@tensorflow/tfjs-core
Version:
Hardware-accelerated JavaScript library for machine intelligence
297 lines • 15.7 kB
JavaScript
"use strict";
var __awaiter = (this && this.__awaiter) || function (thisArg, _arguments, P, generator) {
return new (P || (P = Promise))(function (resolve, reject) {
function fulfilled(value) { try { step(generator.next(value)); } catch (e) { reject(e); } }
function rejected(value) { try { step(generator["throw"](value)); } catch (e) { reject(e); } }
function step(result) { result.done ? resolve(result.value) : new P(function (resolve) { resolve(result.value); }).then(fulfilled, rejected); }
step((generator = generator.apply(thisArg, _arguments || [])).next());
});
};
var __generator = (this && this.__generator) || function (thisArg, body) {
var _ = { label: 0, sent: function() { if (t[0] & 1) throw t[1]; return t[1]; }, trys: [], ops: [] }, f, y, t, g;
return g = { next: verb(0), "throw": verb(1), "return": verb(2) }, typeof Symbol === "function" && (g[Symbol.iterator] = function() { return this; }), g;
function verb(n) { return function (v) { return step([n, v]); }; }
function step(op) {
if (f) throw new TypeError("Generator is already executing.");
while (_) try {
if (f = 1, y && (t = op[0] & 2 ? y["return"] : op[0] ? y["throw"] || ((t = y["return"]) && t.call(y), 0) : y.next) && !(t = t.call(y, op[1])).done) return t;
if (y = 0, t) op = [op[0] & 2, t.value];
switch (op[0]) {
case 0: case 1: t = op; break;
case 4: _.label++; return { value: op[1], done: false };
case 5: _.label++; y = op[1]; op = [0]; continue;
case 7: op = _.ops.pop(); _.trys.pop(); continue;
default:
if (!(t = _.trys, t = t.length > 0 && t[t.length - 1]) && (op[0] === 6 || op[0] === 2)) { _ = 0; continue; }
if (op[0] === 3 && (!t || (op[1] > t[0] && op[1] < t[3]))) { _.label = op[1]; break; }
if (op[0] === 6 && _.label < t[1]) { _.label = t[1]; t = op; break; }
if (t && _.label < t[2]) { _.label = t[2]; _.ops.push(op); break; }
if (t[2]) _.ops.pop();
_.trys.pop(); continue;
}
op = body.call(thisArg, _);
} catch (e) { op = [6, e]; y = 0; } finally { f = t = 0; }
if (op[0] & 5) throw op[1]; return { value: op[0] ? op[1] : void 0, done: true };
}
};
var _this = this;
Object.defineProperty(exports, "__esModule", { value: true });
/**
* @license
* Copyright 2018 Google LLC. All Rights Reserved.
* Licensed under the Apache License, Version 2.0 (the "License");
* you may not use this file except in compliance with the License.
* You may obtain a copy of the License at
*
* http://www.apache.org/licenses/LICENSE-2.0
*
* Unless required by applicable law or agreed to in writing, software
* distributed under the License is distributed on an "AS IS" BASIS,
* WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
* See the License for the specific language governing permissions and
* limitations under the License.
* =============================================================================
*/
var tf = require("../index");
var jasmine_util_1 = require("../jasmine_util");
var test_util_1 = require("../test_util");
jasmine_util_1.describeWithFlags('complex64', jasmine_util_1.ALL_ENVS, function () {
it('tf.complex', function () { return __awaiter(_this, void 0, void 0, function () {
var real, imag, complex, _a;
return __generator(this, function (_b) {
switch (_b.label) {
case 0:
real = tf.tensor1d([3, 30]);
imag = tf.tensor1d([4, 40]);
complex = tf.complex(real, imag);
expect(complex.dtype).toBe('complex64');
expect(complex.shape).toEqual(real.shape);
_a = test_util_1.expectArraysClose;
return [4 /*yield*/, complex.data()];
case 1:
_a.apply(void 0, [_b.sent(), [3, 4, 30, 40]]);
return [2 /*return*/];
}
});
}); });
it('tf.real', function () { return __awaiter(_this, void 0, void 0, function () {
var complex, real, _a;
return __generator(this, function (_b) {
switch (_b.label) {
case 0:
complex = tf.complex([3, 30], [4, 40]);
real = tf.real(complex);
expect(real.dtype).toBe('float32');
expect(real.shape).toEqual([2]);
