small improvements

noxfile.py

@@ -3,6 +3,6 @@
 
 @nox.session
 def tests(session):
-    session.install("pip", "numpy", "pytest", "scipy")
+    session.install("pip", "numpy", "pytest", "pytest-benchmark" "scipy")
     session.run("pip", "install", ".", "-v")
     session.run("pytest")

src/emd_classification.rs

@@ -27,13 +27,19 @@ fn argsort<T: Ord>(data: &[T]) -> Vec<usize> {
 /// * `v1` - A 1-dimensional view of f64 data.
 /// * `v2` - A 1-dimensional view of f64 data.
 ///
+/// # Panics
+/// Panics if the input arrays `v1` and `v2` have different lengths.
+///
 /// # Returns
 /// The Euclidean distance as a floating-point number.
 fn euclidean_distance(v1: &ArrayView1<f64>, v2: &ArrayView1<f64>) -> f64 {
-    v1.iter()
-        .zip(v2.iter())
-        .map(|(&x, &y)| (x - y).powi(2))
-        .sum::<f64>()
+    if v1.len() != v2.len() {
+        panic!("Input arrays must have the same length");
+    }
+    Zip::from(v1)
+        .and(v2)
+        .map_collect(|&x, &y| (x - y).powi(2))
+        .sum()
         .sqrt()
 }
 
@@ -49,6 +55,9 @@ fn euclidean_rdist_row(
     x: &ArrayView1<'_, f64>,
     y: &ArrayView2<'_, f64>,
 ) -> Array1<OrderedFloat<f64>> {
+    if x.is_empty() || y.is_empty() {
+        panic!("Input arrays must not be empty");
+    }
     Zip::from(y.rows()).map_collect(|row| OrderedFloat::from(euclidean_distance(&row, &x)))
 }
 
@@ -104,7 +113,7 @@ pub fn compute_emd_between_2dtensors(
 ) -> 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 (emd_dist, _assignments) = kuhn_munkres_min(&weights);
+    let (emd_dist, _) = kuhn_munkres_min(&weights);
     Ok(emd_dist)
 }
 
@@ -122,7 +131,7 @@ pub fn compute_emd_between_2dtensors_par(
 ) -> Result<OrderedFloat<f64>, MatrixFormatError> {
     let costs = euclidean_rdist_par(x, y);
     let weights = Matrix::from_vec(costs.nrows(), costs.ncols(), costs.into_raw_vec())?;
-    let (emd_dist, _assignments) = kuhn_munkres_min(&weights);
+    let (emd_dist, _) = kuhn_munkres_min(&weights);
     Ok(emd_dist)
 }
 

src/lib.rs

@@ -51,9 +51,10 @@ fn compute_emd<'py>(x: PyReadonlyArray2<'py, f64>, y: PyReadonlyArray2<'py, f64>
 
     match z {
         Ok(z) => Ok(*z),
-        Err(_e) => Err(exceptions::PyTypeError::new_err(
-            "Failed to compute EMD distance.",
-        )),
+        Err(e) => Err(exceptions::PyTypeError::new_err(format!(
+            "Failed to compute EMD distance: {}",
+            e
+        ))),
     }
 }
 
@@ -68,9 +69,10 @@ fn compute_emd_parallel<'py>(
 
     match z {
         Ok(z) => Ok(*z),
-        Err(_e) => Err(exceptions::PyTypeError::new_err(
-            "Failed to compute EMD distance.",
-        )),
+        Err(e) => Err(exceptions::PyTypeError::new_err(format!(
+            "Failed to compute EMD distance: {}",
+            e
+        ))),
     }
 }
 

tests/test_benchmark.py

@@ -8,20 +8,16 @@
 
 def compute_earth_mover_dist(first, second):
     """Compute earth's mover distance (EMD) between two data tensors using numpy and scipy."""
-    emds = []
-    for el in second:
-        d = cdist(first, el)
-        row_ind, col_ind = linear_sum_assignment(d)
-        emd = d[row_ind, col_ind].sum()
-        emds.append(emd)
-    return emds
+    d = cdist(first, second)
+    row_ind, col_ind = linear_sum_assignment(d)
+    return d[row_ind, col_ind].sum()
 
 
 def setup_data():
     """Generates random data for benchmarking."""
     rng = np.random.default_rng()
     data1 = rng.random((50, 50))
-    data2 = rng.random((10, 50, 50))
+    data2 = rng.random((50, 50))
     return data1, data2