Validating an A Priori Enclosure Using. High-Order Taylor Series. George F. Corliss and Robert Rihm. Abstract

Similar documents
Geometric Series Bounds for the Local Errors of Taylor Methods for Linear n-th Order ODEs

On the existence of an eigenvalue below the essential spectrum

Integrating ODE's in the Complex Plane-Pole Vaulting

Interval Methods and Taylor Model Methods for ODEs

Propagation of Uncertainties in Nonlinear Dynamic Models

Verified Calculation of the Nodes and Weights for Gaussian Quadrature Formulas

On a Unied Representation of Some Interval Analytic. Dedicated to the professors of mathematics. L. Berg, W. Engel, G. Pazderski, and H.- W. Stolle.

Optimization of Rational Approximations by Continued Fractions

INTERVAL MATHEMATICS TECHNIQUES FOR CONTROL THEORY COMPUTATIONS

Value Problem for an Ordinary Differential Equation. Nedialko Stoyanov Nedialkov. A thesis submitted in conformity with the requirements

D0 D1 s(t0,d0,t1) D2 s(t1,d1,t2)

ON EXPLICIT INTERVAL METHODS OF ADAMS-BASHFORTH TYPE

Validated Solutions of Initial Value Problems for Parametric ODEs

EXISTENCE VERIFICATION FOR SINGULAR ZEROS OF REAL NONLINEAR SYSTEMS

Our goal is to solve a general constant coecient linear second order. this way but that will not always happen). Once we have y 1, it will always

W. Kramer, U. Kulisch, R. Lohner. Numerical Toolbox. for Veried Computing II. Advanced Numerical Problems

9 Linear Interpolation and Estimation using Interval Analysis

A Finite Element Method for an Ill-Posed Problem. Martin-Luther-Universitat, Fachbereich Mathematik/Informatik,Postfach 8, D Halle, Abstract

Taylor series based nite dierence approximations of higher-degree derivatives

The Algebraic Multigrid Projection for Eigenvalue Problems; Backrotations and Multigrid Fixed Points. Sorin Costiner and Shlomo Ta'asan

ON TAYLOR MODEL BASED INTEGRATION OF ODES

Ratio-Like and Recurrence Relation Tests for Convergence of Series

ODEs and Redefining the Concept of Elementary Functions

Spurious Chaotic Solutions of Dierential. Equations. Sigitas Keras. September Department of Applied Mathematics and Theoretical Physics

16.7 Multistep, Multivalue, and Predictor-Corrector Methods

IMPLICIT INTERVAL MULTISTEP METHODS FOR SOLVING THE INITIAL VALUE PROBLEM

Automatica, 33(9): , September 1997.

Distributed computation of the number. of points on an elliptic curve

October 7, :8 WSPC/WS-IJWMIP paper. Polynomial functions are renable

Applied Mathematics &Optimization

Verified Global Optimization with Taylor Model based Range Bounders

Optimal Preconditioners for Interval Gauss Seidel Methods

UNIVERSITÄT KARLSRUHE

Rigorous numerical computation of polynomial differential equations over unbounded domains

CONSTRUCTIVE APPROXIMATION 2001 Springer-Verlag New York Inc.

Centro de Processamento de Dados, Universidade Federal do Rio Grande do Sul,

16.7 Multistep, Multivalue, and Predictor-Corrector Methods

Numerical Results on the Transcendence of Constants. David H. Bailey 275{281

Second Order ODEs. CSCC51H- Numerical Approx, Int and ODEs p.130/177

Numerical Integration exact integration is not needed to achieve the optimal convergence rate of nite element solutions ([, 9, 11], and Chapter 7). In

Enclosure Methods. G. Alefeld Institut für Angewandte Mathematik Universität Karlsruhe Federal Republic of Germany

INRIA Rocquencourt, Le Chesnay Cedex (France) y Dept. of Mathematics, North Carolina State University, Raleigh NC USA

Arbitrary Accuracy with Variable Precision Arithmetic

Butterworth Filter Properties

Investigation of a Subdivision Based Algorithm. for Solving Systems of Polynomial Equations. University of Applied Sciences / FH Konstanz

