Mathematics Stack Exchange is a question and answer site for people studying math at any level and professionals in related fields. 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

It is well known that for $n=2$, this holds. The polar decomposition provides the topology of $\operatorname{SL}(n,\mathbb R)$ as the product of symmetric matrices and orthogonal matrices, which can be written as the product of exponentials of skew symmetric and symmetric traceless matrices. However I could not find out the proof that $\exp: \mathfrak{sl}(n,\mathbb R)\to\operatorname{SL}(n,\mathbb R)$ is not surjective for $n\geq 3$.

share|cite|improve this question
Lie algebras are commonly written with lower case fraktur letters. Hence it should be \mathfrak{sl} instead of \text{sl}. – kahen May 13 '13 at 20:21

Over $\mathbb{R}$, for a general $n$, a real matrix has a real logarithm if and only if it is nonsingular and in its (complex) Jordan normal form, every Jordan block corresponding to a negative eigenvalue occurs an even number of times. So, you may verify that $\pmatrix{-1&1\\ 0&-1}$ (as given by rschwieb's answer) and $\operatorname{diag}(-2,-\frac12,1,\ldots,1)$ (as given by Gokler's answer) are not matrix exponentials of a real traceless matrix.

share|cite|improve this answer

I came up with an example. Consider diag(-2,-1/2, 1, ..., 1) in SL(n,R). Clearly the lattice of logarithm of this does not intersect sl(n,R).

share|cite|improve this answer

I remember learning somewhere that every matrix in the image of the exponential from $\mathfrak{sl}(n,\Bbb R)$ is either diagonalizable or is unipotent. So, any defective element which is not unipotent would be missed.

Defective matrices with determinant 1 can be built by installing a defective block. For example:

$$ \begin{bmatrix}-1&1\\0&-1\end{bmatrix},\begin{bmatrix}-1&1&0\\0&-1&0\\0&0&1\end{bmatrix},\begin{bmatrix}-1&1&0&0\\0&-1&0&0\\0&0&-1&0\\0&0&0&-1\end{bmatrix}\text,{etc.} $$

share|cite|improve this answer
Your second example has determinant $-1$. Also, what you describe in the first paragraph is true for $n=2$, but are you sure that it is true for a general $n$? – user1551 May 14 '13 at 4:05
@user1551 The comment: Doh good point... I managed to forget about the determinant. I originally had in mind just to give the matrix a defective block, but yeah, what I wrote did not do that. The question: I don't remember it depending on $n$, I thought it depended upon nondiagonalizable elements of $\mathfrak{sl}(n,\Bbb R)$ being nilpotent. Do you know if that conclusion is confined to $n=2$? – rschwieb May 14 '13 at 11:13
It doesn't seem to be true for a general $n$. Consider $R=\pmatrix{0&-1\\ 1&0}$, $e^R=\pmatrix{\cos1&-\sin1\\ \sin1&\cos1}$ and $X=\pmatrix{R&I\\ 0&R}$. We have $e^X=\pmatrix{e^R&e^R\\ 0&e^R}$, which is neither diagonalisable nor unipotent. – user1551 May 14 '13 at 14:22
@user1551 OK, thanks for letting me know. I found the original place I learned this, and it was ambiguous about whether it was true in general. It looks like you still believe the example I gave, however, so I'll have to leave this solution in place for your other solution to reference it properly – rschwieb May 14 '13 at 15:10

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.