Make package work with persistent solvers; update README.md

README.md

@@ -7,9 +7,8 @@ Table of contents
 -----------------
   * [Features](#features)
   * [Installation](#installation)
-* [Basic usage](#basic-usage)
+  * [Basic Usage](#basic-usage)
      * [Using LearningSolver](#using-learningsolver)
-    * [Selecting the internal MIP solver](#selecting-the-internal-mip-solver)
      * [Describing problem instances](#describing-problem-instances)
      * [Obtaining heuristic solutions](#obtaining-heuristic-solutions)
      * [Saving and loading solver state](#saving-and-loading-solver-state)
@@ -17,6 +16,8 @@ Table of contents
   * [Benchmarking](#benchmarking)
      * [Using BenchmarkRunner](#using-benchmarkrunner)
      * [Saving and loading benchmark results](#saving-and-loading-benchmark-results)
+  * [Customization](#customization)
+     * [Selecting the internal MIP solver](#selecting-the-internal-mip-solver)
   * [Current Limitations](#current-limitations)
   * [References](#references)
   * [Authors](#authors)
@@ -60,17 +61,6 @@ for instance in all_instances:
 
 During the first call to `solver.solve(instance)`, the solver will process the instance from scratch, since no historical information is available, but it will already start gathering information. By calling `solver.fit()`, we instruct the solver to train all the internal Machine Learning models based on the information gathered so far. As this operation can be expensive, it may  be performed after a larger batch of instances has been solved, instead of after every solve. After the first call to `solver.fit()`, subsequent calls to `solver.solve(instance)` will automatically use the trained Machine Learning models to accelerate the solution process.
 
-### Selecting the internal MIP solver
-
-By default, `LearningSolver` uses Gurobi as its internal MIP solver. Alternative solvers can be specified through the `parent_solver`a argument, as follows. To select CPLEX, for example:
-```python
-from miplearn import LearningSolver
-import pyomo.environ as pe
-
-cplex = pe.SolverFactory("cplex")
-solver = LearningSolver(parent_solver=cplex)
-```
-
 ### Describing problem instances
 
 Instances to be solved by `LearningSolver` must derive from the abstract class `miplearn.Instance`. The following three abstract methods must be implemented:
@@ -198,6 +188,25 @@ benchmark.load_fit("training_data.bin")
 benchmark.parallel_solve(test_instances)
 ```
 
+Customization
+-------------
+
+### Selecting the internal MIP solver
+
+By default, `LearningSolver` uses [Gurobi](https://www.gurobi.com/) as its internal MIP solver. Alternative solvers can be specified through the `internal_solver_factory` constructor argument. This argument should provide a function (with no arguments) that constructs, configures and returns the desired solver. To select CPLEX, for example:
+```python
+from miplearn import LearningSolver
+import pyomo.environ as pe
+
+def cplex_factory():
+    cplex = pe.SolverFactory("cplex_persistent")
+    cplex.options["threads"] = 4
+    return cplex
+
+solver = LearningSolver(internal_solver_factory=cplex_factory)
+```
+
+
 Current Limitations
 -------------------
 

miplearn/solvers.py

@@ -13,6 +13,10 @@ from joblib import Parallel, delayed
 import multiprocessing
 
 
+def _gurobi_factory():
+    solver = pe.SolverFactory('gurobi_persistent')
+    solver.options["threads"] = 4
+    return solver
 
 class LearningSolver:
     """
@@ -22,11 +26,11 @@ class LearningSolver:
 
     def __init__(self,
                  threads=4,
-                 parent_solver=pe.SolverFactory('gurobi'),
+                 internal_solver_factory=_gurobi_factory,
                  ws_predictor=KnnWarmStartPredictor(),
                  mode="exact"):
-        self.parent_solver = parent_solver
+        self.internal_solver_factory = internal_solver_factory
-        self.parent_solver.options["threads"] = threads
+        self.internal_solver = self.internal_solver_factory()
         self.mode = mode
         self.x_train = {}
         self.y_train = {}
@@ -86,8 +90,11 @@ class LearningSolver:
         return solve_results
                 
     def parallel_solve(self, instances, n_jobs=4, label="Solve"):
+        self.parentSolver = None
+        
         def _process(instance):
-            solver = deepcopy(self)
+            solver = copy(self)
+            solver.internal_solver = solver.internal_solver_factory()
             results = solver.solve(instance)
             return {
                 "x_train": solver.x_train,
@@ -143,8 +150,8 @@ class LearningSolver:
             self.ws_predictors = self.ws_predictors
 
     def _solve(self, model, tee=False):
-        if hasattr(self.parent_solver, "set_instance"):
+        if hasattr(self.internal_solver, "set_instance"):
-            self.parent_solver.set_instance(model)
+            self.internal_solver.set_instance(model)
-            return self.parent_solver.solve(tee=tee, warmstart=True)
+            return self.internal_solver.solve(tee=tee, warmstart=True)
         else:
-            return self.parent_solver.solve(model, tee=tee, warmstart=True)
+            return self.internal_solver.solve(model, tee=tee, warmstart=True)