FROM EQUILIBRIUM TO CHAOS

Computation of the Bernstein Coecients on Subdivided. are computed directly from the coecients on the subdivided triangle from the

Using an Extension of Z. Viktor Friesen. Technische Universitat Berlin

Parallel Iterations for Cholesky and Rohn s Methods for Solving over Determined Linear Interval Systems

15 Nonlinear Equations and Zero-Finders

Numerical analysis meets number theory: using rootfinding methods to calculate inverses mod p n

AIMS Exercise Set # 1

Module 6 : Solving Ordinary Differential Equations - Initial Value Problems (ODE-IVPs) Section 1 : Introduction

Series Solutions Near an Ordinary Point

NUMERICAL SOLUTION OF ODE IVPs. Overview

Finite Differences: Consistency, Stability and Convergence

Structural Grobner Basis. Bernd Sturmfels and Markus Wiegelmann TR May Department of Mathematics, UC Berkeley.

Counterexamples to witness conjectures. Notations Let E be the set of admissible constant expressions built up from Z; + ;?;;/; exp and log. Here a co

On the Computational Procedure of Solving Boundary Value Problems of Class M Using the Power Series Method

CONSISTENCY TECHNIQUES FOR SIMULATION OF WASTEWATER TREATMENT PROCESSES WITH UNCERTAINTIES

t x 0.25

Novel Approach to Analysis of Nonlinear Recursions. 1 Department of Physics, Bar-Ilan University, Ramat-Gan, ISRAEL

Fachrichtung 6.1 Mathematik

reality is complex process

Deterministic Global Optimization for Dynamic Systems Using Interval Analysis

Numerical Methods. King Saud University

AN INTERVAL METHOD FOR INITIAL VALUE PROBLEMS IN LINEAR ORDINARY DIFFERENTIAL EQUATIONS

For δa E, this motivates the definition of the Bauer-Skeel condition number ([2], [3], [14], [15])

An average case analysis of a dierential attack. on a class of SP-networks. Distributed Systems Technology Centre, and

Calculation of maximum entropy densities with application to income distribution

Integration of Ordinary Differential Equations

Mathematical Constraint on Functions with Continuous Second Partial Derivatives

Bounds for nonlinear eigenvalue problems

Euler s Method applied to the control of switched systems

ON THE DIAMETER OF THE ATTRACTOR OF AN IFS Serge Dubuc Raouf Hamzaoui Abstract We investigate methods for the evaluation of the diameter of the attrac


Tetrational as special function

OUTLINE 1. Introduction 1.1 Notation 1.2 Special matrices 2. Gaussian Elimination 2.1 Vector and matrix norms 2.2 Finite precision arithmetic 2.3 Fact

460 HOLGER DETTE AND WILLIAM J STUDDEN order to examine how a given design behaves in the model g` with respect to the D-optimality criterion one uses

Review of Power Series

Discrete Projection Methods for Incompressible Fluid Flow Problems and Application to a Fluid-Structure Interaction

below, kernel PCA Eigenvectors, and linear combinations thereof. For the cases where the pre-image does exist, we can provide a means of constructing

Validated Explicit and Implicit Runge-Kutta Methods

Uniquely 2-list colorable graphs

Filter Banks with Variable System Delay. Georgia Institute of Technology. Atlanta, GA Abstract

1. Introduction Bottom-Up-Heapsort is a variant of the classical Heapsort algorithm due to Williams ([Wi64]) and Floyd ([F64]) and was rst presented i

Institute for Advanced Computer Studies. Department of Computer Science. On the Convergence of. Multipoint Iterations. G. W. Stewart y.

Math 1270 Honors ODE I Fall, 2008 Class notes # 14. x 0 = F (x; y) y 0 = G (x; y) u 0 = au + bv = cu + dv

Chapter 5 Exercises. (a) Determine the best possible Lipschitz constant for this function over 2 u <. u (t) = log(u(t)), u(0) = 2.

