Abstract
A new bound for the condition number of the matrix exponential is presented. Using the bound, we propose an efficient approximation to the condition number, denoted by κg(s, X), that avoids the computation of the Fréchet derivative of the matrix exponential that underlies condition number estimation in the existing algorithms. We exploit the identity eX=(eX/2s)2s for a nonnegative integer s with the properties of the Fréchet derivative operator to obtain the bound. Our cost analysis reveals that considerable computational savings are possible since estimating the condition number by the existing algorithms requires several invocation of the Fréchet derivative of the matrix exponential whose single invocation costs as twice as the cost of the matrix exponential itself. The bound and hence κg(s, X) only involve Fréchet derivative of a monomial of degree 2s, which can be computed exactly in 2s matrix multiplications. We propose two versions of the scaling and squaring algorithm that implement κg(s, X). Our numerical experiments show that κg(s, X) captures the behavior of the condition number and moreover outperforms the condition number in the estimation of relative forward errors for a wide range of problems.