modularise linalg ops #3464

tafsiri · 2020-06-17T01:04:35Z

To see the logs from the Cloud Build CI, please join either our discussion or announcement mailing list.

This change is

lina128

Reviewable status: 0 of 1 approvals obtained (waiting on @annxingyuan and @tafsiri)

tfjs-core/src/ops/band_part.ts, line 71 at r1 (raw file):

function bandPart_<T extends Tensor>(
    a: T|TensorLike, numLower: number, numUpper: number): T {
  if (numLower % 1 !== 0) {

Use util.assert?

tfjs-core/src/ops/band_part.ts, line 86 at r1 (raw file):

  }

  const shape = $a.shape, [M, N] = $a.shape.slice(-2);

One variable per declaration. https://google.github.io/styleguide/jsguide.html#features-one-variable-per-declaration

tfjs-core/src/ops/band_part.ts, line 106 at r1 (raw file):

  }

  const i = range(0, M, 1, 'int32').reshape([-1, 1]),

One variable per declaration.

tfjs-core/src/ops/band_part.ts, line 106 at r1 (raw file):

  }

  const i = range(0, M, 1, 'int32').reshape([-1, 1]),

Remove chained op.

tfjs-core/src/ops/band_part.ts, line 115 at r1 (raw file):

  const zero = zeros([M, N], $a.dtype);

  return stack(unstack($a.reshape([-1, M, N]))

Remove chained op.

tfjs-core/src/ops/band_part.ts, line 117 at r1 (raw file):

  return stack(unstack($a.reshape([-1, M, N]))
                   .map(mat => where(inBand, mat, zero)))
             .reshape(shape) as T;

Remove chained op.

tfjs-core/src/ops/gram_schmidt.ts, line 91 at r1 (raw file):

      if (i > 0) {
        for (let j = 0; j < i; ++j) {
          const proj = sum(ys[j].mul(x)).mul(ys[j]);

Remove chained op.

tfjs-core/src/ops/gram_schmidt.ts, line 92 at r1 (raw file):

        for (let j = 0; j < i; ++j) {
          const proj = sum(ys[j].mul(x)).mul(ys[j]);
          x = x.sub(proj);

Remove chained op.

tfjs-core/src/ops/gram_schmidt.ts, line 95 at r1 (raw file):

        }
      }
      return x.div(norm(x, 'euclidean'));

Remove chained op.

tfjs-core/src/ops/qr.ts, line 70 at r1 (raw file):

 */
function qr_(x: Tensor, fullMatrices = false): [Tensor, Tensor] {
  if (x.rank < 2) {

util.assert

tfjs-core/src/ops/qr.ts, line 81 at r1 (raw file):

    //   perform QR decomposition on them and then stack the results back
    //   together. We should explore whether this can be parallelized.
    const outerDimsProd = x.shape.slice(0, x.shape.length - 2)

Remove chained op.

tfjs-core/src/ops/qr.ts, line 84 at r1 (raw file):

                              .reduce((value, prev) => value * prev);
    const x2ds = unstack(
        x.reshape([

Removed chained op.

tfjs-core/src/ops/qr.ts, line 96 at r1 (raw file):

      r2ds.push(r2d);
    });
    const q = stack(q2ds, 0).reshape(x.shape);

Remove chained op.

tfjs-core/src/ops/qr.ts, line 97 at r1 (raw file):

    });
    const q = stack(q2ds, 0).reshape(x.shape);
    const r = stack(r2ds, 0).reshape(x.shape);

Remove chained op.

tfjs-core/src/ops/qr.ts, line 104 at r1 (raw file):

function qr2d(x: Tensor2D, fullMatrices = false): [Tensor2D, Tensor2D] {
  return ENGINE.tidy(() => {
    if (x.shape.length !== 2) {

util.assert

tfjs-core/src/ops/qr.ts, line 113 at r1 (raw file):

    let q = eye(m);     // Orthogonal transform so far.
    let r = x.clone();  // Transformed matrix so far.

Remove chained op.

tfjs-core/src/ops/qr.ts, line 116 at r1 (raw file):

    const one2D = tensor2d([[1]], [1, 1]);
    let w: Tensor2D = one2D.clone();

Remove chained op.

tfjs-core/src/ops/qr.ts, line 127 at r1 (raw file):

      [w, r, q] = ENGINE.tidy((): [Tensor2D, Tensor2D, Tensor2D] => {
        // Find H = I - tau * w * w', to put zeros below R(j, j).
        const rjEnd1 = r.slice([j, j], [m - j, 1]);

Remove chained ops here and below.

tafsiri

Thanks Na!

Reviewable status: 0 of 1 approvals obtained (waiting on @annxingyuan and @lina128)

tfjs-core/src/ops/band_part.ts, line 71 at r1 (raw file):