Study on Truncation and Roundoff Errors in Numerical Methods

History is full of many examples where errors in numerical calculation have played important role or sometimes proved fatal also. (e.g. The Patriot and the Scud [6], The short flight of the Ariane 5 [7], The Vancouver Stock Exchange [8,9,10], Parliamentary elections in Schleswig-Holstein [11]). So in this paper we review two types of errors occurred while performing calculations: first is truncation error and second is round off error . Examples are given in support of the theory part.

COMPUTER TOOLS FOR SOLVING MATHEMATICAL PROBLEMS: A REVIEW

The rapid development of digital computer hardware and software has had a dramatic influence on mathematics, and contrary. The advanced hardware and modern sophistical software such as computer visualization, symbolic computation, computerassisted proofs, multi-precision arithmetic and powerful libraries, have provided resolving many open problems, a huge very difficult mathematical problems, and discovering new patterns and relationships, far beyond a human capability. In the first part of the paper we give a short review of some typical mathematical problems solved by computer tools. In the second part we present some new original contributions, such as intriguing consequence of the presence of roundoff errors, distribution of zeros of random polynomials, dynamic study of zero-finding methods, a new three-point family of methods for solving nonlinear equations and two algorithms for the inclusion of a simple complex zero of a polynomial.

Rigorous Roundoff Error Analysis of Probabilistic Floating-Point Computations

We present a detailed study of roundoff errors in probabilistic floating-point computations. We derive closed-form expressions for the distribution of roundoff errors associated with a random variable, and we prove that roundoff errors are generally close to being uncorrelated with their generating distribution. Based on these theoretical advances, we propose a model of IEEE floating-point arithmetic for numerical expressions with probabilistic inputs and an algorithm for evaluating this model. Our algorithm provides rigorous bounds to the output and error distributions of arithmetic expressions over random variables, evaluated in the presence of roundoff errors. It keeps track of complex dependencies between random variables using an SMT solver, and is capable of providing sound but tight probabilistic bounds to roundoff errors using symbolic affine arithmetic. We implemented the algorithm in the PAF tool, and evaluated it on FPBench, a standard benchmark suite for the analysis of roundoff errors. Our evaluation shows that PAF computes tighter bounds than current state-of-the-art on almost all benchmarks.

A Two-Phase Approach for Conditional Floating-Point Verification

Tools that automatically prove the absence or detect the presence of large floating-point roundoff errors or the special values NaN and Infinity greatly help developers to reason about the unintuitive nature of floating-point arithmetic. We show that state-of-the-art tools, however, support or provide non-trivial results only for relatively short programs. We propose a framework for combining different static and dynamic analyses that allows to increase their reach beyond what they can do individually. Furthermore, we show how adaptations of existing dynamic and static techniques effectively trade some soundness guarantees for increased scalability, providing conditional verification of floating-point kernels in realistic programs.

Iterative method to the coupled operator matrix equations with sub-matrix constraint and its application in control

A finite iterative algorithm is presented for solving the numerical solutions to the coupled operator matrix equations in Zhang (2017b). In this paper, a new finite iterative algorithm is presented for solving the constraint solutions to the coupled operator matrix equations [Formula: see text], where the constraint solutions include symmetric solutions, bisymmetric solutions and reflexive solutions as special cases. If this system is consistent, for any initial constraint matrices, the exact constraint solutions can be obtained by the introduced algorithm within finite iterative steps in the absence of the roundoff errors. Also, if this system is not consistent, the least-norm constraint solutions can be obtained within the finite iteration steps in the absence of the roundoff errors. Furthermore, if a group of suitable matrices are given, the optimal approximation solutions can be derived. Finally, several numerical examples are given to show the effectiveness of the presented iterative algorithm.