_a = test_util_1.expectArraysClose;
return [4 /*yield*/, real.data()];
case 1:
_a.apply(void 0, [_b.sent(), [3, 30]]);
return [2 /*return*/];
}
});
}); });
it('tf.imag', function () { return __awaiter(_this, void 0, void 0, function () {
var complex, imag, _a;
return __generator(this, function (_b) {
switch (_b.label) {
case 0:
complex = tf.complex([3, 30], [4, 40]);
imag = tf.imag(complex);
expect(imag.dtype).toBe('float32');
expect(imag.shape).toEqual([2]);
_a = test_util_1.expectArraysClose;
return [4 /*yield*/, imag.data()];
case 1:
_a.apply(void 0, [_b.sent(), [4, 40]]);
return [2 /*return*/];
}
});
}); });
it('throws when shapes dont match', function () {
var real = tf.tensor1d([3, 30]);
var imag = tf.tensor1d([4, 40, 50]);
var re = /real and imag shapes, 2 and 3, must match in call to tf.complex\(\)/;
expect(function () { return tf.complex(real, imag); }).toThrowError(re);
});
});
var BYTES_PER_COMPLEX_ELEMENT = 4 * 2;
jasmine_util_1.describeWithFlags('complex64 memory', jasmine_util_1.BROWSER_ENVS, function () {
it('usage', function () { return __awaiter(_this, void 0, void 0, function () {
var numTensors, numBuffers, startTensors, real1, imag1, complex1, real2, imag2, complex2, result, _a, real, _b, imag, _c;
return __generator(this, function (_d) {
switch (_d.label) {
case 0:
numTensors = tf.memory().numTensors;
numBuffers = tf.memory().numDataBuffers;
startTensors = numTensors;
real1 = tf.tensor1d([1]);
imag1 = tf.tensor1d([2]);
complex1 = tf.complex(real1, imag1);
// 5 new Tensors: real1, imag1, complex1, and two internal clones.
expect(tf.memory().numTensors).toBe(numTensors + 5);
// Only 3 new data buckets are actually created.
expect(tf.memory().numDataBuffers).toBe(numBuffers + 3);
numTensors = tf.memory().numTensors;
numBuffers = tf.memory().numDataBuffers;
real2 = tf.tensor1d([3]);
imag2 = tf.tensor1d([4]);
complex2 = tf.complex(real2, imag2);
// 5 new Tensors: real1, imag1, complex1, and two internal clones.
expect(tf.memory().numTensors).toBe(numTensors + 5);
// Only 3 new data buckets are actually created.
expect(tf.memory().numDataBuffers).toBe(numBuffers + 3);
numTensors = tf.memory().numTensors;
numBuffers = tf.memory().numDataBuffers;
result = complex1.add(complex2);
// A complex tensor is created, which is composed of 2 underlying tensors.
expect(tf.memory().numTensors).toBe(numTensors + 3);
numTensors = tf.memory().numTensors;
expect(result.dtype).toBe('complex64');
expect(result.shape).toEqual([1]);
_a = test_util_1.expectArraysClose;
return [4 /*yield*/, result.data()];
case 1:
_a.apply(void 0, [_d.sent(), [4, 6]]);
real = tf.real(result);
expect(tf.memory().numTensors).toBe(numTensors + 1);
numTensors = tf.memory().numTensors;
_b = test_util_1.expectArraysClose;
return [4 /*yield*/, real.data()];
case 2:
_b.apply(void 0, [_d.sent(), [4]]);
imag = tf.imag(result);
expect(tf.memory().numTensors).toBe(numTensors + 1);
numTensors = tf.memory().numTensors;
_c = test_util_1.expectArraysClose;
return [4 /*yield*/, imag.data()];
case 3:
_c.apply(void 0, [_d.sent(), [6]]);
// After disposing, there should be no tensors.
real1.dispose();
imag1.dispose();
real2.dispose();
imag2.dispose();
complex1.dispose();
complex2.dispose();
result.dispose();
real.dispose();
imag.dispose();
expect(tf.memory().numTensors).toBe(startTensors);
return [2 /*return*/];
}
});
}); });
it('tf.complex disposing underlying tensors', function () { return __awaiter(_this, void 0, void 0, function () {
var numTensors, real, imag, complex, _a;
return __generator(this, function (_b) {
switch (_b.label) {
case 0:
numTensors = tf.memory().numTensors;
real = tf.tensor1d([3, 30]);
imag = tf.tensor1d([4, 40]);
expect(tf.memory().numTensors).toEqual(numTensors + 2);
complex = tf.complex(real, imag);
// real and imag are cloned.
expect(tf.memory().numTensors).toEqual(numTensors + 5);
real.dispose();
imag.dispose();
// A copy of real and imag still exist, the one owned by the complex tensor.
