Abstract: A high-performance computing (HPC) framework for accelerating sparse Cholesky factorization on field-programmable gate arrays (FPGAs) is provided. The proposed framework includes an FPGA kernel implementing a throughput-optimized hardware architecture for accelerating a supernodal multifrontal algorithm for sparse Cholesky factorization. The proposed framework further includes a host program implementing a novel scheduling algorithm for finding the optimal execution order of supernode computations for an elimination tree on the FPGA to eliminate the need for off-chip memory access for storing intermediate results. Moreover, the proposed scheduling algorithm minimizes on-chip memory requirements for buffering intermediate results by resolving the dependency of parent nodes in an elimination tree through temporal parallelism.

Abstract: A simulated participant system is configured to interoperate with existing simulators to create an integrated simulation platform with enhanced capabilities to provide, on a readily-scalable basis, simulated participants of one or more different types that are based on dynamically constructed virtual models of live simulation participants. The system observes live participants as they train with a simulator using a closed-loop, self-training configuration by continuously monitoring conversations and actions to create and maintain the virtual models on-the-fly. The virtual models may then be utilized to generate simulated participants to stand in for the live participants when they are absent from future simulator training sessions. Because a simulated participant was virtually modeled from observations of a live participant, the simulated behaviors, as expressed by actions and responses, can typically closely match the expected behavior of the absent live participant.

Abstract: A simulated participant system is configured to interoperate with existing simulators to create an integrated simulation platform with enhanced capabilities to provide, on a readily-scalable basis, simulated participants of one or more different types that are based on dynamically constructed virtual models of live simulation participants. The system observes live participants as they train with a simulator using a closed-loop, self-training configuration by continuously monitoring conversations and actions to create and maintain the virtual models on-the-fly. The virtual models may then be utilized to generate simulated participants to stand in for the live participants when they are absent from future simulator training sessions. Because a simulated participant was virtually modeled from observations of a live participant, the simulated behaviors, as expressed by actions and responses, can typically closely match the expected behavior of the absent live participant.