We propagate the wave packet in time and absorb the wave packet at the edges of the grid after each step (for absorption parameters, see Table I). To simplify the evaluation of we use the sorting scheme from Ref. . Also we use a scheme to delete DVR-points that are in regions that are physically not accessible, thus decreasing the calculation time and the amount of memory needed to store the reduced potential matrix . For more details about this part of the calculation we refer to the appendix of the paper.
After each Nt steps (see Table I) we calculate the derivative of the wave function at R=Rs in the H + O2 channel and at r=rs in the O + OH exit channel. For the value of Rs and rs see Table I. Together with the wave functions themselves at these surfaces we store the derivatives on disk. After the integration we use these vectors to calculate the energy-dependent (non)-reaction probability as outlined in Sec. IIC. We also use them to calculate the time-dependent reaction probability, given as , the time-dependent non-reaction probability, given as , and . For all our calculations we find that never deviates more than 10-6 from 1.