Merge pull request #94 from chrhansk/feature-mps-solver

Add solver for MPS-encoded LPs

docs/conf.py

@@ -14,7 +14,7 @@
 project = "pygradflow"
 copyright = "2023, Christoph Hansknecht"
 author = "Christoph Hansknecht"
-release = "0.5.1"
+release = "0.5.2"
 
 # -- General configuration ---------------------------------------------------
 # https://www.sphinx-doc.org/en/master/usage/configuration.html#general-configuration

pygradflow/runners/mps_runner.py

@@ -0,0 +1,91 @@
+import os
+
+import mpspy
+import numpy as np
+
+from pygradflow.problem import Problem
+from pygradflow.util import sparse_zero
+
+from .instance import Instance
+from .runner import Runner
+
+
+class LinearProblem(Problem):
+    def __init__(self, instance):
+        self.instance = instance
+
+        super().__init__(
+            var_lb=instance.var_lb,
+            var_ub=instance.var_ub,
+            cons_lb=instance.cons_lb,
+            cons_ub=instance.cons_ub,
+        )
+
+    def obj(self, x):
+        instance = self.instance
+        return np.dot(instance.obj, x)
+
+    def obj_grad(self, x):
+        instance = self.instance
+        return instance.obj
+
+    def cons(self, x):
+        instance = self.instance
+        coeffs = instance.coeffs
+        return coeffs.dot(x)
+
+    def cons_jac(self, x):
+        instance = self.instance
+        return instance.coeffs
+
+    def lag_hess(self, x, y):
+        num_vars = self.num_vars
+        return sparse_zero(shape=(num_vars, num_vars))
+
+
+class MPSInstance(Instance):
+    def __init__(self, instance):
+        self.instance = instance
+
+        num_vars = instance.num_vars
+        num_cons = instance.num_cons
+        name = instance.name
+
+        super().__init__(name, num_vars, num_cons)
+
+    def problem(self):
+        return LinearProblem(self.instance)
+
+    def x0(self):
+        instance = self.instance
+        num_vars = instance.num_vars
+
+        x0 = np.zeros((num_vars,))
+
+        return np.clip(x0, instance.var_lb, instance.var_ub)
+
+
+class MPSRunner(Runner):
+    def __init__(self):
+        super().__init__(name="mps")
+
+    def get_instances(self, args):
+        instances = []
+        for file in os.listdir(args.directory):
+            is_qplib = file.endswith(".mps") or file.endswith(".mps.gz")
+
+            if is_qplib:
+                filename = os.path.join(args.directory, file)
+                description = mpspy.read_mps(filename)
+                instances.append(MPSInstance(description))
+
+        return instances
+
+    def parser(self):
+        parser = super().parser()
+        parser.add_argument("directory")
+        return parser
+
+
+if __name__ == "__main__":
+    MPSRunner().main()

pygradflow/runners/qplib_runner.py

@@ -35,10 +35,10 @@ def lag_hess(self, x, y):
         return self.problem.lag_hess(x, y)
 
     def x0(self):
-        return self.problem.x0()
+        return self.problem.x0
 
     def y0(self):
-        return self.problem.y0()
+        return self.problem.y0
 
 
 class QPLIBInstance(Instance):

pyproject.toml

@@ -1,6 +1,6 @@
 [tool.poetry]
 name = "pygradflow"
-version = "0.5.1"
+version = "0.5.2"
 description = "PyGradFlow is a simple implementation of the sequential homotopy method to be used to solve general nonlinear programs."
 authors = ["Christoph Hansknecht <[email protected]>"]
 readme = "README.md"
@@ -25,6 +25,7 @@ mypy = "^1.7.1"
 types-pyyaml = "^6.0.12"
 pytest-timeout = "^2.3.1"
 pyflakes = "^3.2.0"
+mpspy = "^0.1.3"
 
 [tool.poetry.group.docs.dependencies]
 sphinx = "^7.2.6"