Application of the auxiliary performance index for automatic optimality control of discrete Kalman filter

Предложен новый метод автоматического контроля оптимальности дискретного фильтра Калмана, основанный на равенстве нулю градиента вспомогательного функционала качества (ВФК) по параметрам адаптивного дискретного фильтра. Для вычисления градиента ВФК применяется численно устойчивый к ошибкам машинного округления алгоритм модифицированной взвешенной ортогонализации Грама-Шмидта (MWGS-ортогонализации). Алгоритм реализован на языке Matlab. Результаты проведенных численных экспериментов подтверждают эффективность предложенного метода The paper proposes a new method for automatic control of the nominal operating mode of a dynamic stochastic system, based on a combination of two previously developed methods: the auxiliary performance index (API) method and the LD modification of an adaptive filter numerically robust to roundoff errors. The API method was previously developed to solve the problems of identification, adaptation, and control of stochastic systems with control and filtering. We suggest using the API not only as a tool for identifying the parameters of the stochastic system model from the measurement data but also for automatically monitoring the optimality of the adaptive filter, namely, the condition that the API gradient is close to zero should be satisfied (with the necessity and sufficiency) at the point corresponding to the optimal value of the vector parameter in the adaptive Kalman filter. The main result is the new eLD-KF-AC algorithm (extended LD Kalman-like adaptive filtering algorithm with automatic optimality control). The advantages of the obtained solution are as follows: 1) the choice of the adaptive filter structure in the form of an extended LD algorithm can significantly reduce the effect of machine roundoff errors on the calculation results when supplemented by the ability to calculate the sensitivity functions by the system vector parameter of the adaptive filter; 2) the application of the API method allows controlling the optimality of the adaptive filter by the condition that the API gradient is zero at the minimum point, which corresponds to the optimal value of the parameter in the adaptive filter; 3) the calculation of the API gradient in the adaptive extended LD filter does not require significant computational costs and such a control method can be carried out in real-time. The results of the work will be applied to solving problems of joint control and identification of parameters in the class of discrete-time linear stochastic systems represented by equations in the state-space form.

Modified CGLS iterative algorithm for solving the generalized Sylvester-conjugate matrix equation

By introducing the real inner product, this paper offers an modified conjugate gradient least squares iterative algorithm (MCGLS)for solving the generalized Sylvester-conjugate matrix equation. The properties of this algorithm are discussed and the finite convergence of this algorithm is proven. This new iterative method can obtain the symmetric least squares Frobenius norm solution within finite iteration steps in the absence of roundoff errors. Finally, two numerical examples are offered to illustrate the effectiveness of the proposed algorithm.

Counterexample- and Simulation-Guided Floating-Point Loop Invariant Synthesis

We present an automated procedure for synthesizing sound inductive invariants for floating-point numerical loops. Our procedure generates invariants of the form of a convex polynomial inequality that tightly bounds the values of loop variables. Such invariants are a prerequisite for reasoning about the safety and roundoff errors of floating-point programs. Unlike previous approaches that rely on policy iteration, linear algebra or semi-definite programming, we propose a heuristic procedure based on simulation and counterexample-guided refinement. We observe that this combination is remarkably effective and general and can handle both linear and nonlinear loop bodies, nondeterministic values as well as conditional statements. Our evaluation shows that our approach can efficiently synthesize loop invariants for existing benchmarks from literature, but that it is also able to find invariants for nonlinear loops that today’s tools cannot handle.

Toward the Synthesis of Gauss Pivoting Code for Linear Systems Resolution : Application Mechanical Problems

The purpose of this talk is primarily to introduce a new methodology to synthesize numerically accurate programs for the Gaussian elimination method in order to solve linear systems coming from mechanical problems. The synthesis is based on program transformation techniques and it is guided in its estimation of accuracy by interval arithmetic that computes the propagation of roundoff errors. Besides a discussion on numerical accuracy issues related to floating-points arithmetics and roundoff errors, we present our approach used to compute the error bound during the resolution process. Finally, some experimental results will be presented to prove the efficiency of our synthesizer tool and show that the specialized produced code to solve the family of systems given in input is far more accurate and faster than the standard implementation of the Gauss method.