updated zoomerjoin_euclidean_join function

LinkOrgs/R/zoomerjoin_euclidean_join.R

@@ -3,21 +3,31 @@ zoomerjoin_euclidean_join <- function(
   y = NULL, embedy = NULL,
   MaxDist = NULL,
   AveMatchNumberPerAlias = NULL,
-  n_bands = 40L,
-  band_width = 5L,
-  r = 5L,
   ...
 ) {
 
-  stopifnot(
-    "AveMatchNumberPerAlias not supported for zoomerjoin" =
-      !is.null(AveMatchNumberPerAlias)
-  )
+  if(!is.null(AveMatchNumberPerAlias)){
+    warning("Setting AveMatchNumberPerAlias may result in long runtimes when using zoomerjoin")
+
+    threshold <- GetCalibratedDistThres(x = embedy,
+                                      y = embedx,
+                                      AveMatchNumberPerAlias = AveMatchNumberPerAlias)
+
+  } else {
+      threshold <- MaxDist
+  }
+
+    # Find zoomerjoin hyperparameters that xyz in time
+ hyperparameters <- euclidean_hyper_search(3*threshold, threshold, .01, .995)
+
+ n_bands <- hyperparameters$n_bands
+ band_width <- hyperparameters$band_width
+ r <- hyperparameters$r
 
-  match_table <- zoomerjoin::rust_p_norm_join(
+  match_table <- zoomerjoin:::rust_p_norm_join(
     a_mat = as.matrix(embedx),
     b_mat = as.matrix(embedy),
-    radius = MaxDist,
+    radius = threshold,
     band_width = band_width,
     n_bands = n_bands,
     r = r,
@@ -29,5 +39,5 @@ zoomerjoin_euclidean_join <- function(
   colnames(x) <- paste(colnames(x), ".x", sep = "")
 
   z <- cbind(x[match_table[,1],], y[match_table[,2],])
-  return(DeconflictNames(z))
+  return(z)
 }

LinkOrgs/R/zoomerjoin_hyperparameter_searches.R

@@ -63,7 +63,7 @@ euclidean_hyper_search <- function(d1, d2, p1, p2,
     nvars = 3,
     max = FALSE,
     Domains = matrix(
-      c(1, 2000, 1, 500, 0, d1),
+      c(1, 5000, 1, 5000, 0, d1),
       ncol = 2,
       byrow = TRUE
     ),