testing

Author: Giorgio Savastano

Cargo.lock

@@ -11,7 +11,7 @@ crate-type = ["cdylib"]
 pyo3 = { version = "0.21.2", features = ["extension-module", "multiple-pymethods"] }
 ndarray = { version = "0.15.6", features = ["rayon"] }
 numpy = "0.21.0"
-pathfinding = "4.9.1"
+pathfinding = "4.10.0"
 ordered-float = "4.2.0"
 
 [build-dependencies]

Cargo.toml

@@ -109,8 +109,18 @@ pub fn compute_emd_between_2dtensors(
     x: ArrayView2<'_, f64>,
     y: ArrayView2<'_, f64>,
 ) -> Result<OrderedFloat<f64>, MatrixFormatError> {
-    let costs = euclidean_rdist_rust(x, y);
-    let weights = Matrix::from_vec(costs.nrows(), costs.ncols(), costs.into_raw_vec())?;
+    let x_rows = x.nrows();
+    let y_rows = y.nrows();
+
+    let mut costs = Vec::with_capacity(x_rows * y_rows);
+
+    for row_x in x.rows() {
+        for row_y in y.rows() {
+            let dist = euclidean_distance(&row_x, &row_y);
+            costs.push(OrderedFloat::from(dist));
+        }
+    }
+    let weights = Matrix::from_vec(x_rows, y_rows, costs)?;
     let (emd_dist, _) = kuhn_munkres_min(&weights);
     Ok(emd_dist)
 }

src/emd_classification.rs

@@ -13,27 +13,63 @@ def compute_earth_mover_dist(first, second):
     return d[row_ind, col_ind].sum()
 
 
-def setup_data():
+def setup_data(rows, cols):
     """Generates random data for benchmarking."""
     rng = np.random.default_rng()
-    data1 = rng.random((50, 50))
-    data2 = rng.random((50, 50))
+    data1 = rng.random((rows, cols), dtype=np.float64)
+    data2 = rng.random((rows, cols), dtype=np.float64)
     return data1, data2
 
 
-def test_rust_emd_benchmark(benchmark):
-    """Benchmark the Rust-backed EMD calculation."""
-    data1, data2 = setup_data()
+def test_rust_emd_benchmark_small(benchmark):
+    """Benchmark the Rust-backed EMD calculation on small data."""
+    data1, data2 = setup_data(17, 11)
     benchmark(compute_earth_movers_distance_2d, data1, data2, False)
 
 
-def test_rust_emd_par_benchmark(benchmark):
-    """Benchmark the Rust-backed EMD calculation."""
-    data1, data2 = setup_data()
+def test_rust_emd_benchmark_medium(benchmark):
+    """Benchmark the Rust-backed EMD calculation on medium data."""
+    data1, data2 = setup_data(50, 50)
+    benchmark(compute_earth_movers_distance_2d, data1, data2, False)
+
+
+def test_rust_emd_benchmark_large(benchmark):
+    """Benchmark the Rust-backed EMD calculation on large data."""
+    data1, data2 = setup_data(100, 100)
+    benchmark(compute_earth_movers_distance_2d, data1, data2, False)
+
+
+def test_rust_emd_par_benchmark_small(benchmark):
+    """Benchmark the Rust-backed EMD calculation with parallel processing on small data."""
+    data1, data2 = setup_data(17, 11)
+    benchmark(compute_earth_movers_distance_2d, data1, data2, True)
+
+
+def test_rust_emd_par_benchmark_medium(benchmark):
+    """Benchmark the Rust-backed EMD calculation with parallel processing on medium data."""
+    data1, data2 = setup_data(50, 50)
     benchmark(compute_earth_movers_distance_2d, data1, data2, True)
 
 
-def test_numpy_emd_benchmark(benchmark):
-    """Benchmark the numpy/scipy EMD calculation."""
-    data1, data2 = setup_data()
+def test_rust_emd_par_benchmark_large(benchmark):
+    """Benchmark the Rust-backed EMD calculation with parallel processing on large data."""
+    data1, data2 = setup_data(100, 100)
+    benchmark(compute_earth_movers_distance_2d, data1, data2, True)
+
+
+def test_numpy_emd_benchmark_small(benchmark):
+    """Benchmark the numpy/scipy EMD calculation on small data."""
+    data1, data2 = setup_data(17, 11)
+    benchmark(compute_earth_mover_dist, data1, data2)
+
+
+def test_numpy_emd_benchmark_medium(benchmark):
+    """Benchmark the numpy/scipy EMD calculation on medium data."""
+    data1, data2 = setup_data(50, 50)
+    benchmark(compute_earth_mover_dist, data1, data2)
+
+
+def test_numpy_emd_benchmark_large(benchmark):
+    """Benchmark the numpy/scipy EMD calculation on large data."""
+    data1, data2 = setup_data(100, 100)
     benchmark(compute_earth_mover_dist, data1, data2)