Showing papers by "Jean Utke published in 2010"

Interpretative adjoints for numerical simulation codes using MPI

Abstract: An essential performance and correctness factor in numerical simulation and optimization is access to exact derivative information. Adjoint derivative models are particularly useful if a function’s number of inputs far exceeds the number of outputs. The propagation of adjoints requires the data flow to be reversed, implying the reversal of all communication in programs that use message-passing. This paper presents recent advances made in developing the adjoint MPI library AMPI. The described proof of concept aims to serve as the basis for coupling other overloading AD tools with AMPI. We illustrate its use in the context of a specific overloading tool for algorithmic differentiation (AD) for C++ programs. A simplified but representative application problem is discussed as a case study.

Using Automatic Differentiation in Sensitivity Analysis of Nuclear Simulation Models, invited

Abstract: Sensitivity analysis is an important tool in the study of nuclear systems. In our recent work, we introduced a hybrid method that combines sampling techniques with first-order sensitivity analysis to approximate the effects of uncertainty in parameters of a nuclear reactor simulation model. For elementary examples, the approach offers a substantial advantage (in precision, computational efficiency, or both) over classical methods of uncertainty quantification.

Efficient generated libraries for asynchronous derivative computation

Abstract: The computation of derivatives via automatic differentiation is a valuable technique in many science and engineering applications. While the implementation of automatic differentiation via source transformation yields the highest-efficiency results, the implementation via operator overloading remains a viable alternative for some application contexts, such as the computation of higher-order derivatives or in cases where C++ still proves to be too complicated for the currently available source transformation tools. The Rapsodia code generator creates libraries that overload intrinsics for derivative computation. In this paper, we discuss modifications to Rapsodia to improve the efficiency of the generated code, first via limited loop unrolling and second via multithreaded asynchronous derivative computation. We introduce the approaches and present runtime results.