Numerical Solution for the Extrapolation Problem of Analytic Functions

Numerical Solution for the Extrapolation Problem of Analytic Functions

PDF

Nikolaos P. Bakas

Research. Vol 2019 Article ID 3903187

Less

Research. Vol 2019 Article ID 3903187

• Research Article •

Numerical Solution for the Extrapolation Problem of Analytic Functions

Full

Nikolaos P. Bakas

Affiliations

Intelligent Systems Lab & Civil Engineering Department, School of Architecture, Engineering, Land and Environmental Sciences, Neapolis University Pafos, 2 Danais Avenue, 8042 Paphos Cyprus

Published: 2019-05-20 doi: 10.34133/2019/3903187

Outline

Abstract

Less

In this work, a numerical solution for the extrapolation problem of a discrete set of n values of an unknown analytic function is developed. The proposed method is based on a novel numerical scheme for the rapid calculation of higher order derivatives, exhibiting high accuracy, with error magnitude of O(10⁻¹⁰⁰) or less. A variety of integrated radial basis functions are utilized for the solution, as well as variable precision arithmetic for the calculations. Multiple alterations in the function’s direction, with no curvature or periodicity information specified, are efficiently foreseen. Interestingly, the proposed procedure can be extended in multiple dimensions. The attained extrapolation spans are greater than two times the given domain length. The significance of the approximation errors is comprehensively analyzed and reported, for 5832 test cases.

Cite this Article

Nikolaos P. Bakas. Numerical Solution for the Extrapolation Problem of Analytic Functions[J]. Research, 2019 , 2019 (5) : 3903187 - . DOI: 10.34133/2019/3903187

Full Text

Less

1. Introduction

Less

Prediction always differs from observation [1], and extrapolation remains an open challenge [2] or even a hopelessly ill-conditioned [3] problem, in a vast variety of scientific disciplines. It relies on numerical methods which attempt to predict the future, unknown values of a studied phenomenon, given a limited set of observations. Its importance is reflected in the Scopus database [4], by a search with the term extrapolation, yielding nearly 150 thousands of research items, in a wide range of scientific fields (Table S1). The visual presentation of the corresponding keywords (Figure S1) in a bibliometric map [5] reveals that extrapolation is closely associated with modeling, simulation, uncertainty, and interpolation, which is the reciprocal problem to simulate the unknown mechanism which produced the given data, aiming to predict utilizing the constructed model. Contradictorily to interpolation and despite the significance of extrapolation and forecasting methods, they exhibit a decreasing pattern in literature since the early 1980s (Figure S2), indicating the futility to predict for extended time frames. The given data is often called time-series and their extrapolation forecasting, which usually depend on machine learning algorithms such as support vector machines [6–8] and artificial neural networks [2, 9, 10], which construct nonlinear regression models, to predict beyond the known domain. Prediction algorithms are useful in a wide range of scientific disciplines, such as earth sciences [7, 9], finance [8, 11], computer science [12, 13], and engineering [14–16]. Several methods have been proposed for time-series forecasting [2] and competitions regarding accuracy have been conducted, utilizing statistical [11, 17] and machine learning procedures [2, 10, 15]; however, the prediction horizon regards only a small percentage (~20-30%) of the given data extension. Similarly, for less uncertain problems as the extrapolation of curves defined by polynomials (splines) [18–20], extrapolation can cause unpredictable results, and their extension should be short [21]. Apparently, prediction procedures are essential for a vast variety of scientific fields, always based on numerical interpolation methods.

More specifically, in the case of analytic functions, where the given data follow an unknown rigor mathematical law, any extrapolation results are too weak to be interesting [3]. Previous theoretical works focus on the interpolation problem, such as the ability of neural networks with one hidden layer to approximate analytic functions [22], proofs regarding the degree of rational approximation of an analytic function [23], and analysis of the rate of decrease of the best approximations of analytic functions [24]. Accuracy and convergence techniques are demonstrated in [25], yet it is shown theoretically [26] that the approximation of analytic functions cannot converge exponentially. Approximation of analytic functions is investigated utilizing a variety of approximators [27–29], such as Hermite [30, 31] and Airy functions [32]. Computational works regard the approximation of a function [33] as well as its derivatives [34, 35], investigating the interpolation errors among the given nodal points. Such errors are higher than the approximation errors (Runge phenomenon) especially at the boundaries [33], affecting dramatically the extrapolation outside the given domain. Moreover, the round-off errors of computing machines follow some stochastic law [36] and recent works deal with the high impact of infinitesimal errors in the given data, utilizing extended arithmetic precision [37, 38]. Accordingly, the constitution of an accurate numerical method to approximate an analytic function and its derivatives, which is vital for the extrapolation problem, remains a challenge.

