/JuliaSmoothOptimizers/QPSReader.jl).

src/MathProgBenchmarks.jl

@@ -1,61 +1,19 @@
 module MathProgBenchmarks
 
+import CodecBzip2
 using DataDeps
-using GZip
+import GZip
 using Logging
 using QPSReader
 
 include("datasets.jl")
 include("instances.jl")
 include("lists.jl")
+include("mittelmann.jl")
+include("init.jl")
 
-export Dataset, MIPLIB2017, MIPLIB2017Benchmark, Netlib
+export Dataset, MIPLIB2017, Netlib, MittelmannLP
 export read_instance
 export list_instances
 
-function __init__()
-    register(
-        DataDep(
-            "miplib2017-collection",
-            """
-            All instances in the MIPLIB 2017 collection set (size: 3.5 GB).
-            Source: https://miplib.zib.de/index.html.""",
-            "https://miplib.zib.de/downloads/collection.zip",
-            post_fetch_method = unpack,
-        ),
-    )
-    register(
-        DataDep(
-            "miplib2017-benchmark",
-            """
-            All instances in the MIPLIB 2017 benchmark set (size: 317 MB).
-            Source: https://miplib.zib.de/index.html.""",
-            "https://miplib.zib.de/downloads/benchmark.zip",
-            post_fetch_method = unpack,
-            "c756eefd544d83b31809306b45d3549a1a5b9378e6aa78b68738b1a3b6a418fa"
-        ),
-    )
-    register(
-        DataDep(
-            "miplib2017-collection-list",
-            """
-            List of instances in the MIPLIB 2017 collection.
-            Source: https://miplib.zib.de/index.html.""",
-            "https://miplib.zib.de/downloads/collection-v1.test",
-            "c416f857083e2b7ec6edad87acc80437e2ea4be65debdbe7c4f3072881981c3f"
-        ),
-    )
-    register(
-        DataDep(
-            "miplib2017-benchmark-list",
-            """
-            List of instances in the MIPLIB 2017 benchmark.
-            Source: https://miplib.zib.de/index.html.""",
-            "https://miplib.zib.de/downloads/benchmark-v2.test",
-            "0ee9758dba64bb1689cc121c794ee547992a046bcabe1cce9ea79158375dabd5"
-        ),
-    )
-    return nothing
-end
-
 end # module MathProgBenchmarks

src/datasets.jl

@@ -5,34 +5,35 @@ Enum type for mathematical programming datasets.
 
 Possible values:
 
-- [`MIPLIB2017`](@ref), [`MIPLIB2017Benchmark`](@ref)
+- [`MIPLIB2017`](@ref)
 - [`Netlib`](@ref)
+- [`MittelmannLP`](@ref)
 """
-@enum Dataset MIPLIB2017 MIPLIB2017Benchmark Netlib
+@enum Dataset MIPLIB2017 Netlib MittelmannLP
 
 """
     MIPLIB2017
 
 Mixed Integer Linear Programs from the MIPLIB2017 collection dataset.
 
-Source: <https://miplib.zib.de/>.
+Source: <https://miplib.zib.de/tag_collection.html>.
 """
 MIPLIB2017
 
 """
-    MIPLIB2017Benchmark
+    Netlib
 
-Mixed Integer Linear Programs from the MIPLIB2017 benchmark dataset (much smaller than the collection dataset).
+Linear Programs from the Netlib dataset.
 
-Source: <https://miplib.zib.de/>.
+Source: <https://www.netlib.org/lp/data/index.html>.
 """
-MIPLIB2017Benchmark
+Netlib
 
 """
-    Netlib
+    MittelmannLP
 
-Linear Programs from the Netlib dataset.
+Linear Programs from the Mittelmann benchmark dataset.
 
-Source: <https://www.netlib.org/lp/data/index.html>.
+Source: <https://plato.asu.edu/ftp/lptestset/>.
 """
-Netlib
+MittelmannLP