TOWARDS AUTOMATED CHAOS VERIFICATION


Extended Runge Kutta-like formulae

(1.) For any subset P S we denote by L(P ) the abelian group of integral relations between elements of P, i.e. L(P ) := ker Z P! span Z P S S : For ea

Ordinary Differential Equations

Journal of Computational and Applied Mathematics

Development of a New One-Step Scheme for the Solution of Initial Value Problem (IVP) in Ordinary Differential Equations

In Advances in Neural Information Processing Systems 6. J. D. Cowan, G. Tesauro and. Convergence of Indirect Adaptive. Andrew G.

Krylov Subspace Methods for the Evaluation of Matrix Functions. Applications and Algorithms

cells [20]. CAs exhibit three notable features, namely massive parallelism, locality of cellular interactions, and simplicity of basic components (cel

The Hopf bifurcation with bounded noise

Transcription:

Validating an A Priori Enclosure Using High-Order Taylor Series George F. Corliss and Robert Rihm Abstract We use Taylor series plus an enclosure of the remainder term to validate the existence of a unique solution for initial value problems in ordinary dierential equations and to compute a coarse enclosure of that solution. The signicance of this result is its application to Lohner's AWA algorithm for validated solutions, not to the theory of ordinary dierential equations. By using high-order Taylor series in Lohner's Algorithm I, we are able to validate the solution over much longer time steps than is done in the current AWA code. For Lohner's enclosure by polynomials, the enclosures are expensive to compute, but it is easy to check for enclosure. For our enclosure by Taylor series, the enclosures are free because they are already being computed, but checking for enclosure requires 2 n polynomial rootndings. Work is continuing on an implementation that will allow direct computational comparisons of the eectiveness of the two methods. Keywords: ordinary dierential equations, Lohner's algorithm, validated computation, Taylor series, enclosure methods. Introduction The AWA (Anfangswertaufgabe) program by Rudolf Lohner [4, 5, 6, 7, 8] computes an enclosure of the solution of an initial value problem (IVP) in ordinary dierential equations (ODE) u = f(u); u(t) = u; (1) where we only know an interval enclosure [u] of the vector u in general. Without loss of generality, Lohner assumes the system is autonomous only to simplify the proofs. We assume that f is at least p? 1 times continuously dierentiable in a domain D with [u] D IR n, p 2. Then there is a unique at least p times continuously dierentiable solution u(t) in a neighborhood of t. AWA is a single-step method. At each integration time step, AWA applies two algorithms: Algorithm I: (Existence and enclosure) Find a step size h and a coarse enclosure interval [u ] D such that for t 2 [t] := [t; t + h], the solution u(t) exists and satises u(t) 2 [u ].

2 George F. Corliss and Robert Rihm Algorithm II: (Tightening) Compute a tight enclosure for u(t) at t = t1 := t + h. Throughout this paper, we use the term \validation" to mean \prove existence of u(t) for some t and nd a coarse enclosure [u ]". The eciency of AWA has been limited by the Euler step size enforced by Algorithm I as used in the current program. We show that Taylor series plus an enclosure of the remainder term can be used not only in Algorithm II to tighten the enclosure, but also in Algorithm I for the validation step. The consequence of this result for AWA is that Algorithm I can validate the existence of the unique solution over time steps appropriate for high-order methods employed for tightening during Algorithm II. The basic idea of AWA is to enclose the Taylor coecients and the remainder term of the solution. It goes back to Moore who presented his method in 1965 (see [1, 11]). The Taylor coecients are computed using recurrence relations derived from the ODE: f () (u) := f(u); (u) := u; f (i) @f (i?1) (u) := @u f (u) ; i = 1; 2; : : :; p? 1 ; (u) i := 1 i f(i?1) (u); i = 1; 2; : : :; p : (2) The recurrences given by Equation (2) are not as intimidating as they might appear; this is just an expression of the computation of the Taylor coecients for the solution. They can be evaluated by using automatic dierentiation (see e.g. [12]). If we begin the recurrences given by Equation (2) with the initial value u and compute in exact arithmetic, we get the Taylor coecients (u) i for the solution expanded at t = t. If we begin the recurrences given by Equation (2) with an interval vector [u] and compute in rounded interval arithmetic, then we get enclosures ([u]) i of the Taylor coecients for all solutions starting from u 2 [u]. We can also begin with a coarse interval [u ] and compute the interval vectors ([u ]) i. Then (b t? t ) p ([u ]) p contains the remainder of the p? 1st degree Taylor polynomial of u(b t), if u(t) 2 [u ](t) for all t 2 [t ;b t]. We emphasize that the computation of these values does not pre-suppose the existence of a solution. We are just computing sequences of numbers. If we cannot compute any of these values, then we are not able to proceed with validation. However, if we know an a priori enclosure [u ] of the solution on [t], then p?1 u(t) 2 (t? t) i (u) i + (t? t) p ([u ]) p =: T p (t; u; [u ]) or p?1 u(t) 2 (t? t) i ([u]) i + (t? t) p ([u ]) p = T p (t; [u]; [u ]), resp. (3)

