no mem leaks

HannaMeyer · Jan 7, 2025 · f2456fb · f2456fb
1 parent 0076023
commit f2456fb
Show file tree

Hide file tree

Showing 5 changed files with 15 additions and 15 deletions.
diff --git a/R/aoa.R b/R/aoa.R
@@ -452,7 +452,7 @@ aoa <- function(newdata,
             maxLPD = maxLPD) {
     if (method == "L2") {
       # Euclidean Distance
-      return(FNN::knnx.dist(reference, point, k = maxLPD))
+      return(FNN::knnx.dist(reference, point, k = maxLPD, algorithm = "brute"))
     } else if (method == "MD") {
       return(t(sapply(1:dim(point)[1],
                       function(y)
@@ -470,7 +470,7 @@ aoa <- function(newdata,
             maxLPD = maxLPD) {
     if (method == "L2") {
       # Euclidean Distance
-      return(FNN::knnx.index(reference, point, k = maxLPD))
+      return(FNN::knnx.index(reference, point, k = maxLPD, algorithm = "brute"))
     } else if (method == "MD") {
       stop("MD currently not implemented for LPD")
     }

diff --git a/R/geodist.R b/R/geodist.R
@@ -301,7 +301,7 @@ sample2sample <- function(x, type,variables,time_unit,timevar, catVars){
     for (i in 1:nrow(x_clean)){
 
       if(is.null(catVars)) {
-        trainDist <-  FNN::knnx.dist(x_clean[i,],x_clean,k=1)
+        trainDist <-  FNN::knnx.dist(x_clean[i,],x_clean,k=1, algorithm = "brute")
       } else {
         trainDist <- gower::gower_dist(x_clean[i,],x_clean)
       }
@@ -378,7 +378,7 @@ sample2prediction = function(x, modeldomain, type, samplesize,variables,time_uni
     for (i in 1:nrow(modeldomain)){
 
       if(is.null(catVars)) {
-        trainDist <-  FNN::knnx.dist(modeldomain[i,],x_clean,k=1)
+        trainDist <-  FNN::knnx.dist(modeldomain[i,],x_clean,k=1, algorithm = "brute")
       } else {
         trainDist <- gower::gower_dist(modeldomain[i,], x_clean)
       }
@@ -460,7 +460,7 @@ sample2test <- function(x, testdata, type,variables,time_unit,timevar, catVars){
     for (i in 1:nrow(testdata)){
 
       if(is.null(catVars)) {
-        testDist <- FNN::knnx.dist(testdata[i,],x_clean,k=1)
+        testDist <- FNN::knnx.dist(testdata[i,],x_clean,k=1, algorithm = "brute")
       } else {
         testDist <- gower::gower_dist(testdata[i,], x_clean)
       }
@@ -550,7 +550,7 @@ cvdistance <- function(x, cvfolds, cvtrain, type, variables,time_unit,timevar, c
       for (k in 1:nrow(testdata_i)){
 
         if(is.null(catVars)) {
-          trainDist <-  tryCatch(FNN::knnx.dist(testdata_i[k,],traindata_i,k=1),
+          trainDist <-  tryCatch(FNN::knnx.dist(testdata_i[k,],traindata_i,k=1, algorithm = "brute"),
                                  error = function(e)e)
           if(inherits(trainDist, "error")){
             trainDist <- NA

diff --git a/R/knndm.R b/R/knndm.R
@@ -392,9 +392,9 @@ knndm_geo <- function(tpoints, predpoints, k, maxp, clustering, linkf, islonglat
     units(Gij) <- NULL
     Gij <- apply(Gij, 1, min)
   }else{
-    Gj <- c(FNN::knn.dist(tcoords, k = 1))
+    Gj <- c(FNN::knn.dist(tcoords, k = 1, algorithm = "brute"))
     Gij <- c(FNN::knnx.dist(query = sf::st_coordinates(predpoints)[,1:2],
-                            data = tcoords, k = 1))
+                            data = tcoords, k = 1, algorithm = "brute"))
   }
 
   # Check if Gj > Gij (warning suppressed regarding ties)
@@ -576,8 +576,8 @@ knndm_feature <- function(tpoints, predpoints, k, maxp, clustering, linkf, catVa
 
     } else {
       # use FNN with Euclidean distances if no categorical variables are present
-      Gj <- c(FNN::knn.dist(tpoints, k = 1))
-      Gij <- c(FNN::knnx.dist(query = predpoints, data = tpoints, k = 1))
+      Gj <- c(FNN::knn.dist(tpoints, k = 1, algorithm = "brute"))
+      Gij <- c(FNN::knnx.dist(query = predpoints, data = tpoints, k = 1, algorithm = "brute"))
     }
 
 
@@ -793,7 +793,7 @@ distclust_euclidean <- function(tr_coords, folds){
   alldist <- rep(NA, length(folds))
   for(f in unique(folds)){
     alldist[f == folds] <- c(FNN::knnx.dist(query = tr_coords[f == folds,,drop=FALSE],
-                                            data = tr_coords[f != folds,,drop=FALSE], k = 1))
+                                            data = tr_coords[f != folds,,drop=FALSE], k = 1, algorithm = "brute"))
   }
   alldist
 }

diff --git a/R/nndm.R b/R/nndm.R
@@ -350,7 +350,7 @@ nndm <- function(tpoints, modeldomain = NULL, predpoints = NULL,
     if(is.null(catVars)) {
 
       # Euclidean distances if no categorical variables are present
-      Gij <- c(FNN::knnx.dist(query = predpoints, data = tpoints, k = 1))
+      Gij <- c(FNN::knnx.dist(query = predpoints, data = tpoints, k = 1, algorithm = "brute"))
       tdist <- as.matrix(stats::dist(tpoints, upper = TRUE))
       diag(tdist) <- NA
       Gj <- apply(tdist, 1, function(x) min(x, na.rm=TRUE))

diff --git a/R/trainDI.R b/R/trainDI.R
@@ -521,7 +521,7 @@ aoa_get_variables <- function(variables, model, train){
 .mindistfun <- function(point, reference, method, S_inv=NULL){
 
   if (method == "L2"){ # Euclidean Distance
-    return(c(FNN::knnx.dist(reference, point, k = 1)))
+    return(c(FNN::knnx.dist(reference, point, k = 1, algorithm = "brute")))
   } else if (method == "MD"){ # Mahalanobis Distance
     return(sapply(1:dim(point)[1],
                   function(y) min(sapply(1:dim(reference)[1],
@@ -533,9 +533,9 @@ aoa_get_variables <- function(variables, model, train){
 
   if (method == "L2"){ # Euclidean Distance
     if(sorted){
-      return(FNN::knnx.dist(reference, point, k = dim(reference)[1]))
+      return(FNN::knnx.dist(reference, point, k = dim(reference)[1], algorithm = "brute"))
     } else {
-      return(FNN::knnx.dist(point,reference,k=1))
+      return(FNN::knnx.dist(point,reference,k=1, algorithm = "brute"))
     }
   } else if (method == "MD"){ # Mahalanobis Distance
     if(sorted){