8
$\begingroup$

Is there something like a power series expansion for a determinant? I mean the following thing: if $k$ is a field (of characteristic zero) and $M$ and $N$ are two square matrices of the same size over $k$, can we then express $\det(M + tN)$ in a nice way as a power series - which should actually be a polynomial - in $t$?

$\endgroup$
2
  • 3
    $\begingroup$ How about $$\det(I-tA)=\exp\left(-\sum_{n=1}^\infty\frac{t^n}{n}\mathrm{tr}(A^n)\right)?$$ $\endgroup$
    – anon
    Sep 8 '11 at 9:16
  • $\begingroup$ Also, Newton's identities "in reverse" allow one to write the characteristic polynomial $\det(tI-A)$ as a polynomial in $\mathrm{tr}(A^k)$, for $k=1,2,\dots,d$, and this can be generalized to what you want. The coefficients will be defined recursively, so it's arguable how nice of an expression it is. Also, the case when $\det M = 0$ might cause technical issues. $\endgroup$
    – anon
    Sep 8 '11 at 9:26
9
$\begingroup$

To expand on anon's comment:

$$\begin{align}\det (I-A) &= \exp\log\det (I-A) \\&=\exp\text{Tr}\log (I-A) \\ &=\exp\text{Tr}\left(-\sum_{n=1}^\infty\frac{A^n}n\right) \\&=\exp\left(-\sum_{n=1}^\infty\frac{\text{Tr}A^n}n\right)\;.\end{align}$$

$\endgroup$
1
  • $\begingroup$ Is the formula $\log\det=\text{tr}\log$ true also when the matrix $I-A$ does not admit a logarithm, so it is just a formal power series? Because I found proofs of this using the relation $\det\exp(A)=\exp\text{tr}(A)$ and writing $\log(B)=A$, but if $\log(B)$ is not defined I don't know how to work. $\endgroup$
    – Losten
    May 21 '19 at 12:17
5
$\begingroup$

Well, if a row of a matrix is a sum of two rows then the determinant is equal to the sum of the corresponding determinants. So $\det(M + tN)$ can be represented as $2^k$ summands: $$ \det(M + tN)=\sum_{m=0}^k t^m\sum \det A_{i_1,\ldots,i_m}^{i_{m+1},\ldots,i_{k}} $$ where $A_{i_1,\ldots,i_m}^{i_{m+1},\ldots,i_{k}}\;$ is a matrix which has rows $i_1,\ldots,i_m$ of the matrix $N$ and the rows $i_{m+1},\ldots,i_{k}$ from the matrix $M$.

$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.