β Scribed by Paul Stangerup
No coin nor oath required. For personal study only.
This paper describes the implementation of sensitivity calculations in a general-purpose simulation program ESACAP. It is well known that exact derivatives of a steady state or time domain solution with respect to any parameter in a nonlinear dynamic system can be obtained at very little additional computational effort. The algorithm using already available Jacobians in a numerical integration scheme is applied to backward-differentiation formulas (BDFs) of variable step and order. Sensitivity calculation in the ESACAP version of the BDFs based on scaled differences is described. The main purpose of this paper is to identify the additional pieces of code required for sensitivity calculations in connection with a general problem-description language. The paper also describes the computation of exact derivatives in small-signal frequency-domain analyses including poles and zeros of transfer functions. Results from typical ESACAP simulations with sensitivities are presented. (~) 1999 Elsevier Science B.V.
π SIMILAR VOLUMES
The quality of compiler-optimized code for high-performance applications is far behind what optimization and domain experts can achieve by hand. Although it may seem surprising at first glance, the performance gap has been widening over time, due to the tremendous complexity increase in microprocess
Special-purpose parallel machines that are plugged into a workstation to accelerate molecular dynamics (MD) simulations are attracting a considerable amount of interest. These machines comprise scalable homogeneous multiprocessors for calculating nonbonded forces (Coulombic and van der Waals forces)
## Author of original program: p~G. Burke No. of cards required to effect adaptation (including directive cards): 4 Card punching code: BCD ## Nature ofphysical problem provement in time, but typically is of the order of 30%. The purpose of the program is to express the matrix Another factor w