MathOverflow is a question and answer site for professional mathematicians. Join them; it only takes a minute:

Sign up
Here's how it works:
  1. Anybody can ask a question
  2. Anybody can answer
  3. The best answers are voted up and rise to the top

Does there exist an analytical method by which i can exponentiate a 4 by 4 matrix, in the same way as the general 2 by 2 matrix case in pauli matrix basis. I have dirac matrices (which are composed of direct products of pauli matrices) as my basis for 4 by 4 matrices. I need an analytical way ! Any reply is appreciated. regards

share|cite|improve this question
What is the result you are alluding to for $2\times 2? matrices? – Igor Rivin Aug 12 '12 at 23:22
By "analytically" you mean "explicitely"? Put your matrix $X$ in Jordan form. Then $X=S+N$ where $S$ is diagonal and $N$ nilpotent and $SN=NS$, and $\exp(X)=\exp(S)\cdot\exp(N)$ which is quite explicit to compute... – Qfwfq Aug 12 '12 at 23:47
@Qfwfq: that is computationally tractable, but NOT explicit (try writing a formula in terms of matrix elements of $X$) – Igor Rivin Aug 13 '12 at 0:12
Also, why all the votes to close? – Igor Rivin Aug 13 '12 at 0:12

Perhaps I misunderstand the question. When you say you have Dirac matrices, does that mean that you are computing the exponential of a liner combination of Dirac matrices? If so, then there is a very simple analytical formula in any dimension: just use the Clifford relations in the exponential series.

More concretely, suppose that you would like to compute the exponential of a matrix $X := \sum_i x^i \Gamma_i$, where the Dirac matrices $\Gamma_i$ obey the Clifford relation $$ \Gamma_i \Gamma_j + \Gamma_j \Gamma_i = - 2 g_{ij} I~, $$ with $I$ the identity matrix. Then it follows from this relation that $$ X^2 = - x^2 I~, $$ where I have introduced the (indefinite, if $g_{ij}$ has indefinite signature) "squared norm" $$ x^2 = \sum_{i,j} x^i x^j g_{ij}~. $$

If $x^2$ = 0, then $$ \exp X = I + X $$ and if $x^2 \neq 0$, then letting $x = \sqrt{x^2}$ (which could be imaginary), $$ \exp X = \cos x I + \frac{\sin x}{x} X~. $$

Added (for the "heathens")

Quiaochu's comment is correct. Here are some more details.

Let $V$ be a finite-dimensional real vector space with a non-degenerate inner product $\left<-,-\right>$. Let $Cl(V)$ be the corresponding Clifford algebra. Let $\rho: Cl(V) \to \operatorname{End}(M)$ be an irreducible representation of $Cl(V)$. Let $(e_i)$ be a basis for $V$. Then $\Gamma_i := \rho(e_i)$ are called Dirac matrices of $CL(V)$ in the representation $M$.

share|cite|improve this answer
For us heathens: what are Dirac matrices? – Igor Rivin Aug 13 '12 at 13:35
@Igor Rivin: as far as I understand (which is not to say much) the first display (Clifford relation) is sort-of the defintion. – user9072 Aug 13 '12 at 14:08
@Igor: they are a particular matrix representation of a certain Clifford algebra. – Qiaochu Yuan Aug 13 '12 at 16:05

There is a completely explicit formula in this paper of Bensauod and Mouline (rendicotti Palermo, 2005), which is quite compact for low dimensions.

share|cite|improve this answer
(There seems to be a problem with the link.) – Andrés E. Caicedo Aug 13 '12 at 0:12
@Andres: should be fixed now... – Igor Rivin Aug 13 '12 at 0:35
It's explicit in terms of the solution of a differential equation related to the characteristic polynomial. Of course, to solve that differential equation explicitly you need the eigenvalues... – Robert Israel Aug 13 '12 at 1:29

Your Answer


By posting your answer, you agree to the privacy policy and terms of service.

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