Validating an A Priori Enclosure 3 for t 2 [t], if only the interval polynomial T p stays in [u ]. Moore as well as Lohner use (modications of) this formula to tighten an a priori enclosure [u ] (see e.g. [13]). 1 Current AWA Approach Validation and the computation of a tight enclosure are separate, though related, issues. This paper addresses only the issue of validation, the role of Lohner's Algorithm I. The techniques of his Algorithm II for tightening the enclosure are discussed in [4, 5, 6, 7, 8]. Algorithm I uses the following theorem for validation. Theorem 1 ([6]) Let [u ] D be an interval vector satisfying [u 1 ] := u + [; h] f? [u ] [u ] : Then u(t) 2 [u 1 ] (and hence u(t) 2 [u ]) for t 2 [t] = [t; t + h]. It follows directly from Theorem 2 Let [u](t) and [v](t) D be interval vector valued functions, and let each p times continuously dierentiable function v(t) 2 [v](t) satisfy Z t (v(t)) := u + for t 2 [t]. Then the solution u of Equation (1) satises for t 2 [t]. t f(v()) d 2 [u](t) [v](t) (4) u(t) 2 [u](t) Proof: If (4) holds and we start a Picard-Lindelof iteration Z t u (k) (t) = (u (k?1) (t)) = u + t f(u (k?1) ()) d ; k = 1; 2; : : : with a p times dierentiable function u () (t) 2 [v](t), then every successor is again p times dierentiable and lies in [u](t). However, the sequence of Picard iterates should leave this interval function if the solution did so.

4 George F. Corliss and Robert Rihm It seems strange to let v(t) be p times dierentiable. However, this assumption is used in the following section. AWA nds a step size h and an interval [u ] such that [u ] [u 1 ] := [u] + [; h]f? [u ]? [u ] ; (5) for all t 2 [t] := [t; t+h], and hence validates that the solution of Equation (1) exists on [t] and is contained in [u ]. Figure 1 shows the a priori bound [u ] = [:2; :5] and ([u ]) for the logistic equation u = u(1? u), u() = :3, on the interval t 2 [; :6]. In this gure, ([u ]) [u ] in the interval t 2 [; :5]. If we tighten [u ] = [:29; :5], then ([u ]) [u ] in the larger interval t 2 [; :563]..6.5.4 u.3.2.1.1.2.3.4.5.6 t Figure 1. A priori bounds with [u ] = [:2; :5]. 2 Lohner's Enclosure of u (p) The allowable step size h for which validation can be done using Theorem 1 as described in Section 1 is limited to a step appropriate for Euler's method, no matter how high the order of the method used during Algorithm II to tighten the enclosure. Stetter showed in [14] that for the problem u = Au, A 2 IR nn, the step size h obtained by applying Theorem 1 may be multiplied by the degree p of the Taylor

