1 files changed, 29 insertions, 29 deletions

diff --git a/src/grossmeister/search.rs b/src/grossmeister/search.rs
index 378349c..d0fddbb 100644
--- a/src/grossmeister/search.rs
+++ b/src/grossmeister/search.rs
@@ -246,43 +246,46 @@ impl Grossmeister {
         pv
     }
 
+    /// Memory-enhanced Test Driver
+    /// Given a guess score (from previous iteration), use it for series of small-window searches
+    fn mtdf(&mut self, score_guess: Score, depth: u8) -> Score {
+        let window_size = 1.00;
+        let mut score = score_guess;
+        let mut lowerbound = -INFINITY;
+        let mut upperbound = INFINITY;
+
+        while lowerbound < upperbound {
+            let beta = score + if score == lowerbound {
+                window_size
+            } else {
+                0.0
+            };
+
+            score = self.negamax_search(beta - window_size, beta, depth, 0);
+
+            if score < beta {
+                upperbound = score
+            } else {
+                lowerbound = score
+            }
+        }
+
+        score
+    }
+
     pub fn iterative_deepening(&mut self, max_depth: u8) -> (Score, Vec<Move>) {
         let mut best_score = 0.0;
         let mut pv = Vec::new();
-
-        let mut depth = 1;
-        let mut alpha = -INFINITY;
-        let mut beta = INFINITY;
-        let window_size = 0.25;
-        let mut gradual_widening_counter = 0;
+        let mut depth = 0;
 
         while depth <= max_depth {
-            if self.debug {
-                println!("info string window {:?}", (alpha, beta));
-            }
-
-            let score = self.negamax_search(alpha, beta, depth, 0);
+            let score = self.mtdf(best_score, depth);
 
             if self.should_halt.load(std::sync::atomic::Ordering::SeqCst) {
                 println!("info string halting search");
                 break;
             }
 
-            if score <= alpha { // Alpha-cutoff
-                gradual_widening_counter += 1;
-                beta = alpha + window_size * 0.1;
-                alpha = score - window_size * 2.0f32.powi(gradual_widening_counter);
-                println!("info depth {} score upperbound {:.0}", depth, beta * 100.0);
-                continue;
-            }
-            if score >= beta { // Beta-cutoff
-                gradual_widening_counter += 1;
-                alpha = beta - window_size * 0.1;
-                beta = score + window_size * 2.0f32.powi(gradual_widening_counter);
-                println!("info depth {} score lowerbound {:.0}", depth, alpha * 100.0);
-                continue;
-            }
-
             best_score = score;
             pv = self.reconstruct_pv(depth);
 
@@ -305,9 +308,6 @@ impl Grossmeister {
             println!();
 
             depth += 1;
-            gradual_widening_counter = 0;
-            alpha = score - window_size;
-            beta = score + window_size;
 
             // If our score is mate in N, we break at depth N
             if depth as Score >= SCORE_MATE - score.abs() {