feat: feat: removed ExtraIn and ExtraOut from LocalSearchOptimizer interface

README.md

@@ -6,7 +6,7 @@ Rust library for local search optimization
 All of the algorithms are parallelized with Rayon.
 
 1. Hill Climbing.
-2. Tabu Search.
+2. Tabu Search. To use this optimizer you also need to implement your problem specific tabu list.
 3. Simulated Annealing
 4. Epsilon Greedy Search, a variant of Hill Climbing which accepts the trial solution with a constant probabilith even if the score of the trial solution is worse than the previous one.
 5. Relative Annealing, a variant of Simulated Annealing which uses relative score diff to calculate transition probability.
@@ -110,7 +110,7 @@ fn main() {
         pb.set_position(op.iter as u64);
     };
 
-    let res = opt.run(&model, None, n_iter, time_limit, Some(&callback), ());
+    let res = opt.run(&model, None, n_iter, time_limit, Some(&callback));
     pb.finish();
     dbg!(res.unwrap());
 }

examples/quadratic_model.rs

@@ -91,7 +91,7 @@ fn main() {
         pb.set_position(op.iter as u64);
     };
 
-    let res = opt.run(&model, None, n_iter, time_limit, Some(&callback), ());
+    let res = opt.run(&model, None, n_iter, time_limit, Some(&callback));
     pb.finish();
     dbg!(res.unwrap());
 }

examples/tsp_model.rs

@@ -179,24 +179,32 @@ impl DequeTabuList {
     }
 }
 
-impl TabuList for DequeTabuList {
-    type Item = (SolutionType, TransitionType);
+impl Default for DequeTabuList {
+    fn default() -> Self {
+        Self::new(10)
+    }
+}
 
-    fn contains(&self, item: &Self::Item) -> bool {
+impl TabuList<TSPModel> for DequeTabuList {
+    fn contains(&self, item: &(SolutionType, TransitionType)) -> bool {
         let (_, (_, inserted_edges)) = item;
         inserted_edges
             .iter()
             .any(|edge| self.buff.iter().any(|e| *e == *edge))
     }
 
-    fn append(&mut self, item: Self::Item) {
+    fn append(&mut self, item: (SolutionType, TransitionType)) {
         let (_, (removed_edges, _)) = item;
         for edge in removed_edges {
             if self.buff.iter().all(|e| *e != edge) {
                 self.buff.append(edge);
             }
         }
     }
+
+    fn set_size(&mut self, n: usize) {
+        self.buff = RingBuffer::new(n);
+    }
 }
 
 // The output is wrapped in a Result to allow matching on errors
@@ -261,14 +269,13 @@ fn main() {
 
     println!("run hill climbing");
     let optimizer = HillClimbingOptimizer::new(1000, 200);
-    let (final_solution, final_score, _) = optimizer
+    let (final_solution, final_score) = optimizer
         .run(
             &tsp_model,
             initial_solution.clone(),
             n_iter,
             time_limit,
             Some(&callback),
-            (),
         )
         .unwrap();
     println!(
@@ -280,16 +287,14 @@ fn main() {
     pb.reset();
 
     println!("run tabu search");
-    let tabu_list = DequeTabuList::new(20);
-    let optimizer = TabuSearchOptimizer::new(patience, 200, 10);
-    let (final_solution, final_score, _) = optimizer
+    let optimizer = TabuSearchOptimizer::<TSPModel, DequeTabuList>::new(patience, 200, 10, 20);
+    let (final_solution, final_score) = optimizer
         .run(
             &tsp_model,
             initial_solution.clone(),
             n_iter,
             time_limit,
             Some(&callback),
-            tabu_list,
         )
         .unwrap();
     println!(
@@ -301,15 +306,14 @@ fn main() {
     pb.reset();
 
     println!("run annealing");
-    let optimizer = SimulatedAnnealingOptimizer::new(patience, 200);
-    let (final_solution, final_score, _) = optimizer
+    let optimizer = SimulatedAnnealingOptimizer::new(patience, 200, 200.0, 50.0);
+    let (final_solution, final_score) = optimizer
         .run(
             &tsp_model,
             initial_solution.clone(),
             n_iter,
             time_limit,
             Some(&callback),
-            (200.0, 50.0),
         )
         .unwrap();
     println!(
@@ -322,14 +326,13 @@ fn main() {
 
     println!("run epsilon greedy");
     let optimizer = EpsilonGreedyOptimizer::new(patience, 200, 10, 0.3);
-    let (final_solution, final_score, _) = optimizer
+    let (final_solution, final_score) = optimizer
         .run(
             &tsp_model,
             initial_solution.clone(),
             n_iter,
             time_limit,
             Some(&callback),
-            (),
         )
         .unwrap();
     println!(
@@ -342,14 +345,13 @@ fn main() {
 
