Revisiting Individual Discipline Feasible using matrix-free Inexact-Newton-Krylov

Abstract

The individual-discipline-feasible (IDF) formulation was proposed to simplify the implementation of MDO problems. The IDF formulation introduces coupling variables into the optimization problem that eliminate the need for a full multidisciplinary analysis at each optimization iteration; this simplifies the solution of MDO problems by maintaining modularity of the discipline software. Historically, the MDO community has used conventional optimization algorithms to solve IDF-formulated problems. Conventional optimizers are not well suited to IDF, because they use limited-memory quasi-Newton methods (linear convergence) and require the constraint Jacobian explicitly. The cost of computing the coupling-variable constraint Jacobian is prohibitively expensive for high-fidelity IDF problems. Matrix-free Reduced-Space inexact-Newton-Krylov (RSNK) algorithms overcome these issues, because they scale superlinearly and do not require the constraint Jacobian explicitly. Therefore, this class of algorithm has great potential to solve IDF-formulated MDO problems in a scalable and efficient manner. In this paper, we describe the application of RSNK to the IDF formulation and compare its performance to the multidisciplinary feasible architecture.

Publication
10th AIAA Multidisciplinary Design Optimization Conference, National Harbor, Maryland, USA