The purpose of this study was to provide a generic numerical solution for the extrapolation problem. It was attained for extended extrapolation horizons of even greater than 200% the given domain length if the set of data are derived from an unknown analytic function and their precision is high. The rationale of the proposed method adheres to the following three stages: (a) interpolation of the set of values

(f 1, f 2, …, f N)

at specified points

(x 1, x 2, …, x N)

using integrated radial basis functions (IRBFs), (b) computation of high order derivatives

{f ′, f ′ ′, f (3), …}

at any point of the closed domain

[x 1, x N]

with high accuracy based on a novel numerical scheme, and (c) successive application of the Taylor series formula to extrapolate

f

at points outside the given domain. The method is capable of interpolating within the given data with high precision, avoiding the Runge phenomenon at the boundaries [33], as well as of computing the higher order derivatives with remarkable accuracy, which are fundamental problems in numerous applications. The effects of the method’s parameters on the prediction errors were extensively investigated by their analysis of variance (ANOVA) and feature extraction utilizing the Random Forest [39] method, for 5832 test cases. Illustrative examples of highly nonlinear analytic functions in two and three dimensions demonstrate the prediction extents attained by the proposed method.

2. Numerical Solution

Less

2.1. Extrapolation of One Infinitesimal dx

Let

f

be an analytic function, which is unknown. It is given that the function takes values

b = (f 1, f 2, …, f N)

at specified points

x = (x 1, x 2, …, x N)

as in Figure 1(a), for a generic analytic function. The extrapolation problem, that is, to predict the values which take the function outside the given domain

[x 1, x N]

, can be transformed into the computation of

n

ordinary derivatives of

f

(f ′, f ′ ′, …, f (n))

, at point

x N

. In particular, applying the Taylor formula from the point

x N

to the next point

x N + 1 = x N + d x

, the corresponding value of

f

is given by

(1) f x N + d x = f x N + d x f ′ x N + d x 2 2 f ′ ′ x N + ⋯ + d x n n! f n x N,

for an infinitesimal step

d x

f (x N + d x)

is the value of

f

at point

x N + d x

, depending on the

n

ordinary derivatives of

f

x N

. These

n

derivatives should be computed, as they are unknown.

2.2. Numerical Evaluation of the Derivatives

Around the endpoint

x N

, we used the Taylor formula for computing the value of the function at the first predicted point

x N + 1

. In order to calculate the derivatives, we need to consider the function values within the end interval

d x

from the point with abscissa

x N - 1

x N

. Accordingly, the Taylor expansion from the endpoint

x N

to any previous point

h j → x N

(Figure 1(b)) may also be written in the form

(2) f x N - h j = f x N - h j f ′ x N + h j 2 2 f ′ ′ x N - ⋯ + - h j n n! f n x N .

By interchanging the positions of

f x N

and

f x N - h j

, we may write

(3) - f x N = - f x N - h j - h j f ′ h j + h j 2 2 f ′ ′ h j - ⋯ + - h j n n! f n h j,

and by applying it to

n + 1

points

h j w i t h i n t h e e n d i n t e r v a l

dx (Figure 1(b)) and writing the resulting system of equations in matrix form, we obtain

(4) - f x N 11 ⋮ 1 = - f x N - h 1 - h 1 f ′ x N + h 1 2 2 f ′ ′ x N - ⋯ + - h 1 n n! f n x N - f x N - h 2 - h 2 f ′ x N + h 2 2 2 f ′ ′ x N - ⋯ + - h 2 n n! f n x N ⋮ - f x N - h n + 1 - h n + 1 f ′ x N + h n + 1 2 2 f ′ ′ x N - ⋯ + - h n + 1 n n! f n x N .

It is important to note that all the derivatives are considered at the endpoint

x N

. Hence, extracting the derivatives as a column vector, we have

(5) - f x N 11 ⋮ 1 = - f x N - h 1 - h 1 h 1 2 2 ⋯ - h 1 n n! - f x N - h 2 - h 2 h 2 2 2 ⋯ - h 2 n n! ⋮ ⋮ ⋮ ⋱ ⋮ - f x N - h n + 1 - h n + 1 h n + 1 2 2 ⋯ - h n + 1 n n! 1 f ′ x N f ′ ′ x N ⋮ f n x N .

