Orthogonal matrix and trigonometric functions of matrices

154 Views Asked by At

Let me try to summarize my question by taking into consideration some elementary notions of orthogonal matrices.

The general $2\times2$ orthogonal matrix has the form:

$$O = \begin{pmatrix} \cos\theta & \sin\theta \\ -\sin\theta & \cos\theta \end{pmatrix} $$

Where $\theta$ is an arbitrary real number. It follows that $O^T O = I$. This matrix is also capable of diagonalizing some symmetric matrix $A$:

$$O^T A O = \text{diag}(A)$$

Now, let's assume that $\theta$ is not a real number, but rather, is a matrix. Therefore:

$$O^T O = \begin{pmatrix}\cos\theta^T \cos\theta + \sin\theta^T \sin\theta & \cos\theta^T\sin\theta - \sin\theta^T\cos\theta \\ -\sin\theta^T\cos\theta + \cos\theta^T\sin\theta & \cos\theta^T\cos\theta + \sin\theta^T\sin\theta \end{pmatrix}$$

It is no longer clear that $O^T O$ is the identity. It actually seems like:

$$O^\prime = \begin{pmatrix}\cos\theta & -\sin\theta \\ \sin\theta & \cos\theta \end{pmatrix}$$

is the inverse of $O$ instead of $O^T$. Moreover, there also seems to be an issue with the diagonalization of our matrix $A$, it seems like $O^T A O$ would no longer yield a diagonal matrix because $O$ and $O^T$ are no longer the inverse of each other (maybe I am wrong with my reasoning?), but if I use $O^\prime A O$ instead, there is still an issue since it would seem as if the diagonal matrix is not symmetric, and, therefore, can't be diagonal.

Is this generalization completely nonsensical or am I missing something here?