Convolutional discrete Fourier transform method for calculating thermal neutron cross section in liquids

Being exact at both short- and long-time limits, the Gaussian approximation is widely used to calculate neutron incoherent inelastic scattering functions in liquids. However, to overcome a few numerical difficulties, extra physical approximations are often employed to ease the evaluation. In this work, a new numerical method, called convolutional discrete Fourier transform, is proposed to perform Fourier transform of $\exp[-f(t)]$. We have applied this method to compute the differential cross sections of light water up to 10eV. The obtained results, thoroughly benchmarked against experimental data, showed a much higher dynamic range than conventional fast Fourier transform. The calculated integral cross sections agree closely with the light water data in the state-of-the-art nuclear data library. It is in evidence that this numerical method can be used in the place of the extra physical approximations.