Thus, only the first column of the matrix in (5) varies as it refers to the values of f within the edge interval. As the objective is to compute the n ordinary derivatives of f, (5) is separated in submatrices, in order to rearrange the equations and develop a formula to calculate the vector D, containing the values of the n derivatives of f at point

x N

. The solid lines indicate the submatrices B, C, H, D, and

J n, 1

; hence, (5) can be written in matrix partitions form

(6) - f x N 1 J n, 1 = A B C H 1 D,

where

6 a A = - f x N - h 1, 6 b B = - h 1 h 1 2 2 ⋯ - h 1 n n!, 6 c C = - f x N - h 2 ⋯ - f x N - h n + 1 T, 6 d H = - h 2 h 2 2 2 ⋯ - h 2 n n! ⋮ ⋮ ⋱ ⋮ - h n + 1 h n + 1 2 2 ⋯ - h n + 1 n n!, 6 e D = f ′ x N f ′ ′ x N ⋯ f n x N T, 6 f J n, 1 = 1 ⋯ 1 T .

Since (6) is a system of

n + 1

equations, in order to calculate the vector D (containing the values of the derivatives of

f

at point

x N

), by (6), we deduce

(7) - f x N J n, 1 = H D + C,

which is a system of n equations. Thus, the matrix of the n ordinary derivatives can be directly computed by

(8) D = H - 1 - C - f x N J n, 1,

where n, and hence the number of the calculated derivatives, can be arbitrarily selected. Equation (8) offers a direct computation of the n ordinary derivatives D. Since C is still unknown, we can compute it, by interpolating f with IRBFs as defined subsequently.

2.3. Function Approximation with IRBFs

Radial Basis Functions (RBFs) networks are universal approximators while Integrated RBFs are capable of attaining precise approximation for a function and its derivatives [34, 35] as well as for the solution of partial differential equations [37, 38]. The essential formulation of RBFs, adopted in this work, is as follows.

Let

x

be a variable taking

N

values in a domain

X

. The known values of the function are given at the positions

x 1, x 2, …, x N

as per Figure 1, while

∀ i ∈ (1,2, …, N - 1)

the interval length between two adjacent points

x i

x i + 1

is equal to

d x

. The given values

f (x i)

of the unknown function

f

can be approximated with RBFs

φ (x)

, centered at the N points. The RBFs approximation is defined by

(9) f x i ≅ ∑ j = 1 N a j φ r i j,

where

α j

are the unknown weights for the approximation with RBFs,

r i j = x i - x j

represents the distance of a given point

x i

in the domain from any other point

x j

, and

φ (r i j)

is the radial basis kernels such as Gaussian, Shifted Multiquadrics, and Sigmoid or their integrals, as in 14a, 14b, 14c, 14d, 14e, 14f. Applying (9) at the

N

given points results in the approximated values of

f

b i = f (x i)

; hence we deduce that

(10) b i = ∑ j = 1 N a j φ r i j .

If we write (10)

N

times

(i = 1, …, N)

and define

Φ

as a matrix containing the values of

φ (r i j)

, with i indicating the rows and j the columns of

Φ

α = α 1, α 2, …, α N T

a vector containing the unknown approximation weights, and

b = b 1, b 2, …, b N T

a vector comprising the values of

f

at the nodal points

x N

, we may write

(11) b = Φ α .

Hence we derive

(12) α = Φ - 1 b .

The vector b is known, and the matrix

Φ

can be calculated from the distances of the known points

x i

and

x j

, so the weights

α

of the RBFs approximation of the function

f

are computed by numerically inverting matrix

Φ

and multiplying it with vector b. After the computation of vectors

α

, the radial basis functions

φ (r)

can be recomputed for each intermediate point

h j

, in order to approximate the values of the unknown function

f

at these points. Accordingly, each approximation vector

φ j

which corresponds to the points

h j

φ j = φ (x i - h j)

, when multiplied with the calculated weights

α

of (12), results in the interpolated values

f (x N - h j)

of 6c, and so

(13) C = φ 2 φ 3 ⋯ φ n + 1 T α,

where

φ j

are the approximation vectors for each h_j and

α

the RBFs weights. The vectors

φ j

are computed utilizing 14a, 14b, 14c, 14d, 14e, 14f, resulting in the vector C, which is the function’s f approximation at points h_j (Figure 1(b)).