expect(tf.memory().numTensors).toEqual(numTensors + 3);
expect(complex.dtype).toBe('complex64');
expect(complex.shape).toEqual(real.shape);
_a = test_util_1.expectArraysClose;
return [4 /*yield*/, complex.data()];
case 1:
_a.apply(void 0, [_b.sent(), [3, 4, 30, 40]]);
complex.dispose();
expect(tf.memory().numTensors).toEqual(numTensors);
return [2 /*return*/];
}
});
}); });
it('reshape', function () { return __awaiter(_this, void 0, void 0, function () {
var memoryBefore, a, b, _a, _b;
return __generator(this, function (_c) {
switch (_c.label) {
case 0:
memoryBefore = tf.memory();
a = tf.complex([[1, 3, 5], [7, 9, 11]], [[2, 4, 6], [8, 10, 12]]);
// 3 new tensors, the complex64 tensor and the 2 underlying float32 tensors.
expect(tf.memory().numTensors).toBe(memoryBefore.numTensors + 3);
// Bytes should be counted once.
expect(tf.memory().numBytes)
.toBe(memoryBefore.numBytes + 6 * BYTES_PER_COMPLEX_ELEMENT);
b = a.reshape([6]);
// 1 new tensor from the reshape.
expect(tf.memory().numTensors).toBe(memoryBefore.numTensors + 4);
// No new bytes from a reshape.
expect(tf.memory().numBytes)
.toBe(memoryBefore.numBytes + 6 * BYTES_PER_COMPLEX_ELEMENT);
expect(b.dtype).toBe('complex64');
expect(b.shape).toEqual([6]);
_a = test_util_1.expectArraysClose;
return [4 /*yield*/, a.data()];
case 1:
_b = [_c.sent()];
return [4 /*yield*/, b.data()];
case 2:
_a.apply(void 0, _b.concat([_c.sent()]));
b.dispose();
// 1 complex tensor should be disposed.
expect(tf.memory().numTensors).toBe(memoryBefore.numTensors + 3);
// Byte count should not change because the refcounts are all 1.
expect(tf.memory().numBytes)
.toBe(memoryBefore.numBytes + 6 * BYTES_PER_COMPLEX_ELEMENT);
a.dispose();
// All the tensors should now be disposed.
expect(tf.memory().numTensors).toBe(memoryBefore.numTensors);
// The underlying memory should now be released.
expect(tf.memory().numBytes).toBe(memoryBefore.numBytes);
return [2 /*return*/];
}
});
}); });
it('clone', function () { return __awaiter(_this, void 0, void 0, function () {
var memoryBefore, a, b, _a, _b;
return __generator(this, function (_c) {
switch (_c.label) {
case 0:
memoryBefore = tf.memory();
a = tf.complex([[1, 3, 5], [7, 9, 11]], [[2, 4, 6], [8, 10, 12]]);
// 3 new tensors, the complex64 tensor and the 2 underlying float32 tensors.
expect(tf.memory().numTensors).toBe(memoryBefore.numTensors + 3);
// Bytes should be counted once
expect(tf.memory().numBytes)
.toBe(memoryBefore.numBytes + 6 * BYTES_PER_COMPLEX_ELEMENT);
b = a.clone();
// 1 new tensor from the clone.
expect(tf.memory().numTensors).toBe(memoryBefore.numTensors + 4);
// No new bytes from a clone.
expect(tf.memory().numBytes)
.toBe(memoryBefore.numBytes + 6 * BYTES_PER_COMPLEX_ELEMENT);
expect(b.dtype).toBe('complex64');
_a = test_util_1.expectArraysClose;
return [4 /*yield*/, a.data()];
case 1:
_b = [_c.sent()];
return [4 /*yield*/, b.data()];
case 2:
_a.apply(void 0, _b.concat([_c.sent()]));
b.dispose();
// 1 complex tensor should be disposed.
expect(tf.memory().numTensors).toBe(memoryBefore.numTensors + 3);
// Byte count should not change because the refcounts are all 1.
expect(tf.memory().numBytes)
.toBe(memoryBefore.numBytes + 6 * BYTES_PER_COMPLEX_ELEMENT);
a.dispose();
// All the tensors should now be disposed.
expect(tf.memory().numTensors).toBe(memoryBefore.numTensors);
// The underlying memory should now be released.
expect(tf.memory().numBytes).toBe(memoryBefore.numBytes);
return [2 /*return*/];
}
});
}); });
});
//# sourceMappingURL=complex_ops_test.js.map