With the help of a simple variational procedure it is possible to convert the partial sums of order N of many divergent series expansions f (g) = ∞ n=0 a n g n into partial sums N n=0 b n g −ωn , where 0 < ω < 1 is a parameter that parametrizes the approach to the large-g limit. The latter are partial sums of a strong-coupling expansion of f (g) which converge against f (g) for g outside a certain divergence radius. The error decreases exponentially fast for large N , like e −const.×N 1−ω. We

more »

... esent a review of the method and various applications.