Vector calculus identities

From Wikipedia, the free encyclopedia
Jump to navigation Jump to search

The following are important identities involving derivatives and integrals in vector calculus.

Operator notation[edit]


For a function in three-dimensional Cartesian coordinate variables, the gradient is the vector field:

where i, j, k are the standard unit vectors for the x, y, z-axes. More generally, for a function of n variables , also called a scalar field, the gradient is the vector field:
where are orthogonal unit vectors in arbitrary directions.

For a vector field written as a 1 × n row vector, also called a tensor field of order 1, the gradient or covariant derivative is the n × n Jacobian matrix:

For a tensor field of any order k, the gradient is a tensor field of order k + 1.


In Cartesian coordinates, the divergence of a continuously differentiable vector field is the scalar-valued function:

The divergence of a tensor field of non-zero order k is written as , a contraction to a tensor field of order k − 1. Specifically, the divergence of a vector is a scalar. The divergence of a higher order tensor field may be found by decomposing the tensor field into a sum of outer products and using the identity,

where is the directional derivative in the direction of multiplied by its magnitude. Specifically, for the outer product of two vectors,


In Cartesian coordinates, for the curl is the vector field:

where i, j, and k are the unit vectors for the x-, y-, and z-axes, respectively. In Einstein notation, the vector field has curl given by:
where = ±1 or 0 is the Levi-Civita parity symbol.


In Cartesian coordinates, the Laplacian of a function is

For a tensor field, , the Laplacian is generally written as:

and is a tensor field of the same order.

When the Laplacian is equal to 0, the function is called a harmonic function. That is,

Special notations[edit]

In Feynman subscript notation,

where the notation ∇B means the subscripted gradient operates on only the factor B.[1][2]

Less general but similar is the Hestenes overdot notation in geometric algebra.[3] The above identity is then expressed as:

where overdots define the scope of the vector derivative. The dotted vector, in this case B, is differentiated, while the (undotted) A is held constant.

For the remainder of this article, Feynman subscript notation will be used where appropriate.

First derivative identities[edit]

For scalar fields , and vector fields , , we have the following derivative identities.

Distributive properties[edit]

Product rule for multiplication by a scalar[edit]

We have the following generalizations of the product rule in single variable calculus.

In the second formula, the transposed gradient is an n × 1 column vector, is a 1 × n row vector, and their product is an n × n matrix (or more precisely, a dyad); This may also be considered as the tensor product of two vectors, or of a covector and a vector.

Quotient rule for division by a scalar[edit]

Chain rule[edit]

Let be a one-variable function from scalars to scalars, a parametrized curve, and a function from vectors to scalars. We have the following special cases of the multi-variable chain rule.

For a coordinate parametrization we have:

Here we take the trace of the product of two n × n matrices: the gradient of A and the Jacobian of .

Dot product rule[edit]

where denotes the Jacobian matrix of the vector field .

Alternatively, using Feynman subscript notation,

See these notes.[4]

As a special case, when A = B,

The generalization of the dot product formula to Riemannian manifolds is a defining property of a Riemannian connection, which differentiates a vector field to give a vector-valued 1-form.

Cross product rule[edit]

Note that the matrix is antisymmetric.

Second derivative identities[edit]

Divergence of curl is zero[edit]

The divergence of the curl of any vector field A is always zero:

This is a special case of the vanishing of the square of the exterior derivative in the De Rham chain complex.

Divergence of gradient is Laplacian[edit]

The Laplacian of a scalar field is the divergence of its gradient:

The result is a scalar quantity.

Divergence of divergence is not defined[edit]

Divergence of a vector field A is a scalar, and you cannot take the divergence of a scalar quantity. Therefore:

Curl of gradient is zero[edit]

The curl of the gradient of any continuously twice-differentiable scalar field is always the zero vector:

This is a special case of the vanishing of the square of the exterior derivative in the De Rham chain complex.

Curl of curl[edit]

Here ∇2 is the vector Laplacian operating on the vector field A.

Curl of divergence is not defined[edit]

The divergence of a vector field A is a scalar, and you cannot take curl of a scalar quantity. Therefore

DCG chart: Some rules for second derivatives.

A mnemonic[edit]

The figure to the right is a mnemonic for some of these identities. The abbreviations used are:

  • D: divergence,
  • C: curl,
  • G: gradient,
  • L: Laplacian,
  • CC: curl of curl.

Each arrow is labeled with the result of an identity, specifically, the result of applying the operator at the arrow's tail to the operator at its head. The blue circle in the middle means curl of curl exists, whereas the other two red circles (dashed) mean that DD and GG do not exist.

Summary of important identities[edit]





  • [5]

Vector dot Del Operator[edit]

  • [6]

Second derivatives[edit]

  • (scalar Laplacian)
  • (vector Laplacian)
  • (Green's vector identity)

Third derivatives[edit]


Below, the curly symbol ∂ means "boundary of" a surface or solid.

Surface–volume integrals[edit]

In the following surface–volume integral theorems, V denotes a three-dimensional volume with a corresponding two-dimensional boundary S = ∂V (a closed surface):

  • \oiint (divergence theorem)
  • \oiint
  • \oiint
  • \oiint (Green's first identity)
  • \oiint \oiint (Green's second identity)
  • \oiint (integration by parts)
  • \oiint (integration by parts)

Curve–surface integrals[edit]

In the following curve–surface integral theorems, S denotes a 2d open surface with a corresponding 1d boundary C = ∂S (a closed curve):

  • (Stokes' theorem)

Integration around a closed curve in the clockwise sense is the negative of the same line integral in the counterclockwise sense (analogous to interchanging the limits in a definite integral):

\ointclockwise \ointctrclockwise

Endpoint-curve integrals[edit]

In the following endpoint–curve integral theorems, P denotes a 1d open path with signed 0d boundary points and integration along P is from to :

  • (Gradient theorem).

See also[edit]


  1. ^ Feynman, R. P.; Leighton, R. B.; Sands, M. (1964). The Feynman Lectures on Physics. Addison-Wesley. Vol II, p. 27–4. ISBN 0-8053-9049-9.
  2. ^ Kholmetskii, A. L.; Missevitch, O. V. (2005). "The Faraday induction law in relativity theory". p. 4. arXiv:physics/0504223.
  3. ^ Doran, C.; Lasenby, A. (2003). Geometric algebra for physicists. Cambridge University Press. p. 169. ISBN 978-0-521-71595-9.
  4. ^ Kelly, P. (2013). "Chapter 1.14 Tensor Calculus 1: Tensor Fields" (PDF). Mechanics Lecture Notes Part III: Foundations of Continuum Mechanics. University of Auckland. Retrieved 7 December 2017.
  5. ^ "lecture15.pdf" (PDF).{{cite web}}: CS1 maint: url-status (link)
  6. ^ Kuo, Kenneth K.; Acharya, Ragini (2012). Applications of turbulent and multi-phase combustion. Hoboken, N.J.: Wiley. p. 520. doi:10.1002/9781118127575.app1. ISBN 9781118127575. Archived from the original on 19 April 2020. Retrieved 19 April 2020.

Further reading[edit]