Journal article demonstrating that multifidelity sampling techniques combining high and low-fidelity network simulations produce estimators with significantly lower variance, making them effective UQ tools for network analysis.
Sandia LDRD report summarizing a three-year project to quantify behavioral (not performance) differences between emulations and real-world systems by running representative workloads on both and comparing collected metrics.
Research infrastructure that executed over 10,000 experiments processing half a petabyte of data to quantify behavioral differences between virtual and physical testbeds for cyber security research validation.
Investigates multifidelity UQ strategies for network simulations by combining limited high-fidelity model runs with numerous lower-fidelity evaluations to achieve more accurate estimators with lower variance.