Time Stepping Classes for Optimization
نویسندگان
چکیده
This report introduces the “Time Stepping Package for Optimization”, or TSOpt, which is an interface for time-stepping simulation written in C++. It packages a simulator together with its derivatives (“sensitivities”) and adjoint derivatives with respect to simulation parameters in a single object called a Jet, which can be used in conjunction with an optimization algorithm to solve a simulation-driven optimization problem. Further, TSOpt interfaces with the Rice Vector Library (RVL), allowing Jet objects to define a Operator subclass. INTRODUCTION We are interested in solving optimization problems of the form min c J(c) = G(u(c, ·)) (1) where the state u(t) and the control c(t) solve the state equation H̄ ( du dt , u, c ) = 0 . (2) Though many types of control and inverse problems fit the mathematical framework above, we note that the procedure of solving such problems numerically oft involve the same procedures (e.g., formation of derivatives). To exploit these commonalities, hence, we created TSOpt, or the “Time-Stepping for Optimization” package. TSOpt is a “middleware” package written in C++, designed to act as an “interface for time-stepping simulation”, providing a way for simulation software to inter-operate with optimization software. TSOpt is capable of encapsulating the reference, linearized and adjoint simulators in a single object, and properly arrange their execution. TSOpt also aids in providing necessary data structures for the optimization algorithm (e.g., the gradient, formed via the adjointstate method). This report is organized as follows: the first section introduces RVL and section two discusses the Alg framework developed by Tony Padula. RVL and the Alg framework provides the foundation for TSOpt. The most notable features of TSOpt include its modular code structure, due to use of the Alg framework from the Rice Vector Library (RVL), and also accommodation of a generic data structure type through templating. The specifics of the structure of TSOpt and its features will be discussed in more detail in section three.
منابع مشابه
A General Solution for Implicit Time Stepping Scheme in Rate-dependant Plasticity
In this paper the derivation of the second differentiation of a general yield surface implicit time stepping method along with its consistent elastic-plastic modulus is studied. Moreover, the explicit, trapezoidal implicit and fully implicit time stepping schemes are compared in rate-dependant plasticity. It is shown that implementing fully implicit time stepping scheme in rate-dependant plasti...
متن کاملStrongly stable multi-time stepping method with the option of controlling amplitude decay in responses
Recently, multi-time stepping methods have become very popular among scientist due to their high stability in problems with critical conditions. One important shortcoming of these methods backs to their high amount of uncontrolled amplitude decay. This study proposes a new multi-time stepping method in which the time step is split into two sub-steps. The first sub-step is solved using the well-...
متن کاملA Time-Stepping Library for Simulation-Driven Optimization
The Timestepping Simulation for Optimization (”TSOpt”) library provides an interface for time-stepping simulation. It packages a simulator together with its derivatives (”sensitivities”) and adjoint derivatives with respect to simulation parameters, and uses the aggregate to define a Rice Vector Library Operator subclass.
متن کاملFractional Order Glucose Insulin System Using Fractional Back-Stepping Sliding Mode Control
In this paper, based on a fractional order Bergman minimal model, a robust strategy for regulationof blood glucose in type 1 diabetic patients is presented. Glucose/insulin concentration in the patientbody is controlled through the injection under the patients skin by the pump. Many various con-trollers for this system have been proposed in the literature. However, most of the...
متن کاملVariational Contact-Implicit Trajectory Optimization
We present a formulation of rigid body dynamics with frictional contact that leverages ideas from discrete mechanics to derive time-stepping schemes with any desired order of integration accuracy. By utilizing the new time-stepping methods as dynamics constraints in a direct trajectory optimization scheme, we are able to perform contact-implicit trajectory optimization with accuracy equal to tr...
متن کاملذخیره در منابع من
با ذخیره ی این منبع در منابع من، دسترسی به آن را برای استفاده های بعدی آسان تر کنید
عنوان ژورنال:
دوره شماره
صفحات -
تاریخ انتشار 2010