15/12/2018 Taylor series expansion

t i
Next: Sherman-Morrison formula Up: Appendices Previous: Review of Linar Algebra

Taylor series expansion

Single-variable functions:

A single-variable function can be expanded around a given point by the Taylor series:

When is small, the higher order terms can be neglected so that the function can be approximated as a quadratic function

or even a linear function

Multi-variable scalar-valued functions:

A multi-variable function can also be expanded by the Taylor series:

http://fourier.eng.hmc.edu/e176/lectures/NM/node45.html 1/4
15/12/2018 Taylor series expansion

which can be expressed in vector form as:

where is a vector and and are respectively the gradient vector and the Hessian matrix (first and second order derivatives

in single variable case) of the function defined as:

If the second derivatives of are continuous, then , i.e., is symmetric.

When is small (i.e., is small), then can be approximated by a quadratic function with a third order error term

or even a linear function with a second order error term

http://fourier.eng.hmc.edu/e176/lectures/NM/node45.html 2/4
15/12/2018 Taylor series expansion

Multi-variable vector-valued functions:

A set of multi-variable functions ( ) can be expressed as a vector function

The Taylor expansion of the ith ( ) component is:

The first two terms of these components can be written in vector form:

where is the Jacobian matrix defined over the vector function :

However, the 2nd order term can no longer expressed in matrix form, as it requires tensor notation.
http://fourier.eng.hmc.edu/e176/lectures/NM/node45.html 3/4
15/12/2018 Taylor series expansion

Note that the Hessian matrix of a function can be obtained as the Jacobian matrix of the gradient vector of :

Note that and may not be always the same. When , there are more equations than variables, the problem is over-constrained.

t i
Next: Sherman-Morrison formula Up: Appendices Previous: Review of Linar Algebra
Ruye Wang 2015-02-12

http://fourier.eng.hmc.edu/e176/lectures/NM/node45.html 4/4