1: \begin{abstract}

2: We consider the closest-point projection with respect to the Frobenius norm

3: of a general real square matrix to the set $\SL(n)$ of matrices with unit determinant.

4: As it turns out, it is sufficient to consider diagonal matrices only.

5: We investigate the structure of the problem both in Euclidean coordinates

6: and in an $n$-dimensional generalization of the classical hyperbolic coordinates

7: of the positive quadrant.

8: Using symmetry arguments we show that the global minimizer is contained

9: in a particular cone. Based on different views of the problem, we propose

10: four different iterative algorithms, and we give convergence results for all of them.

11: Numerical tests show that computing the projection costs essentially as much

12: as a singular value decomposition.

13: Finally, we give an explicit formula for the first derivative of the projection.

14: \end{abstract}

15: