You are on page 1of 3

Determinant and inverse matrix[edit]

See also: Determinant § Relation to eigenvalues and trace, and Characteristic polynomial


§ Properties
For a general n × n invertible matrix A, i.e., one with nonzero determinant, A−1 can thus be written as
an (n − 1)-th order polynomial expression in A: As indicated, the Cayley–Hamilton theorem amounts
to the identity

The coefficients ci are given by the elementary symmetric polynomials of the eigenvalues of A.


Using Newton identities, the elementary symmetric polynomials can in turn be expressed in terms
of power sum symmetric polynomials of the eigenvalues:

where tr(Ak) is the trace of the matrix Ak. Thus, we can express ci in terms of the trace of powers
of A.
In general, the formula for the coefficients ci is given in terms of complete exponential Bell
polynomials as[nb 2]

In particular, the determinant of A equals (−1)nc0. Thus, the determinant can be written as


the trace identity:

Likewise, the characteristic polynomial can be written as

and, by multiplying both sides by A−1 (note −(−1)n = (−1)n−1), one is led to an


expression for the inverse of A as a trace identity,

Another method for obtaining these coefficients ck for a general n × n matrix,


provided no root be zero, relies on the following alternative expression for the
determinant,

Hence, by virtue of the Mercator series,


where the exponential only needs be expanded to order λ−n,
since p(λ) is of order n, the net negative powers of λ automatically
vanishing by the C–H theorem. (Again, this requires a ring containing
the rational numbers.) Differentiation of this expression with respect
to λ allows one to express the coefficients of the characteristic
polynomial for general n as determinants of m × m matrices,[nb 3]

Examples
For instance, the first few Bell polynomials are B0 = 1, B1(x1)
= x1, B2(x1, x2) = x2
1 + x2, and B3(x1, x2, x3) = x
3

1 + 3 x1x2 + x3.

Using these to specify the coefficients ci of the characteristic


polynomial of a 2 × 2 matrix yields

The coefficient c0 gives the determinant of the 2 × 


2 matrix, c1 minus its trace, while its inverse is given by

It is apparent from the general formula for cn−k, expressed in


terms of Bell polynomials, that the expressions

always give the coefficients cn−1 of λn−1 and cn−2 of λn−2 in


the characteristic polynomial of any n × n matrix,
respectively. So, for a 3 × 3 matrix A, the statement of
the Cayley–Hamilton theorem can also be written as

where the right-hand side designates a 3 × 


3 matrix with all entries reduced to zero. Likewise,
this determinant in the n = 3 case, is now

This expression gives the negative of


coefficient cn−3 of λn−3 in the general case, as
seen below.
Similarly, one can write for a 4 × 4 matrix A,
where, now, the determinant is cn−4,

and so on for larger matrices. The


increasingly complex expressions for
the coefficients ck is deducible
from Newton's identities or
the Faddeev–LeVerrier algorithm.

You might also like