Fast Algorithms for Imaging Simulation through Turbulence

The Phase I work has shown the technical feasibility of creating new, equivalent-accuracy algorithms for optical simulation with smaller operations counts than existing Fourier methods. These will be helpful in speeding up extended scene imaging simulations used in laser weapons' systems performance evaluations. However, the computational efficiency advantage of these new algorithms over Fourier methods depends on the ability to optimize the new algorithms to the same level as has been done for FFTs. The primary technical goal of the Phase II is to develop and test a robust optical propagator containing generic software optimizations that will make these new algorithms more efficient than an optimized Fourier propagator, both on single processors (nodes) and multi-node computational clusters. Added to these efficiencies will be absorbing and impedance boundary conditions which both broaden the physical applications of optical simulations and further enhance their efficiency by allowing for removal of “guard” points in the computational grid included only to avoid wrap-around errors. These boundary conditions complete the transition from periodic to general spatial domains which is required for more general propagation scenarios, particularly those close to the ground. The combined efficiencies of new algorithms and boundary conditions is expected to decrease optical simulation times by at least a factor of 10 compared to current Fourier propagators. The efficiency and boundary conditions enhancements produced in the Phase II will first be included and tested in WaveTrain, but will also be offered as a portable package suitable for inclusion in other optical simulation software.