Sensitivity theory for nonlinear systems. II. Extensions to additional classes of responses

Technical Report · Sat Aug 01 04:00:00 EDT 1981

OSTI ID:6083779

Cacuci, D G

This work extends a recent, functional-analytic formulation of sensitivity theory to include treatment of additional types of responses. There are physical systems where a critical point of a function that depends on the system's state vector and parameters defines the location in phase-space where the response functional is evaluated. The Gateaux-differentials giving the sensitivities of both the functional and the critical point to changes in the system's parameters are obtained by alternative formalisms. The Forward Sensitivity Formalism is the simpler and more general, but may be prohibitively expensive for problems with large data bases. The Adjoint Sensitivity Formalism, although less generally applicable and requiring several adjoint calculations, is likely to be the only practical approach. Sensitivity theory is also extended to include treatment of general operators, acting on the system's state vector and parameters, as responses. In this case, the Forward Sensitivity Formalism is the same as for functional responses, but the Adjoint Sensitivity Formalism is considerably different. The Adjoint Sensitivity Formalism requires expanding the indirect effect term, an element of a Hilbert space, in terms of elements of an orthonormal basis. Since as many calculations of adjoint functions are required as there are non-zero terms in this expansion, careful consideration of truncating the expansion is needed to assess the advantages of the Adjoint Sensitivity Formalism over the Forward Sensitivity Formalism.

• Research Organization: Oak Ridge National Lab., TN (USA)
• DOE Contract Number: W-7405-ENG-26
• OSTI ID: 6083779
• Report Number(s): ORNL/TM-7511; ON: DE81030258
• Country of Publication: United States
• Language: English

Related Subjects

99 GENERAL AND MISCELLANEOUS
990200* -- Mathematics & Computers
ANALYTICAL SOLUTION
COMPARATIVE EVALUATIONS
NONLINEAR PROBLEMS
SENSITIVITY ANALYSIS