src/init.jl

@@ -0,0 +1,67 @@
+function __init__()
+    # MIPLIB 2017
+    register(
+        DataDep(
+            "miplib2017-collection",
+            """
+            All instances in the MIPLIB 2017 collection set (size: 3.5 GB).
+            Source: https://miplib.zib.de/index.html.""",
+            "https://miplib.zib.de/downloads/collection.zip",
+            post_fetch_method = unpack,
+        ),
+    )
+    register(
+        DataDep(
+            "miplib2017-benchmark",
+            """
+            All instances in the MIPLIB 2017 benchmark set (size: 317 MB).
+            Source: https://miplib.zib.de/index.html.""",
+            "https://miplib.zib.de/downloads/benchmark.zip",
+            post_fetch_method = unpack,
+            "c756eefd544d83b31809306b45d3549a1a5b9378e6aa78b68738b1a3b6a418fa"
+        ),
+    )
+    register(
+        DataDep(
+            "miplib2017-collection-list",
+            """
+            List of instances in the MIPLIB 2017 collection.
+            Source: https://miplib.zib.de/index.html.""",
+            "https://miplib.zib.de/downloads/collection-v1.test",
+            "c416f857083e2b7ec6edad87acc80437e2ea4be65debdbe7c4f3072881981c3f"
+        ),
+    )
+    register(
+        DataDep(
+            "miplib2017-benchmark-list",
+            """
+            List of instances in the MIPLIB 2017 benchmark.
+            Source: https://miplib.zib.de/index.html.""",
+            "https://miplib.zib.de/downloads/benchmark-v2.test",
+            "0ee9758dba64bb1689cc121c794ee547992a046bcabe1cce9ea79158375dabd5"
+        ),
+    )
+    # Mittelmann
+    for (folder, names) in pairs(MITTELMANN_LP_INSTANCES)
+        for name in names
+            @info "$folder / $name"
+            datadep_name = "mittelmann-lp-$name"
+            # TODO: some paths are just .bz2, following no visible pattern
+            if isnothing(folder)
+                datadep_msg = """
+                Instance $name from the Mittelmann LP benchmark set.
+                Source: https://plato.asu.edu/ftp/lptestset/."""
+                datadep_url = "https://plato.asu.edu/ftp/lptestset/$name.mps.bz2"
+            else
+                datadep_msg = """
+                Instance $name from the Mittelmann LP benchmark set (folder $folder).
+                Source: https://plato.asu.edu/ftp/lptestset/."""
+                datadep_url = "https://plato.asu.edu/ftp/lptestset/$folder/$name.mps.bz2"
+            end
+            register(
+                DataDep(datadep_name, datadep_msg, datadep_url)
+            )
+        end
+    end
+    return nothing
+end

src/instances.jl

@@ -8,15 +8,17 @@ Read the instance identified by `name` from a given `dataset` and return a `QPSR
 - [`Dataset`](@ref)
 """
 function read_instance(dataset::Dataset, name::String)
-    if dataset == MIPLIB2017 || dataset == MIPLIB2017Benchmark
-        return read_miplib2017_instance(dataset, name)
+    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)
     end
 end
 
-function read_miplib2017_instance(dataset::Dataset, name::String)
-    folder = if dataset == MIPLIB2017Benchmark
+function read_miplib2017_instance(name::String)
+    folder = if name in list_miplib2017_instances(; benchmark_only = true)
         datadep"miplib2017-benchmark"
     else
         datadep"miplib2017-collection"
@@ -39,6 +41,17 @@ function read_netlib_instance(name::String)
     return read_mps(sif_path; mpsformat)
 end
 
+function read_mittelmann_lp_instance(name::String)
+    folder_path = @datadep_str("mittelmann-lp-$name")
+    path1 = joinpath(folder_path, "$name.mps.bz2")
+    path2 = joinpath(folder_path, "$name.bz2")
+    if ispath(path1)
+        return read_mps(path1)
+    else
+        return read_mps(path2)
+    end
+end
+
 function read_mps(path::String; mpsformat::Symbol = :free)
     if endswith(path, ".mps.gz")
         contents = GZip.open(path, "r") do f
@@ -48,6 +61,13 @@ function read_mps(path::String; mpsformat::Symbol = :free)
         open(mps_path, "w") do f
             write(f, contents)
         end
+    elseif endswith(path, ".bz2") || endswith(path, ".mps.bz2")
+        compressed = CodecBzip2.read(path)
+        contents = String(CodecBzip2.transcode(CodecBzip2.Bzip2Decompressor, compressed))
+        mps_path = string(tempname(), ".mps")
+        open(mps_path, "w") do f
+            write(f, contents)
+        end
     else
         @assert endswith(path, ".mps") || endswith(path, ".SIF")
         mps_path = path

src/lists.jl

@@ -8,15 +8,17 @@ Return the names of all available instances in a given `dataset`.
 - [`Dataset`](@ref)
 """
 function list_instances(dataset::Dataset)
-    if dataset == MIPLIB2017 || dataset == MIPLIB2017Benchmark
-        return list_miplib2017_instances(dataset)
+    if dataset == MIPLIB2017
+        return list_miplib2017_instances()
     elseif dataset == Netlib
         return list_netlib_instances()
+    elseif dataset == MittelmannLP
+        return list_mittelmann_lp_instances()
     end
 end
 
-function list_miplib2017_instances(dataset::Dataset)
-    list_path = if dataset == MIPLIB2017Benchmark
+function list_miplib2017_instances(; benchmark_only::Bool = false)
+    list_path = if benchmark_only
         joinpath(datadep"miplib2017-benchmark-list", "benchmark-v2.test")
     else
         joinpath(datadep"miplib2017-collection-list", "collection-v1.test")
@@ -32,3 +34,7 @@ function list_netlib_instances()
     valid_instances = filter(n -> endswith(n, ".SIF"), readdir(netlib_path))
     return map(n -> lowercase(chopsuffix(n, ".SIF")), valid_instances)
 end
+
+function list_mittelmann_lp_instances()
+    return reduce(vcat, values(MITTELMANN_LP_INSTANCES))
+end