Rework API to avoid awkward &Option

ego/src/egor_solver.rs

@@ -541,7 +541,7 @@ where
         yt: &ArrayBase<impl Data<Elem = f64>, Ix2>,
         init: bool,
         recluster: bool,
-        clustering: &Option<Clustering>,
+        clustering: Option<&Clustering>,
         model_name: &str,
     ) -> Box<dyn ClusteredSurrogate> {
         let mut builder = self.surrogate_builder.clone();
@@ -567,9 +567,9 @@ where
             );
             model
         } else {
-            let clustering = clustering.as_ref().unwrap().clone();
+            let clustering = clustering.unwrap();
             let model = builder
-                .train_on_clusters(&xt.view(), &yt.view(), &clustering)
+                .train_on_clusters(&xt.view(), &yt.view(), clustering)
                 .expect("GP training failure");
             model
         }
@@ -615,7 +615,7 @@ where
                             &yt.slice(s![.., k..k + 1]).to_owned(),
                             init && i == 0,
                             recluster,
-                            &clusterings[k],
+                            clusterings[k].as_ref(),
                             &name,
                         )
                     })

gp/src/parameters.rs

@@ -109,8 +109,8 @@ impl<F: Float, Mean: RegressionModel<F>, Corr: CorrelationModel<F>> GpValidParam
     }
 
     /// Get number of components used by PLS
-    pub fn kpls_dim(&self) -> &Option<usize> {
-        &self.kpls_dim
+    pub fn kpls_dim(&self) -> Option<&usize> {
+        self.kpls_dim.as_ref()
     }
 
     /// Get the number of internal optimization restart

gp/src/sparse_parameters.rs

@@ -86,8 +86,8 @@ impl<F: Float, Corr: CorrelationModel<F>> SgpValidParams<F, Corr> {
     }
 
     /// Get the number of components used by PLS
-    pub fn kpls_dim(&self) -> &Option<usize> {
-        &self.gp_params.kpls_dim
+    pub fn kpls_dim(&self) -> Option<&usize> {
+        self.gp_params.kpls_dim.as_ref()
     }
 
     /// Get starting theta value for optimization
@@ -121,8 +121,8 @@ impl<F: Float, Corr: CorrelationModel<F>> SgpValidParams<F, Corr> {
     }
 
     /// Get seed
-    pub fn seed(&self) -> &Option<u64> {
-        &self.seed
+    pub fn seed(&self) -> Option<&u64> {
+        self.seed.as_ref()
     }
 }
 

moe/src/clustering.rs

@@ -119,15 +119,14 @@ pub fn find_best_number_of_clusters<R: Rng + Clone>(
                 .ok();
 
             if let Some(gmm) = maybe_gmm {
-                let gmm = Box::new(gmm);
                 // Cross Validation
                 for (train, valid) in dataset.fold(5).into_iter() {
                     if let Ok(mixture) = GpMixParams::default()
                         .n_clusters(n_clusters)
                         .regression_spec(regression_spec)
                         .correlation_spec(correlation_spec)
                         .kpls_dim(kpls_dim)
-                        .gmm(Some(gmm.clone()))
+                        .gmm(gmm.clone())
                         .fit(&train)
                     {
                         let xytrain =

moe/src/gp_algorithm.rs

@@ -109,7 +109,7 @@ impl<R: Rng + SeedableRng + Clone> GpMixValidParams<f64, R> {
         let dataset = Dataset::from(training);
 
         let gmx = if self.gmx().is_some() {
-            *self.gmx().as_ref().unwrap().clone()
+            self.gmx().unwrap().clone()
         } else {
             trace!("GMM training...");
             let gmm = GaussianMixtureModel::params(n_clusters)

moe/src/gp_parameters.rs

@@ -38,9 +38,9 @@ pub struct GpMixValidParams<F: Float, R: Rng + Clone> {
     /// Number of GP hyperparameters optimization restarts
     n_start: usize,
     /// Gaussian Mixture model used to cluster
-    gmm: Option<Box<GaussianMixtureModel<F>>>,
+    gmm: Option<GaussianMixtureModel<F>>,
     /// GaussianMixture preset
-    gmx: Option<Box<GaussianMixture<F>>>,
+    gmx: Option<GaussianMixture<F>>,
     /// Random number generator
     rng: R,
 }
@@ -100,13 +100,13 @@ impl<F: Float, R: Rng + Clone> GpMixValidParams<F, R> {
 
     /// An optional gaussian mixture to be fitted to generate multivariate normal
     /// in turns used to cluster
-    pub fn gmm(&self) -> &Option<Box<GaussianMixtureModel<F>>> {
-        &self.gmm
+    pub fn gmm(&self) -> Option<&GaussianMixtureModel<F>> {
+        self.gmm.as_ref()
     }
 