Validating an A Priori Enclosure 5 polynomial used in Algorithm II. Algorithm II yields a validated tight enclosure for t = t + ph because p!([u ]) p = A p [u ] contains the p-th derivative u (p) (t) for all t 2 [t; t + ph], even though [u ] does not enclose the solution on this enlarged interval, in general. Lohner [9] suggests using coarse enclosures for higher derivatives also in the general case. Actually, he uses Theorem 2 instead of Theorem 1 and a special kind of interval polynomials for [v](t) instead of a constant coarse enclosure as in the current AWA program. Lohner guesses an enclosure [v ] of u (p) (t)=p! satisfying p?1 [v](t) := (t? t) i (u) i + (t? t) p [v ] D: The Picard-Lindelof operator reproduces the Taylor coecients of the exact solution. There is an interval vector [v 1 ] such that each p times dierentiable function v(t) 2 [v](t) satises p?1 (v(t)) 2 [u](t) := (t? t) i (u) i + (t? t) p [v 1 ]: If [v 1 ] [v ] holds, then we also have [u](t) 2 [v](t), and according to Theorem 2, u(t) 2 [u](t). To obtain [v 1 ], one has to compute an enclosure of the remainder coecient of the p? 2nd degree Taylor polynomial of f(v(t)) for all p times dierentiable functions v(t) 2 [v](t). It also contains the remainder coecient of the p? 1st degree Taylor polynomial of (v(t)). For computing or enclosing these coecients on a computer, one has to apply polynomial interval machine arithmetic given by Eiermann [3], which provides the enclosure of a xed number of Taylor coecients of standard operations and standard functions on interval polynomials as well as the enclosure of the respective remainders. The advantage of this idea of Lohner over the validation strategy discussed in Section 1 is that validation is possible for step sizes appropriate for a method of local order p. Example 1 The solution of the initial value problem u =?u; u() = 1 satises u(t) = e?t 2 [; 1] for t. However, the application of Theorem 1 for [u ] = [; 1] requires a step size 1 since u + [; h]f([u ]) [u ], 1 + [; h](?[; 1]) [; 1], h 1:

6 George F. Corliss and Robert Rihm However, Lohner's approach yields validation for much longer steps. p = 6 and [v ] = [; 1], we get If we choose [v](t) = 1? t + t2 2? t3 6 + t4 24? t5 12 + t6 [; 1]; 72 f([v](t)) =?[v](t)?1 + t? t2 2 + t3 6? t4 24 + t5 12 [1? h ; 1]; and 6 [u](t) = ([v](t)) = 1? t + t2 2? t3 6 + t4 24? t5 12 + t6 72 [1? h 6 ; 1] for t 2 [t; t + h]. We have [1? h 6 ; 1] [v ] = [; 1], h 6 : Of course, we could enclose the sixth Taylor coecient of f instead of the fth one to obtain h = 7. However, in general we only assume f to be p? 1 times dierentiable. In this example, the coecients of f(v(t)) can easily be calculated. However, there is no implementation of the method for the general case yet, although Lohner has described [9] such an implementation using the Eiermann operators for interval polynomial arithmetic. 3 Validated Enclosure Using Taylor Series The use of Taylor series for validation, as well as for tight enclosure, is suggested by pictures such as those in Figure 2, which shows Taylor series enclosures for the solution of the Lorenz system. 2 18 16 14 12 1 8 6 4 2.1.2.3.4.5 t

Validating an A Priori Enclosure 7 3 2 1.1.2.3.4.5 t -1 4 3 2 1.1.2.3.4.5 t Figure 2. Taylor Series Enclosures for the Solution of the Lorenz System. As already mentioned, AWA uses the recurrence relations (2) and the formula (3) for computing a tight enclosure in Algorithm II after having validated a coarse enclosure [u ] for a step size h in Algorithm I. It would cause no additional costs if the Taylor expansion (3) could also be used in Algorithm I. The following theorem shows that this can actually be done. Theorem 3 Let u 2 int([u ]) D. Let for t 2 [t]. Then p?1 [u](t) := (t? t) i (u) i + (t? t) p ([u ]) p [u ] ; (6) u(t) 2 [u](t) for t 2 [t] : Proof: It follows from our assumptions that Equation (1) has a solution u in some neighborhood of the point (t; u), for every u 2 [u]. We must prove that that neighborhood includes all of [t]. If (6) holds, then u(t) 2 [u](t) as long as u(t) 2 [u ].

