Time-integrated form of the semiclassical initial value method

A method is presented that greatly improves the efficiency of semiclassical initial value representation treatments by transforming phase space integration variables to time, energy, and additional coordinates and momenta on a Poincare surface. Since the integration over time can be treated as an integration along the classical motion, the number of trajectories needed to obtain convergence is significantly reduced. The technique is applied to test cases involving bounded motion with very encouraging results. (C) 1999 American Institute of Physics. [S0021-9606(99)01918-2].