SIMBAD references

A pulsar's frequency has never been discovered from γ-ray data. Severe problems arise because of the sparse data and long observation times, but the excellent sensitivity of the EGRET instrument on the Compton Gamma Ray Observatory offers a solution. For strong sources, it is possible to split an observation into smaller time intervals and retain useful signals which can later be combined. Although there is signal loss overall, frequency searches on the separate intervals accumulate many fewer trials than a normal search because the need to consider frequency derivatives can be eliminated and the sensitivity to source location uncertainty is much reduced. We present an `evolutionary' method in which the trials are further reduced by testing data only at those frequencies which have given some indication of periodicity in all previous data intervals. For a typical two-week EGRET dataset, the method reduces the total number of trials by a factor of around 300 and the computation time by a factor of more than 10³. The evolutionary technique is applied to the now known X-ray and γ-ray pulsar Geminga and is shown to identify the correct spin period. The implications for the identification of other EGRET sources are discussed.