• Corpus ID: 11459307

Algorithmic Differentiation of Numerical Methods : Second-Order Tangent and Adjoint Solvers for Systems of Parametrized Nonlinear Equations

@inproceedings{Safiran2014AlgorithmicDO,
  title={Algorithmic Differentiation of Numerical Methods : Second-Order Tangent and Adjoint Solvers for Systems of Parametrized Nonlinear Equations},
  author={Niloofar Safiran and Johannes Lotz and Uwe Naumann},
  year={2014}
}
Forward and reverse modes of algorithmic differentiation (AD) transform implementations of multivariate vector functions F : IR → IR as computer programs into tangent and adjoint code, respectively. The reapplication of the same ideas yields higher derivative code. In particular, second derivatives play an important role in nonlinear programming. Second-order methods based on Newton’s algorithm promise faster convergence in the neighbourhood of the minimum by taking into account second… 

Figures and Tables from this paper

Second-order Tangent Solvers for Systems of Parameterized Nonlinear Equations
TLDR
The local computational overhead for the computation of second-order tangents of the solution vector with respect to the parameters by Algorithmic Differentiation depends on the number of iterations performed by the nonlinear solver, but this dependence can be eliminated by taking a second- order symbolic approach to differentiation of the non linear system.
Symbolic vs . Algorithmic Differentiation of GSL Integration Routines
Forward and reverse modes of algorithmic differentiation (AD) transform implementations of multivariate vector functions F : IR → IR as computer programs into tangent and adjoint code, respectively.
Meta Adjoint Programming in C + +
Adjoint code development for many-core architectures like GPUs is a major challenge. To address issues arising from the parallelization, currently hand-writing the adjoint code is the most promising
Reachability analysis of non-linear hybrid systems using Taylor Models
TLDR
This thesis focuses on the techniques to compute all reachable states over a bounded time horizon and finitely many jumps for a hybrid system with non-linear dynamics, and presents the use of Taylor models as the over-approximate representations for nonlinear ODE solutions.
Java program analysis by symbolic execution
TLDR
This thesis presents a preprocessing step for Java Bytecode programs in which all complicated language constructs are handled and a new analysis technique is presented which helps end users identify parts of the analyzed code which are irrelevant for the desired outcome.
Inferring Lower Bounds for Runtime Complexity
TLDR
This work presents the first approach to deduce lower bounds for innermost runtime complexity of term rewrite systems (TRSs) automatically, to generate suitable families of rewrite sequences of a TRS and to find a relation between the length of such a rewrite sequence and the size of the first term in the sequence.
Information Hiding in the Public RSA Modulus
The manufacturer of an asymmetric backdoor for a public key cryptosystem manipulates the key generation process in such a way that he can extract the private key or other secret information from the
CD2Alloy: A Translation of Class Diagrams to Alloy and back from Alloy Instances to Object Diagrams
TLDR
The translations are defined and an example is given, showing a class diagram and its complete representation in Alloy, as well as a corresponding Alloy instance and itscomplete representation back in an object diagram.
Abstraction and refinement of probabilistic automata using modal stochastic games
TLDR
This paper describes the development of probabilistic automata using modal Stochastic Games and describes the design and refinement of these automata with real-time constraints.
Towards Privacy-Preserving Multi-party Bartering
TLDR
This work has devised the first privacy-preserving protocol for automatically determining an actual trade between multiple parties without involving a trusted third party.

References

SHOWING 1-10 OF 175 REFERENCES
Algorithmic Differentiation of Numerical Methods : Tangent-Linear and Adjoint Solvers for Systems of Nonlinear Equations
We discuss software tool support for the Algorithmic Differentiation (also known as Automatic Differentiation; AD) of numerical simulation programs that contain calls to solvers for parameterized
Algorithmic Differentiation of Numerical Methods : Tangent-Linear and Adjoint Direct Solvers for Systems of Linear Equations
We consider the Algorithmic Differentiation (also know as Automatic Differentiation; AD) of numerical simulation programs that contain calls to direct solvers for systems of n linear equations. AD of
Evaluating derivatives - principles and techniques of algorithmic differentiation, Second Edition
TLDR
This second edition has been updated and expanded to cover recent developments in applications and theory, including an elegant NP completeness argument by Uwe Naumann and a brief introduction to scarcity, a generalization of sparsity.
The Art of Differentiating Computer Programs - An Introduction to Algorithmic Differentiation
  • U. Naumann
  • Computer Science
    Software, environments, tools
  • 2012
TLDR
This is the first entry-level book on algorithmic differentiation (AD), providing fundamental rules for the generation of first- and higher-order tangent-linear and adjoint code.
Second-order adjoints by source code manipulation of numerical programs
TLDR
A first research prototype of the NAGWare Fortran compiler with the ability to generate adjoint code automatically is presented and an approach to generating second-order adjointcode for use in Newtontype algorithms for unconstrained nonlinear optimization is discussed.
Toward Low Static Memory Jacobian Accumulation
Derivatives are essential ingredients of a wide range of numerical algorithms. We focus on the accumulation of Jacobian matrices by Gaussian elimination on a sparse implementation of the extended
Syntax-Directed Derivative Code (Part I: Tangent-Linear Code)
This is the first instance in a series of papers on single-pass generation of various types of derivative code by syntax-directed translation. We consider the automatic generation of tangent-linear
Optimal vertex elimination in single-expression-use graphs
TLDR
This article presents an efficient algorithm for solving a relevant subclass of this combinatorial optimization problem that aims to minimize the number of arithmetic operations performed by the elimination algorithm and relies on the ability to compute vertex covers in bipartite graphs in polynomial time.
dco/c++: Derivative Code by Overloading in C++
Boosted by advanced type genericity and support for template metaprogramming techniques, the role of C++ as the preferred language for largescale numerical simulation in Computational Science,
Algorithm 755: ADOL-C: a package for the automatic differentiation of algorithms written in C/C++
The C++ package ADOL-C described here facilitates the evaluation of first and higher derivatives of vector functions that are defined by computer programs written in C or C++. The resulting
...
1
2
3
4
5
...