v0.24.4 patch (#57)

* add to arrowspace config. bump version
* minor changes

Author: Mec-iS

Cargo.lock

@@ -1,6 +1,6 @@
 [package]
 name = "arrowspace"
-version = "0.24.3"
+version = "0.24.4"
 edition = "2024"
 description = "Spectral vector search with taumode (λτ) indexing"
 authors = ["Lorenzo <[email protected]>"]
@@ -26,7 +26,7 @@ exclude = [
 
 [dependencies]
 rand = "0.9.2"
-smartcore = {version = "^0.4.5"}
+smartcore = {version = "^0.4.7"}
 ordered-float = "5.1.0"
 rayon = "1.11.0"
 sprs = "0.11.4"
@@ -48,7 +48,7 @@ csv = { version = "1.4.0", optional = true }
 
 [dev-dependencies]
 criterion = "0.7.0"
-smartcore = {version = "^0.4.5", features = ["datasets"]}
+smartcore = {version = "^0.4.7", features = ["datasets"]}
 rand_pcg = "0.9.0"
 tempfile = "3.23.0"
 rand_xoshiro = "0.7.0"

Cargo.toml

@@ -1282,7 +1282,7 @@ impl ConfigValue {
     pub fn as_bool(&self) -> Option<bool> {
         match self {
             ConfigValue::Bool(v) => Some(*v),
-            _ => None,
+            _ => panic!("called as_bool but it is not"),
         }
     }
 
@@ -1296,21 +1296,22 @@ impl ConfigValue {
                     None
                 }
             }
-            _ => None,
+            _ => panic!("called as_usize but it is not"),
         }
     }
 
     pub fn as_f64(&self) -> Option<f64> {
         match self {
-            ConfigValue::F64(v) => Some(*v),
-            ConfigValue::OptionF64(v) => {
-                if v.is_some() {
-                    Some(v.unwrap())
-                } else {
-                    None
-                }
-            }
-            _ => None,
+            ConfigValue::F64(v) => match v {
+                val if val.is_nan() => Some(-1.0),
+                val => Some(*val),
+            },
+            ConfigValue::OptionF64(v) => match v {
+                Some(val) if val.is_nan() => Some(-1.0),
+                Some(val) => Some(*val),
+                None => None,
+            },
+            _ => panic!("called as_f64 but it is not"),
         }
     }
 
@@ -1324,7 +1325,7 @@ impl ConfigValue {
                     None
                 }
             }
-            _ => None,
+            _ => panic!("called as_u64 but it is not"),
         }
     }
 
@@ -1400,6 +1401,10 @@ impl ArrowSpaceBuilder {
             "use_dims_reduction".to_string(),
             ConfigValue::Bool(self.use_dims_reduction),
         );
+        config.insert(
+            "extra_dims_reduction".to_string(),
+            ConfigValue::Bool(self.extra_dims_reduction),
+        );
         config.insert("rp_eps".to_string(), ConfigValue::F64(self.rp_eps));
 
