Evaluating the noncentral chi-square distribution for the Cox-Ingersoll-Ross process (Q1768381)

This paper deals with the Cox-Ingersoll-Ross model and process, which can be expressed in terms of the non-central chi-square (NCS hereafter) distribution, and with its evaluation techniques. The author investigates three standard evaluation methods of the NCS distribution, viz. with series of gamma functions, by analytic approximation, and by asymptotic expansion. A fourth method is proposed and developed for evaluating the upper and lower tails of the NCS distribution, using a Bessel function series representation. The paper performs numerical tests for all the four NCS evaluation techniques, compares their efficiency and accuracy, and shows that the fourth proposed method based on Bessel function series has a good computational behaviour. Combining (some of) the four evaluation methods is also suggested to be an efficient technique to obtain an NCS evaluation accuracy of at least six significant digits.