The definitions of the radial basis functions φ utilized in (12), (13) are specified in the following group of 14a, 14b, 14c, 14d, 14e, 14f: the Gaussian and the Shifted Logarithmic, as well as integrations of them, for one and two times. Equations 14a, 14b, 14c, 14d, 14e, 14f are written concerning

r = x - x j

, in order to be generic for any

x ∈ [x 1, x N]

and hence for points x_i and h_j.

The Gaussian RBF is defined by

14 a φ r = e - r 2 / c 2,

where c is a constant influencing the shape of φ [40–42]. By integration of the Gaussian kernel, we obtain

14 b φ r = ∫ e - r 2 / c 2 d x = c 2 π erf ⁡ x - x j c,

and if we integrate for a second time, then

14 c φ r = ∫ c 2 π erf ⁡ x - x j c d x = c 2 e - x - x j 2 / c 2 + c π x - x j erf ⁡ x - x j / c 2,

where erf is the error function,

erf ⁡ (x) = 1 / π ∫ - x x e - t 2 d t

, exhibiting a sigmoid scheme, which is commonly used in ANNs [43].

Respectively, the Shifted Logarithmic RBF is defined by

14 d φ r = ln ⁡ r 2 + c 2,

and by integration of the kernel we obtain

14 e φ r = ∫ ln ⁡ r 2 + c 2 d x = ln ⁡ x - x j 2 + c 2 x - x j - 2 x + 2 c a tan ⁡ x - x j c,

and if we integrate for a second time, then

14 f φ r = ∫ l n x - x j 2 + c 2 x - x j - 2 x + 2 c tan - 1 ⁡ x - x j c d x = x x j - x - x j c 2 + x 2 + x j 2 ln ⁡ x - x j 2 + c 2 - 3 x 2 2 + 2 c x - x j tan - 1 ⁡ x - x j c - x x j ln ⁡ x - x j 2 + c 2 .

2.4. Iterative Implementation of Taylor Series Expansion

Exploiting the computed derivatives with high accuracy at the end of the domain

x N

, the predicted value of the function can be computed at the next step

x N + 1

, by the Taylor series expansion of (1). This procedure is repeated iteratively, for each next point

x N + 2, x N + 3, …

utilizing each time the shifted vectors

15 a b ← b 2 : N f x N + d x, 15 b x ← x 2 : N x N + d x .

The N points are initially the positions where the function values are given. This procedure is iteratively applied using at each step the

N - 1

previous values of the function (eliminating each time the first point) and one new value of the function, calculated by the procedure (15a and 15b). Keeping constant the length dx between a predicted point and its subsequent, the matrix

Φ

remains the same, as in the initial interpolation domain. The same stands for the inverted matrix

Φ - 1

as well as the matrices

φ j

, and despite the sequential shifts outside the original domain, the relative distances r of 14a, 14b, 14c, 14d, 14e, 14f remain the same. The same stands for the matrices H and H^-1. Hence, the interpolation of the function within the end increment dx each time is accomplished by matrix multiplication and not inversion, which is performed only one time for

Φ

and one time for H. Utilizing this approach, during the iterations of the extrapolation procedure, the computational time is decreased dramatically, as these matrices are computed only once.

2.5. The Extrapolation Algorithm

In Algorithm 1, the procedure of extrapolating given values of f is demonstrated in algorithmic form, utilizing at each step the appropriate equations as formulated in Sections 2.1–2.4.

Algorithm 1: Extrapolating given data of an unknown function.

3. Results

Less

3.1. Parameters Affecting the Method and Preparation of Dataset to Test Performance

The efficiency of the procedure was verified through numerical examples for a variety of highly nonlinear functions and their extrapolation spans. The extrapolation span is based on the accuracy of the interpolation method ((9), (10), (12), (13), 14a, 14b, 14c, 14d, 14e, 14f) as well as the numerical differentiation ((1), (5), (6), 6a, 6b, 6c, 6d, 6e, 6f, (8)). As demonstrated in Figure 1(b), the intervals for the differentiation h_j are selected near the

N - 1 s t

node of the domain discretization, in order to achieve the highest possible accuracy, because this region is within the selected dx, and simultaneously, the interpolation error is minimized near this node, due to the Runge phenomenon [33]. Hence we select

16 h j = d x - j - 1 d x 1 0 l,

where l indicates the limit distance near the node

x N - 1

. However, for extremely low values of h, the matrix H 6d contains elements near zero, and its inversion grows into unstable one (high condition number and inversion errors).