     /// An optional multivariate normal used to cluster (take precedence over gmm)
-    pub fn gmx(&self) -> &Option<Box<GaussianMixture<F>>> {
-        &self.gmx
+    pub fn gmx(&self) -> Option<&GaussianMixture<F>> {
+        self.gmx.as_ref()
     }
 
     /// The random generator
@@ -236,8 +236,8 @@ impl<F: Float, R: Rng + Clone> GpMixParams<F, R> {
 
     #[doc(hidden)]
     /// Sets the gaussian mixture (used to find the optimal number of clusters)
-    pub(crate) fn gmm(mut self, gmm: Option<Box<GaussianMixtureModel<F>>>) -> Self {
-        self.0.gmm = gmm;
+    pub(crate) fn gmm(mut self, gmm: GaussianMixtureModel<F>) -> Self {
+        self.0.gmm = Some(gmm);
         self
     }
 
@@ -246,9 +246,7 @@ impl<F: Float, R: Rng + Clone> GpMixParams<F, R> {
     /// Warning: no consistency check is done on the given initialization data
     /// *Panic* if multivariate normal init data not sound
     pub fn gmx(mut self, weights: Array1<F>, means: Array2<F>, covariances: Array3<F>) -> Self {
-        self.0.gmx = Some(Box::new(
-            GaussianMixture::new(weights, means, covariances).unwrap(),
-        ));
+        self.0.gmx = Some(GaussianMixture::new(weights, means, covariances).unwrap());
         self
     }
 
@@ -262,8 +260,8 @@ impl<F: Float, R: Rng + Clone> GpMixParams<F, R> {
             kpls_dim: self.0.kpls_dim(),
             theta_tuning: self.0.theta_tuning().clone(),
             n_start: self.0.n_start(),
-            gmm: self.0.gmm().clone(),
-            gmx: self.0.gmx().clone(),
+            gmm: self.0.gmm().cloned(),
+            gmx: self.0.gmx().cloned(),
             rng,
         })
     }

moe/src/sgp_algorithm.rs

@@ -110,7 +110,7 @@ impl<R: Rng + SeedableRng + Clone> SparseGpMixtureValidParams<f64, R> {
         let dataset = Dataset::from(training);
 
         let gmx = if self.gmx().is_some() {
-            *self.gmx().as_ref().unwrap().clone()
+            self.gmx().unwrap().clone()
         } else {
             debug!("GMM training...");
             let gmm = GaussianMixtureModel::params(n_clusters)

moe/src/sgp_parameters.rs

@@ -42,9 +42,9 @@ pub struct SparseGpMixtureValidParams<F: Float, R: Rng + Clone> {
     /// Inducings
     inducings: Inducings<F>,
     /// Gaussian Mixture model used to cluster
-    gmm: Option<Box<GaussianMixtureModel<F>>>,
+    gmm: Option<GaussianMixtureModel<F>>,
     /// GaussianMixture preset
-    gmx: Option<Box<GaussianMixture<F>>>,
+    gmx: Option<GaussianMixture<F>>,
     /// Random number generator
     rng: R,
 }
@@ -116,13 +116,13 @@ impl<F: Float, R: Rng + Clone> SparseGpMixtureValidParams<F, R> {
 
     /// An optional gaussian mixture to be fitted to generate multivariate normal
     /// in turns used to cluster
-    pub fn gmm(&self) -> &Option<Box<GaussianMixtureModel<F>>> {
-        &self.gmm
+    pub fn gmm(&self) -> Option<&GaussianMixtureModel<F>> {
+        self.gmm.as_ref()
     }
 
     /// An optional multivariate normal used to cluster (take precedence over gmm)
-    pub fn gmx(&self) -> &Option<Box<GaussianMixture<F>>> {
-        &self.gmx
+    pub fn gmx(&self) -> Option<&GaussianMixture<F>> {
+        self.gmx.as_ref()
     }
 
     /// The random generator
@@ -270,9 +270,7 @@ impl<F: Float, R: Rng + SeedableRng + Clone> SparseGpMixtureParams<F, R> {
     /// Warning: no consistency check is done on the given initialization data
     /// *Panic* if multivariate normal init data not sound
     pub fn gmx(mut self, weights: Array1<F>, means: Array2<F>, covariances: Array3<F>) -> Self {
-        self.0.gmx = Some(Box::new(
-            GaussianMixture::new(weights, means, covariances).unwrap(),
-        ));
+        self.0.gmx = Some(GaussianMixture::new(weights, means, covariances).unwrap());
         self
     }
 
@@ -288,8 +286,8 @@ impl<F: Float, R: Rng + SeedableRng + Clone> SparseGpMixtureParams<F, R> {
             n_start: self.0.n_start(),
             sparse_method: self.0.sparse_method(),
             inducings: self.0.inducings().clone(),
-            gmm: self.0.gmm().clone(),
-            gmx: self.0.gmx().clone(),
+            gmm: self.0.gmm().cloned(),
+            gmx: self.0.gmx().cloned(),
             rng,
         })
     }