I am trying to check whether the limit mentioned in the equation (3.5) of this textbook (page 14 of the PDF) does really simplify into the claimed expression.

I am trying to simplify this limit in particular: $$ \lim_{k\to\infty} \left(\mathbf{1} + \alpha_a\frac{X_a}{k}\right)^k = \sum_{m=0}^{\infty} \frac{1}{m!}\left(\alpha_aX_a\right)^m \equiv e^{\alpha_aX_a}, $$ where $X_a$ is a matrix and $\alpha_a$ is a scalar.

To make this case as simple as possible, I tried to simplify the limit using the following concrete matrix $X = \begin{bmatrix}0 & -1\\1 & 0\end{bmatrix}$.

The result of the limit simplification should then be $\begin{bmatrix}\cos\alpha & -\sin\alpha\\\sin\alpha & \cos\alpha\end{bmatrix}$, as is mentioned in equation (3.6) of the textbook.

However, Instead of that result, MMA simplified the limit into the $2\times2$ identity matrix. This is the expression I entered into MMA:

A = {
  {0, -a},
  {a, 0}

Limit[(IdentityMatrix[2] + A/N)^N, N -> Infinity] // MatrixForm

How can I replicate the textbook results?

  • 1
    $\begingroup$ This may not matter, but avoid uppercase symbols in general, and particularly uppercase single letter symbols like A, N. N is a function in its own right (see its documentation. $\endgroup$
    – MarcoB
    Feb 4, 2022 at 2:58
  • $\begingroup$ Thanks for the heads up, I didn't know about this. Changing N to k and A to r did not change the result, as you wrote (i.e. MMA still simplified the expression to $\mathbf{I}$). $\endgroup$
    – jordi
    Feb 4, 2022 at 3:35

1 Answer 1


Use MatrixPower and MatrixFunction

A = {{0, -a}, {a, 0}};

Limit[MatrixPower[IdentityMatrix[2] + A/n, n], n -> Infinity] // 

(* {{Cos[a], -Sin[a]}, {Sin[a], Cos[a]}} *)

MatrixFunction[Exp, A]

(* {{Cos[a], -Sin[a]}, {Sin[a], Cos[a]}} *)
  • 2
    $\begingroup$ There is also MatrixExp[A] $\endgroup$
    – Carl Woll
    Feb 4, 2022 at 4:15
  • $\begingroup$ Thank you, using MatrixPower did the trick. I did not expect Mathematica to require using a special function for matrix powers (I thought the ^ operator was enough). How can I be sure in the future that MMA won't provide misleading results (like the identity matrix) when simplifying arbitrary expressions? Are there any other caveats with symbolic simplification that I should be aware of? $\endgroup$
    – jordi
    Feb 4, 2022 at 15:10
  • $\begingroup$ Power[matrix, n] and MatrixPower[matrix, n] are two completely different mathematical operations. Likewise f[matrix] and MatrixFunction[f, matrix] are radically different. Mathematica cannot guess which operation you want; you need to specify the intended operation. $\endgroup$
    – Bob Hanlon
    Feb 4, 2022 at 15:36
  • $\begingroup$ @BobHanlon Oh, I didn't realize there was a difference, but after reading your comment, it makes complete sense to me! So Power[] performs element-wise power, whereas MatrixPower[] performs repeated matrix multiplication. $\endgroup$
    – jordi
    Feb 4, 2022 at 20:39

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge that you have read and understand our privacy policy and code of conduct.

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