论文部分内容阅读
A new hybrid interpolation method for constructing smooth potential energy surfaces(PESs)from ab initio data is proposed.By successively performing Reproducing Kernel Hilbert Space(RKHS)interpolation and cubic spline interpolation,this method efficiently yields globally accurate potential energy surfaces.In addition to inheriting most advantages of the pure RKHS method,such as correct permutation symmetry and asymptotic behavior,and the capability of providing analytic derivatives,this method offers the following extra benefits: First,a faster initiation time,which would become an important factor to affect the fitting efficiency when the number of available ab initio points is large.Lengthy initiation time may makes the fitting process prohibitively expensive; Second,enhanced stability and accuracy due to the decreased ill-conditionality of linear inverse problem when performing RKHS interpolation; We also proposed a fast algorithm of this hybrid method so that the computational time for the calculation of the PES and its derivatives is greatly reduced and competitive with the pure RKHS method.We have applied this method to N(2D)+H2→ NH+H reaction to test our method.An adiabatic PES of NH2(X2A")state has been constructed from more than 20 000 ab initio points.Accurate quantum dynamics results calculated on the constructed PES demonstrates high accuracy and efficiency of this new fast hybrid interpolation method.