     println!("run relative annealing");
     let optimizer = RelativeAnnealingOptimizer::new(patience, 200, 10, 1e1);
-    let (final_solution, final_score, _) = optimizer
+    let (final_solution, final_score) = optimizer
         .run(
             &tsp_model,
             initial_solution,
             n_iter,
             time_limit,
             Some(&callback),
-            (),
         )
         .unwrap();
     println!(

src/optim/base.rs

@@ -6,11 +6,6 @@ use crate::{callback::OptCallbackFn, Duration, OptModel};
 /// Optimizer that implements local search algorithm
 #[auto_impl(&, Box, Rc, Arc)]
 pub trait LocalSearchOptimizer<M: OptModel> {
-    /// Extra input type
-    type ExtraIn;
-    /// Extra output type
-    type ExtraOut;
-
     /// Start optimization
     fn optimize<F>(
         &self,
@@ -20,8 +15,7 @@ pub trait LocalSearchOptimizer<M: OptModel> {
         n_iter: usize,
         time_limit: Duration,
         callback: Option<&F>,
-        extra_in: Self::ExtraIn,
-    ) -> (M::SolutionType, M::ScoreType, Self::ExtraOut)
+    ) -> (M::SolutionType, M::ScoreType)
     where
         M: OptModel,
         F: OptCallbackFn<M::SolutionType, M::ScoreType>;
@@ -34,8 +28,7 @@ pub trait LocalSearchOptimizer<M: OptModel> {
         n_iter: usize,
         time_limit: Duration,
         callback: Option<&F>,
-        extra_in: Self::ExtraIn,
-    ) -> AnyResult<(M::SolutionType, M::ScoreType, Self::ExtraOut)>
+    ) -> AnyResult<(M::SolutionType, M::ScoreType)>
     where
         M: OptModel,
         F: OptCallbackFn<M::SolutionType, M::ScoreType>,
@@ -51,18 +44,17 @@ pub trait LocalSearchOptimizer<M: OptModel> {
         let (initial_solution, initial_score) =
             model.preprocess_solution(initial_solution, initial_score)?;
 
-        let (solution, score, extra) = self.optimize(
+        let (solution, score) = self.optimize(
             model,
             initial_solution,
             initial_score,
             n_iter,
             time_limit,
             callback,
-            extra_in,
         );
 
         let (solution, score) = model.postprocess_solution(solution, score);
-        Ok((solution, score, extra))
+        Ok((solution, score))
     }
 }
 

src/optim/epsilon_greedy.rs

@@ -39,8 +39,6 @@ impl EpsilonGreedyOptimizer {
 }
 
 impl<M: OptModel> LocalSearchOptimizer<M> for EpsilonGreedyOptimizer {
-    type ExtraIn = ();
-    type ExtraOut = ();
     /// Start optimization
     ///
     /// - `model` : the model to optimize
@@ -49,7 +47,6 @@ impl<M: OptModel> LocalSearchOptimizer<M> for EpsilonGreedyOptimizer {
     /// - `n_iter`: maximum iterations
     /// - `time_limit`: maximum iteration time
     /// - `callback` : callback function that will be invoked at the end of each iteration
-    /// - `_extra_in` : not used
     fn optimize<F>(
         &self,
         model: &M,
@@ -58,8 +55,7 @@ impl<M: OptModel> LocalSearchOptimizer<M> for EpsilonGreedyOptimizer {
         n_iter: usize,
         time_limit: Duration,
         callback: Option<&F>,
-        _extra_in: Self::ExtraIn,
-    ) -> (M::SolutionType, M::ScoreType, Self::ExtraOut)
+    ) -> (M::SolutionType, M::ScoreType)
     where
         M: OptModel + Sync + Send,
         F: OptCallbackFn<M::SolutionType, M::ScoreType>,
@@ -77,7 +73,6 @@ impl<M: OptModel> LocalSearchOptimizer<M> for EpsilonGreedyOptimizer {
             n_iter,
             time_limit,
             callback,
-            _extra_in,
         )
     }
 }

