Variable opsConst

Type declaration

• angleBetween: ((u: ad.Num[], v: ad.Num[]) => ad.Num)

  Return the unsigned angle between vectors u, v, in radians. Assumes that both u and v have nonzero magnitude. The returned value will be in the range [0,pi].

  • • (u, v): ad.Num

    • Parameters

      • u: ad.Num[]
      • v: ad.Num[]

      Returns ad.Num

• angleFrom: ((u: ad.Num[], v: ad.Num[]) => ad.Num)

  Return the signed angle from vector u to vector v, in radians. Assumes that both u and v are 2D vectors and have nonzero magnitude. The returned value will be in the range [-pi,pi].

  • • (u, v): ad.Num

    • Parameters

      • u: ad.Num[]
      • v: ad.Num[]

      Returns ad.Num

• cross2: ((u: ad.Num[], v: ad.Num[]) => ad.Num)

  Return 2D determinant/cross product of 2D vectors

  • • (u, v): ad.Num

    • Parameters

      • u: ad.Num[]
      • v: ad.Num[]

      Returns ad.Num

• cross3: ((u: ad.Num[], v: ad.Num[]) => ad.Num[])

  Return 3D cross product of 3D vectors

  • • (u, v): ad.Num[]

    • Parameters

      • u: ad.Num[]
      • v: ad.Num[]

      Returns ad.Num[]

• dist: ((c1: ad.Num, c2: ad.Num) => ad.Num)

  Return the Euclidean distance between scalars c1, c2.

  • • (c1, c2): ad.Num

    • Parameters

      • c1: ad.Num
      • c2: ad.Num

      Returns ad.Num

• ewmmdiv: ((A1: ad.Num[][], A2: ad.Num[][]) => ad.Num[][])

  Return the elementwise quotient of matrices A1, A2.

  • • (A1, A2): ad.Num[][]

    • Parameters

      • A1: ad.Num[][]
      • A2: ad.Num[][]

      Returns ad.Num[][]

• ewmmmul: ((A1: ad.Num[][], A2: ad.Num[][]) => ad.Num[][])

  Return the elementwise product of matrices A1, A2.

  • • (A1, A2): ad.Num[][]

    • Parameters

      • A1: ad.Num[][]
      • A2: ad.Num[][]

      Returns ad.Num[][]

• ewvvdiv: ((v1: ad.Num[], v2: ad.Num[]) => ad.Num[])

  Return the elementwise quotient of vectors v1 and v2.

  • • (v1, v2): ad.Num[]

    • Parameters

      • v1: ad.Num[]
      • v2: ad.Num[]

      Returns ad.Num[]

• ewvvmul: ((v1: ad.Num[], v2: ad.Num[]) => ad.Num[])

  Return the elementwise product of vectors v1 and v2.

  • • (v1, v2): ad.Num[]

    • Parameters

      • v1: ad.Num[]
      • v2: ad.Num[]

      Returns ad.Num[]

• mmadd: ((A1: ad.Num[][], A2: ad.Num[][]) => ad.Num[][])

  Return the sum of matrices A1, A2.

  • • (A1, A2): ad.Num[][]

    • Parameters

      • A1: ad.Num[][]
      • A2: ad.Num[][]

      Returns ad.Num[][]

• mmmul: ((A: ad.Num[][], B: ad.Num[][]) => ad.Num[][])

  Return the matrix A multiplied by matrix B.

  • • (A, B): ad.Num[][]

    • Parameters

      • A: ad.Num[][]
      • B: ad.Num[][]

      Returns ad.Num[][]

• mmsub: ((A1: ad.Num[][], A2: ad.Num[][]) => ad.Num[][])

  Return the difference of matrices A1, A2.

  • • (A1, A2): ad.Num[][]

    • Parameters

      • A1: ad.Num[][]
      • A2: ad.Num[][]

      Returns ad.Num[][]

• msdiv: ((A: ad.Num[][], c: ad.Num) => ad.Num[][])

  Return the Matrix A divided by scalar c.

  • • (A, c): ad.Num[][]

    • Parameters

      • A: ad.Num[][]
      • c: ad.Num

      Returns ad.Num[][]

• mtrans: ((A: ad.Num[][]) => ad.Num[][])

  Return the transpose of the matrix A.

  • • (A): ad.Num[][]

    • Parameters

      • A: ad.Num[][]

      Returns ad.Num[][]

• mvmul: ((A: ad.Num[][], v: ad.Num[]) => ad.Num[])

  Return the matrix A multiplied by vector v, i.e., Av.

  • • (A, v): ad.Num[]

    • Parameters

      • A: ad.Num[][]
      • v: ad.Num[]

      Returns ad.Num[]

• norm: ((c1: ad.Num, c2: ad.Num) => ad.Num)

  Return the norm of the 2-vector [c1, c2].

  • • (c1, c2): ad.Num

    • Parameters

      • c1: ad.Num
      • c2: ad.Num

      Returns ad.Num

• rot90: (([x, y]: ad.Num[]) => ad.Num[])

  Rotate a 2D point [x, y] by 90 degrees counterclockwise.

  • • ([x, y]): ad.Num[]

    • Parameters

      • [x, y]: ad.Num[]

      Returns ad.Num[]

• smmul: ((c: ad.Num, A: ad.Num[][]) => ad.Num[][])

  Return the scalar c times the Matrix A.