         config

src/builder.rs

@@ -416,9 +416,9 @@ impl Default for ArrowSpace {
             lambdas: Vec::new(),
             lambdas_sorted: SortedLambdas::new(),
             // lambdas normalisation
-            min_lambdas: f64::NAN,
-            max_lambdas: f64::NAN,
-            range_lambdas: f64::NAN,
+            min_lambdas: -1.0,
+            max_lambdas: -1.0,
+            range_lambdas: -1.0,
             // enable synthetic λ with Median τ by default
             taumode: TAUDEFAULT,
             // Clustering defaults
@@ -457,9 +457,9 @@ impl ArrowSpace {
             lambdas: vec![0.0; n_items],        // will be computed later
             lambdas_sorted: SortedLambdas::new(),
             // lambdas normalisation
-            min_lambdas: f64::NAN,
-            max_lambdas: f64::NAN,
-            range_lambdas: f64::NAN,
+            min_lambdas: -1.0,
+            max_lambdas: -1.0,
+            range_lambdas: -1.0,
             taumode,
             // Clustering defaults
             n_clusters: 0,
@@ -634,9 +634,9 @@ impl ArrowSpace {
             lambdas,
             lambdas_sorted,
             // Normalization fields
-            min_lambdas: f64::NAN,
-            max_lambdas: f64::NAN,
-            range_lambdas: f64::NAN,
+            min_lambdas: -1.0,
+            max_lambdas: -1.0,
+            range_lambdas: -1.0,
             taumode,
             n_clusters,
             cluster_assignments: Vec::new(),
@@ -689,9 +689,9 @@ impl ArrowSpace {
             lambdas: vec![0.0; n_items],        // will be computed later
             lambdas_sorted: SortedLambdas::new(),
             // lambdas normalisation
-            min_lambdas: f64::NAN,
-            max_lambdas: f64::NAN,
-            range_lambdas: f64::NAN,
+            min_lambdas: -1.0,
+            max_lambdas: -1.0,
+            range_lambdas: -1.0,
             taumode,
             // Clustering defaults
             n_clusters: 0,
@@ -763,9 +763,9 @@ impl ArrowSpace {
             lambdas: vec![0.0; n_items],        // will be computed later
             lambdas_sorted: SortedLambdas::new(),
             // lambdas normalisation
-            min_lambdas: f64::NAN,
-            max_lambdas: f64::NAN,
-            range_lambdas: f64::NAN,
+            min_lambdas: -1.0,
+            max_lambdas: -1.0,
+            range_lambdas: -1.0,
             taumode: TAUDEFAULT,
             // Clustering defaults
             n_clusters: 0,
@@ -807,9 +807,9 @@ impl ArrowSpace {
             signals: sprs::CsMat::zero((0, 0)),
             lambdas: vec![0.0; nitems],
             lambdas_sorted: SortedLambdas::new(),
-            min_lambdas: f64::NAN,
-            max_lambdas: f64::NAN,
-            range_lambdas: f64::NAN,
+            min_lambdas: -1.0,
+            max_lambdas: -1.0,
+            range_lambdas: -1.0,
             taumode: TAUDEFAULT,
             n_clusters: 0,
             cluster_assignments: Vec::new(),
@@ -843,14 +843,14 @@ impl ArrowSpace {
         );
 
         if let Some(ref proj) = self.projection_matrix {
-            debug!(
+            trace!(
                 "Projecting query: {} → {} dimensions using seed-based projection",
                 self.nfeatures,
                 self.reduced_dim.unwrap()
             );
             proj.project(query)
         } else {
-            debug!("No projection applied, returning original query");
+            trace!("No projection applied, returning original query");
             query.to_vec()
         }
     }
@@ -894,7 +894,7 @@ impl ArrowSpace {
 
             let lambda = sc_lambdas[best_idx];
 
-            debug!(
+            trace!(
                 "Query mapped to subcentroid {}/{} with λ={:.6} (dist={:.4})",
                 best_idx,
                 subcentroids.shape().0,
@@ -1447,6 +1447,19 @@ impl ArrowSpace {
         config.insert("nitems".to_string(), ConfigValue::Usize(self.nitems));
         config.insert("nfeatures".to_string(), ConfigValue::Usize(self.nfeatures));
 
+        config.insert(
+            "min_lambdas".to_string(),
+            ConfigValue::F64(self.min_lambdas),
+        );
+        config.insert(
+            "max_lambdas".to_string(),
+            ConfigValue::F64(self.max_lambdas),
+        );
+        config.insert(
+            "range_lambdas".to_string(),
+            ConfigValue::F64(self.range_lambdas),
+        );
+
         // projection matrix
         if self.projection_matrix.is_some() {
             config.insert(
@@ -1497,6 +1510,19 @@ impl ArrowSpace {
             ConfigValue::F64(self.cluster_radius),
         );
 
+        config.insert(
+            "min_lambdas".to_string(),
+            ConfigValue::F64(self.min_lambdas),
+        );
+        config.insert(
+            "max_lambdas".to_string(),
+            ConfigValue::F64(self.max_lambdas),
+        );
+        config.insert(
+            "range_lambdas".to_string(),
+            ConfigValue::F64(self.range_lambdas),
+        );
+
         config
     }
 }