Add duplicate argument in sample method (#893)

Author: ishii-norimi

lib/util/matrix.js

@@ -652,18 +652,28 @@ export default class Matrix {
 	 * Returns a matrix that sampled along the axis.
 	 * @param {number} n Sampled size
 	 * @param {number} [axis] Axis to be sampled
+	 * @param {number} [duplicate] Allow duplicate index or not
 	 * @returns {[Matrix, number[]]} Sampled matrix and its original indexes
 	 */
-	sample(n, axis = 0) {
+	sample(n, axis = 0, duplicate = false) {
 		const k = this.sizes[axis]
 		const idx = []
-		for (let i = 0; i < n; i++) {
-			idx.push(Math.floor(Math.random() * (k - i)))
-		}
-		for (let i = n - 1; i >= 0; i--) {
-			for (let j = n - 1; j > i; j--) {
-				if (idx[i] <= idx[j]) {
-					idx[j]++
+		if (duplicate) {
+			for (let i = 0; i < n; i++) {
+				idx.push(Math.floor(Math.random() * k))
+			}
+		} else {
+			if (n > k) {
+				throw new MatrixException('Invalid sampled size.')
+			}
+			for (let i = 0; i < n; i++) {
+				idx.push(Math.floor(Math.random() * (k - i)))
+			}
+			for (let i = n - 1; i >= 0; i--) {
+				for (let j = n - 1; j > i; j--) {
+					if (idx[i] <= idx[j]) {
+						idx[j]++
+					}
 				}
 			}
 		}

tests/lib/util/matrix.test.js

@@ -989,6 +989,29 @@ describe('Matrix', () => {
 			expect(expidx).toEqual(idx)
 		})
 
+		test.each([undefined, 0])('row(%p) duplicate index', axis => {
+			const n = 6
+			const org = Matrix.randn(3, 5)
+			const [mat, idx] = org.sample(n, axis, true)
+			expect(idx).toHaveLength(n)
+
+			const expidx = []
+			for (let k = 0; k < n; k++) {
+				for (let i = 0; i < org.rows; i++) {
+					let flg = true
+					for (let j = 0; j < org.cols; j++) {
+						flg &= mat.at(k, j) === org.at(i, j)
+					}
+					if (flg) {
+						expidx.push(i)
+						break
+					}
+				}
+			}
+			expect(expidx).toHaveLength(n)
+			expect(expidx).toEqual(idx)
+		})
+
 		test('col index', () => {
 			const n = 3
 			const org = Matrix.randn(10, 5)
@@ -1012,6 +1035,34 @@ describe('Matrix', () => {
 			expect(expidx).toEqual(idx)
 		})
 
+		test('col duplicate index', () => {
+			const n = 6
+			const org = Matrix.randn(3, 5)
+			const [mat, idx] = org.sample(n, 1, true)
+			expect(idx).toHaveLength(n)
+
+			const expidx = []
+			for (let k = 0; k < n; k++) {
+				for (let j = 0; j < org.cols; j++) {
+					let flg = true
+					for (let i = 0; i < org.rows; i++) {
+						flg &= mat.at(i, k) === org.at(i, j)
+					}
+					if (flg) {
+						expidx.push(j)
+						break
+					}
+				}
+			}
+			expect(expidx).toHaveLength(n)
+			expect(expidx).toEqual(idx)
+		})
+
+		test('fail invalid sampled size %p', () => {
+			const mat = Matrix.randn(5, 10)
+			expect(() => mat.sample(6, 0)).toThrow('Invalid sampled size.')
+		})
+
 		test.each([-1, 2])('fail invalid axis %p', axis => {
 			const mat = Matrix.randn(5, 10)
 			expect(() => mat.sample(4, axis)).toThrow('Invalid axis.')