  • • (c, A): ad.Num[][]

    • Parameters

      • c: ad.Num
      • A: ad.Num[][]

      Returns ad.Num[][]

• vabs: ((v: ad.Num[]) => ad.Num[])

  Return the entrywise absolute value of the vector v

  • • (v): ad.Num[]

    • Parameters

      • v: ad.Num[]

      Returns ad.Num[]

• vadd: ((v1: ad.Num[], v2: ad.Num[]) => ad.Num[])

  Return the sum of vectors v1, v2.

  • • (v1, v2): ad.Num[]

    • Parameters

      • v1: ad.Num[]
      • v2: ad.Num[]

      Returns ad.Num[]

• vdist: ((v: ad.Num[], w: ad.Num[]) => ad.Num)

  Return the Euclidean distance between vectors v and w.

  • • (v, w): ad.Num

    • Parameters

      • v: ad.Num[]
      • w: ad.Num[]

      Returns ad.Num

• vdistsq: ((v: ad.Num[], w: ad.Num[]) => ad.Num)

  Return the Euclidean distance squared between vectors v and w.

  • • (v, w): ad.Num

    • Parameters

      • v: ad.Num[]
      • w: ad.Num[]

      Returns ad.Num

• vdiv: ((v: ad.Num[], c: ad.Num) => ad.Num[])

  Return the vector v divided by scalar c.

  • • (v, c): ad.Num[]

    • Parameters

      • v: ad.Num[]
      • c: ad.Num

      Returns ad.Num[]

• vdot: ((v1: ad.Num[], v2: ad.Num[]) => ad.Num)

  Return the dot product of vectors v1, v2. Note: if you want to compute a norm squared, use vnormsq instead, it generates a smaller computational graph

  • • (v1, v2): ad.Num

    • Parameters

      • v1: ad.Num[]
      • v2: ad.Num[]

      Returns ad.Num

• vmax: ((v1: ad.Num[], v2: ad.Num[]) => ad.Num[])

  Return the maximum value of each component of the vectors v1 and v2

  • • (v1, v2): ad.Num[]

    • Parameters

      • v1: ad.Num[]
      • v2: ad.Num[]

      Returns ad.Num[]

• vmmul: ((v: ad.Num[], A: ad.Num[][]) => ad.Num[])

  Return the vector v multiplied by matrix A, i.e., v^T A.

  • • (v, A): ad.Num[]

    • Parameters

      • v: ad.Num[]
      • A: ad.Num[][]

      Returns ad.Num[]

• vmove: ((v: ad.Num[], c: ad.Num, u: ad.Num[]) => ad.Num[])

  Return v + c * u.

  • • (v, c, u): ad.Num[]

    • Parameters

      • v: ad.Num[]
      • c: ad.Num
      • u: ad.Num[]

      Returns ad.Num[]

• vmul: ((c: ad.Num, v: ad.Num[]) => ad.Num[])

  Return the vector v multiplied by scalar c.

  • • (c, v): ad.Num[]

    • Parameters

      • c: ad.Num
      • v: ad.Num[]

      Returns ad.Num[]

• vneg: ((v: ad.Num[]) => ad.Num[])

  Return the vector v, scaled by -1.

  • • (v): ad.Num[]

    • Parameters

      • v: ad.Num[]

      Returns ad.Num[]

• vnorm: ((v: ad.Num[]) => ad.Num)

  Return the Euclidean norm of vector v.

  • • (v): ad.Num

    • Parameters

      • v: ad.Num[]

      Returns ad.Num

• vnormalize: ((v: ad.Num[]) => ad.Num[])

  Return the vector v, normalized.

  • • (v): ad.Num[]

    • Parameters

      • v: ad.Num[]

      Returns ad.Num[]

• vnormsq: ((v: ad.Num[]) => ad.Num)

  Return the Euclidean norm squared of vector v.

  • • (v): ad.Num

    • Parameters

      • v: ad.Num[]

      Returns ad.Num

• vouter: ((u: ad.Num[], v: ad.Num[]) => ad.Num[][])

  Return outer product matrix uv^T. Vectors u and v must have the same length.

  NOTE: This functionality is duplicated in outerProduct() from Functions.ts. Since outerProduct has a more directly interpretable name, we may wish to deprecate vouter and move outerProduct into Autodiff.ts in a future release.

  • • (u, v): ad.Num[][]

    • Parameters

      • u: ad.Num[]
      • v: ad.Num[]

      Returns ad.Num[][]

• vproduct: ((v1: ad.Num[], v2: ad.Num[]) => ad.Num[])

  Returns the entrywise product of two vectors, v1 and v2

  • • (v1, v2): ad.Num[]

    • Parameters

      • v1: ad.Num[]
      • v2: ad.Num[]

      Returns ad.Num[]

• vrot: (([x, y]: ad.Num[], a: ad.Num) => ad.Num[])

  Rotate a 2D point [x, y] by a degrees counterclockwise.

  • • ([x, y], a): ad.Num[]

    • Parameters

      • [x, y]: ad.Num[]
      • a: ad.Num

      Returns ad.Num[]

• vsub: ((v1: ad.Num[], v2: ad.Num[]) => ad.Num[])

  Return the difference of vectors v1 and v2.

  • • (v1, v2): ad.Num[]

    • Parameters

      • v1: ad.Num[]
      • v2: ad.Num[]

      Returns ad.Num[]

• vsum: ((v: ad.Num[]) => ad.Num)

  Return the sum of elements in vector v.

  • • (v): ad.Num

    • Parameters

      • v: ad.Num[]

      Returns ad.Num