The features which have an effect on the calculations, as well as their values (in parentheses) utilized in the parametric investigation, are the number of Taylor terms utilized, indicated as number of derivatives, (25, 50, 75); the number of computer digits used for the arbitrary precision calculations (500, 1000, 2000); the span of the given domain L (1/2, 1); the number of the divisions N (50, 100, 200); the limit l (5, 10, 20); the number of IRBFs integrations (0, 1, 2); the kernel of RBFs networks (Gaussian = 1, Shifted Logarithmic = 2); their shape parameter c (1/5, 1, 5); and the unknown function which is studied (

sin ⁡ (x)

e c o s (x)

). Accordingly, the resulting database consists of 5832 records of test cases. The output of each parametric execution of the proposed procedure was the condition number of the H and

Φ

matrices, their inversion errors, and the error of the first and second derivatives according to the proposed procedure. Moreover, the error of the first step of the extrapolation was computed (ε), in order to investigate its dependence on the problem statement parameters (N, L, dx, f) as well as the method’s attributes (#derivatives, digits, kernel, c, #integrations).

3.2. Effects of Parameters on Extrapolation Accuracy and Computational Time

The

ε ′ = log 10 ⁡ ε + 10 - 323

was utilized as a measure for the extrapolation error, as the values of

ε

exhibit a variation within the domain

4.623 ∗ 10 - 283

to the IEEE® arithmetic representation for positive infinity [44]. Given that the lower values of error

ε ′

are important, the analysis of variance was conducted for the cases with values of

ε < 10 - 50

; hence the resulting database consists of 2292 records. ANOVA found statistically significant differences between the means (MD) of

ε ′

for 25 and 50 derivatives (MD = 43.5803, p-value = 9.5606

∗

10⁻¹⁰) and for 25 and 75 derivatives (MD = 54.2206, p-value =

9.5606 ∗ 10 - 10

) as demonstrated in Figure S3 and Table S2. Similarly, the number of digits (Figure S4 and Table S3) was found significant for the

ε ′

(MD = 20.3576, p-value = 9.6963

∗

10⁻¹⁰ for 500 to 1000 digits and MD = 33.9628, p-value = 9.5606

∗

10⁻¹⁰ for 500 to 2000 digits). The number of divisions exhibits a clear difference as well, in Figure S5 and Table S4 (MD = 22.7727, p-value = 9.5606

∗

10⁻¹⁰ for 50 to 100 divisions and MD = 49.6978, p-value = 9.5606

∗

10⁻¹⁰ for 50 to 200 divisions). The univariate linear correlation of the condition number of

Φ

with

ε ′

(Figure S12) exhibits an R² of 0.2559 (p-value = 3.1608

∗

10⁻¹⁴⁹). For the condition number of H as well as the inversion errors of H and

Φ

, the correlations are even lower (Figures S13-S15); however, the majority of the values are 10³²³ for the condition of H and 10⁻³²³ for the inversion errors.

In order to further investigate more complex associations among the studied parameters and the extrapolation error

ε ′

, the random forests method [39] was utilized. The numerical dataset was divided into a train set (85% of observations) and a test set (15%) in order to constitute and investigate the reliability of the predictive model. The R² for the predicted versus actual

ε ′

for the test set was 0.8954 (Figure S18), indicating a reliable model. The features significance was evaluated in terms of their contribution to the prediction error in the constituted model (Figure S19A), signifying high values for the studied function, the condition of

Φ

, the number of derivatives, and the domain length. Similarly, considering the computation time as dependent variable, the R² for the test set was 0.9658 and the highest predictive features were the number of divisions, the number of integrations, and number of digits (Figure S19B).

3.3. Illustrative Functions

f

and Their Extrapolation Spans

In Figure 2 the extrapolation results are demonstrated by four test functions as indicated. The initial domain contains only the known given data for

x

and

f (x)

, and after the vertical lines, each graph contains the exact function values, the predicted and the normalized extrapolation error

ε ′ ′ = (log 10 ⁡ ε - min ⁡ (log 10 ⁡ ε)) / (max ⁡ (log 10 ⁡ ε) - min ⁡ (log 10 ⁡ (| ε |)))

, which take values in the

[0,1]

. The text cases include a vast variety of analytic functions as well as their combinations. In Figures 2(a)–2(d) the given curvature or periodicity information is meager, compared to the predicted evolution of the function’s values. Interestingly, the logged error plot exhibits a logarithmic scheme, indicating a weak form of stability [3].