8 George F. Corliss and Robert Rihm Assume that u(t) leaves [u](t) =: [u(t); u(t)], and hence [u ] at b t 2 int([t ]). Without loss of generality, we assume u j (b t) = uj (b t) = sup[u ]j, for some j 2 f1; : : :; ng. From u (p) (t) 2 p!([u ]) p, it follows that p?2 u (t) 2 (t? t) i (u) i + (t? t) p?1 p([u ]) p ; and hence u (t) u (t) on [t;b t]. Therefore, uj (b t) = uj (b t) implies uj (t) = u j (t) for all t 2 [t;b t]. If uj is constant, then we have u = u j (b t) = sup[u ]j, which contradicts the assumption u 2 int([u ]). Otherwise, u j (b t) is an isolated maximum of the p-th degree polynomial u j. The same holds for u j (t), since this function is at least p times continuously dierentiable in [t]. Hence, u(t) cannot leave [u](t) at b t. This approach avoids complicated calculations and nevertheless takes the order p into account. With p = 6, we can validate the coarse enclosure [; 1] in Example 1 for h = 2:1, compared with h = 1 for a constant bound enclosure and h = 6 using Lohner's polynomial enclosure. Example 2 We want to validate the solution u(t) = 1=t of the initial value problem u =?u 2 ; u(1) = 1 in the interval [u ] = [; 2]. For the constant bound enclosure, we have [u 1 ] := 1 + [; h](?[; 2] 2 ) = [1? 4h; 1] [u ], h 1=4 as the maximal step size we can reach by applying Theorem 1. However, the approach of this section for p = 2 yields a longer step: [u](t) = 1? (t? 1) + [; 8](t? 1) 2 [; 2], h 1 + p 33 16 4 Implications for AWA Using the results of the previous section, AWA Algorithm I is :42 : Task 1: Approximate solution. Compute Taylor coecients of an approximate solution p?1 bu(t) := (t? t) i (u) i from recurrence relation (2). In practice, we must compute (u) i using rounded interval arithmetic to capture any round-o errors. The work of Task 1 is already being done in AWA Algorithm II.

Validating an A Priori Enclosure 9 Task 2: Guess a step h. Use heuristics based on the radius of convergence [1] of the truncated series for bu or on previous steps sizes. Task 3: Guess a constant enclosure [u ]. Bound the range of bu([t; t+h]), and inate it a little. Task 4: Compute the remainder terms ([u ]) i, for i = ; : : : ; p from the recurrence relation (2). The work of Task 4 is already being done in AWA Algorithm II. Task 5: Compute stepsize for validation. The stepsize is the point nearest to t (and in [t]) at which p?1 [u](t) := (t? t) i (u) i + (t? t) p ([u ]) p leaves [u ]. This requires 2 n polynomial rootndings. For each polynomial rootnding problem, it is sucient to compute only relatively coarse lower bounds for the smallest real root in the interval (t; t + h], so a few iterations of a simplied interval Newton method should suce. Remark. Variable order. We must compute ([u ]), : : :, ([u ]) p?1 in order to compute ([u ]) p. Hence, it is relatively inexpensive to compute the step size in Task 5 for each order p. The order used for validation in Algorithm I can be dierent from the order used for tightening in Algorithm II. The information is readily available to make both Algorithms I and II fully variable order. Remark. Iteration. If Task 5 nds that [u](t) [u ] on [t], then we can repeat Tasks 4 and 5 with a larger h or else with [u ] = [u]([t]). Remark. Optimization. Given guesses for h from Task 2 and for [u ] from Task 3, we compute in Task 5 the largest step for which we can validate existence and containment. Numerical experiments have shown [2] that a carefully chosen [u ] allows step sizes 1 times as long as step sizes corresponding to [u ] given by reasonable heuristics. That suggests viewing Algorithm I as an optimization problem: Maximize step size h by varying guessed h and [u ] p?1 subject to [u](t) := (t? t) i (u) i + (t? t) p ([u ]) p [u ] Remark. Eectiveness. Either Lohner's enclosure by polynomials or our enclosure by Taylor series is signicantly more expensive than Picard-Lindelof iteration of constant bounds currently used by AWA. However, AWA Algorithm II involves the solution of an ODE system of dimension n n, so almost any eort we expend in Algorithm I that allows AWA to take longer steps improves the overall eciency of AWA.

