Version upgrade to 1.0.0

Finally completed random SVD integration

Added random SVD tests with lanczos comparison

Author: Ian

Cargo.lock

@@ -4,13 +4,13 @@ description = "A Rust port of LAS2 from SVDLIBC"
 keywords = ["svd"]
 categories = ["algorithms", "data-structures", "mathematics", "science"]
 name = "single-svdlib"
-version = "0.8.0"
+version = "1.0.0"
 edition = "2021"
 license-file = "SVDLIBC-LICENSE.txt"
 
 [dependencies]
 anyhow = "1.0.97"
-argmin = { version = "0.10.0", features = ["rayon"] }
+#argmin = { version = "0.10.0", features = ["rayon"] }
 nalgebra-sparse = "0.10.0"
 num-traits = "0.2.19"
 rand = "0.9.0"

Cargo.toml

@@ -168,7 +168,7 @@ where
     Ok(SvdRec {
         // Dimensionality (number of Ut,Vt rows & length of S)
         d: r.d,
-        ut: Array2::from_shape_vec((r.d, r.Ut.cols), r.Ut.value)?,
+        u: Array2::from_shape_vec((r.d, r.Ut.cols), r.Ut.value)?,
         s: Array::from_shape_vec(r.d, r.S)?,
         vt: Array2::from_shape_vec((r.d, r.Vt.cols), r.Vt.value)?,
         diagnostics: Diagnostics {
@@ -1402,7 +1402,7 @@ fn lanso<T: SvdFloat>(
 impl<T: SvdFloat + 'static> SvdRec<T> {
     pub fn recompose(&self) -> Array2<T> {
         let sdiag = Array2::from_diag(&self.s);
-        self.ut.t().dot(&sdiag).dot(&self.vt)
+        self.u.dot(&sdiag).dot(&self.vt)
     }
 }
 

src/lanczos/mod.rs

@@ -166,20 +166,18 @@ mod simple_comparison_tests {
     #[test]
     fn test_random_svd_computation() {
 
-        // Create a matrix with high sparsity (99%)
         let test_matrix = create_sparse_matrix(1000, 250, 0.01); // 1% non-zeros
 
-        // Convert to CSR for processing
         let csr = CsrMatrix::from(&test_matrix);
 
-        // Run randomized SVD with reasonable defaults for a sparse matrix
         let result = randomized::randomized_svd(
             &csr,
-            50,                              // target rank
-            10,                              // oversampling parameter
-            3,                               // power iterations
-            randomized::PowerIterationNormalizer::QR,    // use QR normalization
-            Some(42),                        // random seed
+            50,
+            10,
+            3,
+            randomized::PowerIterationNormalizer::QR,
+            false,
+            Some(42),
         );
 
         // Verify the computation succeeds on a highly sparse matrix
@@ -206,8 +204,8 @@ mod simple_comparison_tests {
             }
 
             // Verify basics of U and V dimensions
-            assert_eq!(svd_result.ut.nrows(), 50, "U transpose should have 50 rows");
-            assert_eq!(svd_result.ut.ncols(), 1000, "U transpose should have 1000 columns");
+            assert_eq!(svd_result.u.nrows(), 50, "U transpose should have 50 rows");
+            assert_eq!(svd_result.u.ncols(), 1000, "U transpose should have 1000 columns");
             assert_eq!(svd_result.vt.nrows(), 50, "V transpose should have 50 rows");
             assert_eq!(svd_result.vt.ncols(), 250, "V transpose should have 250 columns");
 
@@ -232,6 +230,7 @@ mod simple_comparison_tests {
                 10,                              // oversampling parameter
                 7,                               // power iterations
                 randomized::PowerIterationNormalizer::QR,    // use QR normalization
+                false,
                 Some(42),                        // random seed
             )
         });
@@ -263,6 +262,7 @@ mod simple_comparison_tests {
                 10,                              // oversampling parameter
                 2,                               // power iterations
                 randomized::PowerIterationNormalizer::QR,    // use QR normalization
+                false,
                 Some(42),                        // random seed
             )
         });