src/optim/generic.rs

@@ -55,8 +55,6 @@ where
     FT: TransitionProbabilityFn<ST>,
     M: OptModel<ScoreType = ST>,
 {
-    type ExtraIn = ();
-    type ExtraOut = ();
     /// Start optimization
     ///
     /// - `model` : the model to optimize
@@ -65,7 +63,6 @@ where
     /// - `n_iter`: maximum iterations
     /// - `time_limit`: maximum iteration time
     /// - `callback` : callback function that will be invoked at the end of each iteration
-    /// - `_extra_in` : not used
     fn optimize<F>(
         &self,
         model: &M,
@@ -74,8 +71,7 @@ where
         n_iter: usize,
         time_limit: Duration,
         callback: Option<&F>,
-        _extra_in: Self::ExtraIn,
-    ) -> (M::SolutionType, M::ScoreType, Self::ExtraOut)
+    ) -> (M::SolutionType, M::ScoreType)
     where
         F: OptCallbackFn<M::SolutionType, M::ScoreType>,
     {
@@ -141,6 +137,6 @@ where
         }
 
         let best_solution = (*best_solution.borrow()).clone();
-        (best_solution, best_score, ())
+        (best_solution, best_score)
     }
 }

src/optim/hill_climbing.rs

@@ -19,9 +19,6 @@ impl HillClimbingOptimizer {
 }
 
 impl<M: OptModel> LocalSearchOptimizer<M> for HillClimbingOptimizer {
-    type ExtraIn = ();
-    type ExtraOut = ();
-
     /// Start optimization
     ///
     /// - `model` : the model to optimize
@@ -39,8 +36,7 @@ impl<M: OptModel> LocalSearchOptimizer<M> for HillClimbingOptimizer {
         n_iter: usize,
         time_limit: Duration,
         callback: Option<&F>,
-        _extra_in: Self::ExtraIn,
-    ) -> (M::SolutionType, M::ScoreType, Self::ExtraOut)
+    ) -> (M::SolutionType, M::ScoreType)
     where
         F: OptCallbackFn<M::SolutionType, M::ScoreType>,
     {
@@ -52,7 +48,6 @@ impl<M: OptModel> LocalSearchOptimizer<M> for HillClimbingOptimizer {
             n_iter,
             time_limit,
             callback,
-            _extra_in,
         )
     }
 }

src/optim/logistic_annealing.rs

@@ -44,8 +44,6 @@ impl LogisticAnnealingOptimizer {
 }
 
 impl<M: OptModel<ScoreType = NotNan<f64>>> LocalSearchOptimizer<M> for LogisticAnnealingOptimizer {
-    type ExtraIn = ();
-    type ExtraOut = ();
     /// Start optimization
     ///
     /// - `model` : the model to optimize
@@ -54,7 +52,6 @@ impl<M: OptModel<ScoreType = NotNan<f64>>> LocalSearchOptimizer<M> for LogisticA
     /// - `n_iter`: maximum iterations
     /// - `time_limit`: maximum iteration time
     /// - `callback` : callback function that will be invoked at the end of each iteration
-    /// - `_extra_in` : not used
     fn optimize<F>(
         &self,
         model: &M,
@@ -63,8 +60,7 @@ impl<M: OptModel<ScoreType = NotNan<f64>>> LocalSearchOptimizer<M> for LogisticA
         n_iter: usize,
         time_limit: Duration,
         callback: Option<&F>,
-        _extra_in: Self::ExtraIn,
-    ) -> (M::SolutionType, M::ScoreType, Self::ExtraOut)
+    ) -> (M::SolutionType, M::ScoreType)
     where
         F: OptCallbackFn<M::SolutionType, M::ScoreType>,
     {
@@ -82,7 +78,6 @@ impl<M: OptModel<ScoreType = NotNan<f64>>> LocalSearchOptimizer<M> for LogisticA
             n_iter,
             time_limit,
             callback,
-            _extra_in,
         )
     }
 }

src/optim/relative_annealing.rs

@@ -44,9 +44,6 @@ impl RelativeAnnealingOptimizer {
 }
 
 impl<M: OptModel<ScoreType = NotNan<f64>>> LocalSearchOptimizer<M> for RelativeAnnealingOptimizer {
-    type ExtraIn = ();
-    type ExtraOut = ();
-
     /// Start optimization
     ///
     /// - `model` : the model to optimize
@@ -64,8 +61,7 @@ impl<M: OptModel<ScoreType = NotNan<f64>>> LocalSearchOptimizer<M> for RelativeA
         n_iter: usize,
         time_limit: Duration,
         callback: Option<&F>,
-        _extra_in: Self::ExtraIn,
-    ) -> (M::SolutionType, M::ScoreType, Self::ExtraOut)
+    ) -> (M::SolutionType, M::ScoreType)
     where
         F: OptCallbackFn<M::SolutionType, M::ScoreType>,
     {
@@ -83,7 +79,6 @@ impl<M: OptModel<ScoreType = NotNan<f64>>> LocalSearchOptimizer<M> for RelativeA
             n_iter,
             time_limit,
             callback,
-            _extra_in,
         )
     }
 }