1 George F. Corliss and Robert Rihm Remark. Which is better? For Lohner's enclosure by polynomials, the enclosures are expensive to compute, but it is easy to check for enclosure. For our enclosure by Taylor series, the enclosures are free because they are already being computed, but checking for enclosure requires 2n polynomial rootndings. Work is continuing on an implementation that will allow direct computational comparisons of the eectiveness of the two methods. Acknowledgments We thank the anonymous referee for helpful suggestions. The work of Corliss was supported in part by National Science Foundation Grant No. DMS{9413525 and in part by SUN Academic Equipment Grant No. EDUD{US{9428. References [1] Y. F. Chang and G. Corliss, Solving Ordinary Dierential Equations Using Taylor Series, ACM Trans. Math. Software, 8(1982), 114{144. [2] G. Corliss, Validating an A Priori Enclosure Using High-Order Taylor Series, presented at SCAN '95, Wuppertal, 1995. [3] M. Ch. Eiermann, Adaptive Berechnung von Integraltransformationen mit Fehlerschranken, PhD thesis, Universitat Freiburg, 1989. [4] R. J. Lohner, Anfangswertaufgaben im IR n mit kompakten Mengen fur Anfangswerte und Parameter, Diplomarbeit, Inst. f. Angew. Math., Universitat Karlsruhe, 1978. [5] R. J. Lohner, Enclosing the Solutions of Ordinary Initial and Boundary Value Problems, in Computer Arithmetic: Scientic Computation and Programming Languages, E. W. Kaucher, U. Kulisch, and C. Ullrich, eds., Wiley-Teubner Series in Computer Science, Stuttgart, 1987, pp. 255{286. [6] R. J. Lohner, Einschlieung der Losung gewohnlicher Anfangs{ und Randwertaufgaben und Anwendungen, PhD thesis, Universitat Karlsruhe, 1988. [7] R. J. Lohner, Interval Arithmetic in Staggered Correction Format, in Scientic Computing with Automatic Result Verication, E. Adams and U. Kulisch, eds., Academic Press, San Diego, 1993. [8] R. J. Lohner, On Step Size and Order Control in the Veried Solution of Ordinary Initial Value Problems, presented at SCAN '93, September 1993, Vienna.

Validating an A Priori Enclosure 11 [9] R. J. Lohner, Step Size and Order Control in the Veried Solution of IVP with ODE's, presented at the SciCADE '95 International Conference on Scientic Computation and Dierential Equations, Stanford, Calif., March 28 { April 1, 1995. [1] R. E. Moore, The Automatic Analysis and Control of Error in Digital Computation Based on the Use of Interval Numbers, in Error in Digital Computation, Vol. I, L. B. Rall, ed., John Wiley and Sons, New York, 1965, pp. 61{13. [11] R. E. Moore, Interval Analysis, Prentice-Hall, Englewood Clis, N.J., 1966. [12] L. B. Rall, Automatic Dierentiation: Techniques and Applications, vol. 12 of Lecture Notes in Computer Science, Springer Verlag, Berlin, 1981. [13] R. Rihm, Interval Methods for Initial Value Problems in ODEs, in Topics in Validated Computations, J. Herzberger, ed., Elsevier Science B.V., Amsterdam, 1994, pp. 173{27. [14] H. J. Stetter, Validated Solution of Initial Value Problems for ODE, in Computer Arithmetic and Self-Validating Numerical Methods Ch. Ullrich, ed., Academic Press, San Diego, 199, pp. 171{186. Addresses: G.F. Corliss, Marquette University, Department of Mathematics, Statistics, and Computer Science, P.O. Box 1881 Milwaukee, Wisconsin 5321-1881, USA. georgec@mscs.mu.edu R. Rihm, Universitat Karlsruhe, Institut fur Angewandte Mathematik, 76128 Karlsruhe, Germany. robert.rihm@math.uni-karlsruhe.de