Replace multiplication with addition and refactor function signature

kgryte · kgryte · commit 2fbf296b848f · 2018-07-11T04:22:21.000+01:00
diff --git a/lib/node_modules/@stdlib/ml/incr/kmeans/lib/init_kmeansplusplus.js b/lib/node_modules/@stdlib/ml/incr/kmeans/lib/init_kmeansplusplus.js
@@ -41,25 +41,27 @@ var squaredCorrelation = require( './squared_correlation.js' );
 * @param {Array} out - output array
 * @param {Function} dist - distance function to apply
 * @param {PositiveInteger} npts - number of data points
+* @param {PositiveInteger} ndims - number of dimensions
 * @param {ndarray} matrix - data point matrix
 * @param {NonNegativeInteger} ci - centroid row index
 * @returns {Array} output array
 */
-function dapply( out, dist, npts, matrix, ci ) {
+function dapply( out, dist, npts, ndims, matrix, ci ) {
 	var offsetC;
-	var stride;
-	var offset;
+	var offsetD;
+	var strideD;
 	var buf;
-	var N;
 	var i;
 
 	buf = matrix.data;
-	N = matrix.shape[ 1 ];
-	stride = matrix.strides[ 0 ];
-	offsetC = stride * ci;
+
+	strideD = matrix.strides[ 0 ];
+	offsetC = strideD * ci;
+	offsetD = 0;
+
 	for ( i = 0; i < npts; i++ ) {
-		offset = stride * i;
-		out[ i ] = dist( N, buf, 1, offset, buf, 1, offsetC ); // Magic number `1` for stride is based on knowing that the matrix is row-major single-segment contiguous
+		out[ i ] = dist( ndims, buf, 1, offsetD, buf, 1, offsetC ); // Magic number `1` for stride is based on knowing that the matrix is row-major single-segment contiguous
+		offsetD += strideD;
 	}
 	return out;
 }
@@ -203,7 +205,7 @@ function kmeansplusplus( out, buffer, metric, normalize, trials, seed ) {
 	// 2-5. For each data point, compute the distances to each centroid, find the closest centroid, and, based on the distance to the closest centroid, assign a probability to the data point to be chosen as centroid `c_j`...
 	for ( j = 1; j < k; j++ ) {
 		// Note: instead of repeatedly computing centroid distances for each data point, we only need to compute the distances for the most recent centroid and to maintain a hash of closest distance results...
-		dapply( d2, dist, npts, buffer, centroids[ j-1 ] );
+		dapply( d2, dist, npts, ndims, buffer, centroids[ j-1 ] );
 		csum = 0.0; // total cumulative distance
 		for ( i = 0; i < npts; i++ ) {
 			ind = 2 * i;