In Figure 3, the proposed procedure is implied firstly in the x-axis and then in the y-axis for the function

f (x, y) = s i n (e 2 x) + c o s (e 3 y / 2)

. Given only the cyan region, the method can predict the highly nonlinear colored surface in the x, y, z space.

4. Discussion

Less

4.1. Numerical Differentiation

Numerical differentiation is highly sensitive to noise [34], especially for the higher order derivatives. After an extensive examination the literature utilizing IRBFs [35, 45, 46] for the derivatives approximation, or for the solution of specific problems [47, 48] as well as other differentiation methods [49, 50] and a variety of formulations for IRBFs (14a, 14b, 14c, 14d, 14e, 14f), the proposed procedure accomplished striking accuracy, with error magnitude ε of O(10⁻¹⁰⁰) or less (Figures S3-S11 and supplementary database), for the derivatives’ computation. This finally permitted the implementation of the Taylor method for significant extrapolation extents (Figures 2 and 3 and Data File S1). The numerical calculation of the derivatives with variable precision arithmetic offers high accuracy [37, 38]. The inverse problem of numerical integration exhibits lower errors with

ε ~ O (10 - 1000)

; however, the integration is less sensitive to—even small—errors in the given data [51]. Cheng [37], similarly to this work, found that the errors of the derivatives approximation are of one or more order of magnitude higher than the function’s (Figures S16, S17, and Data File S1). However, the digits studied were 100-200, while the precise computation of the derivatives was vital for the extrapolation, and so for a higher number of digits. Similarly, Mai-Duy et al. [52] examined the compactly integrated radial basis functions, with errors for the derivatives

ε ~ O (10 - 10)

for fifty digits accuracy.

4.2. RBFs’ Matrix Φ and Shape Parameter

The approximation scheme ((9), (10), (12), (13), 14a, 14b, 14c, 14d, 14e, 14f) permits a precise interpolation, within the given domain, with

ε ~ O (10 - 100)

or less, as found in the numerical experiments (supplementary database). Accordingly, the approximation of the function’s values is attainable with high accuracy, not only at the nodes x_i, but for any intermediate point between two given points and, hence, for points h_j (Figure 1(b)), permitting the precise calculation of vector C (13). Arbitrary precision found capable of avoiding the ill-conditioning of

Φ

was also indicated by Cheng [37] and Huang [38]. Using arbitrary precision in the calculations, the inversion errors in

Φ Φ - 1

-I were found of O(<10⁻¹⁰⁰) or even exact zero (supplementary database). The arbitrary precision has also been used for the robust computation of

Φ

by Cheng [37], where the Gaussian RBFs and inversed multiquadrics exhibit accuracy of interpolation with

ε ~ O

(~10⁻²⁰-10⁻³⁰).

The association of the condition number of

Φ

with

ε ′

exhibits low R² (0.25595) with slightly negative slope (Figure S12), although the high condition number is considered to increase instability [37, 41, 52] indicating the need for high precision to the calculations; however the majority of the values are 10³²³, that is, the maximum real number considered by the software [44]. Sensitivity analysis was selected instead of optimizing each method’s parameters such as times of RBFs integration, the kernel function, or its shape parameter [40–42], as the interpolation errors in a number of the test cases were equal to zero at the nodes, eliminating any relevant objective function. The number of integrations of RBF φ increased computational time, as it magnifies the

Φ

and

Φ - 1

matrices, due to the more complex formulas (14b, 14c, 14e, 14f).

4.3. Summary of Findings and Limitations of the Proposed Solution

In brief, a method for the extrapolation of analytic functions is developed, accompanied by a systematic investigation of the involved parameters. The proposed scheme for the numerical differentiation exhibited low enough errors to permit adequate extrapolation spans. The constituted database of the numerical experiments highlights the fact that, for the same problem formulation (L, N, dx, f), the derivatives calculation exhibits a high variation, indicating the vagueness of the transference from the presumed theory to the actual calculations. The numerical investigation of the extrapolation errors suggests that only utilizing high accuracy and a precise approximation scheme for a function as well as its derivatives, the extrapolation is attainable. Thus, for real-world phenomena, with laboratory accuracies even of O(10⁻²⁰), the predictions are limited to short length. Only if the measurements contain some hundreds of significant digits, the proposed solution is efficient. As this is difficult to be accomplished by laboratory instruments, this work’s findings provide strong evidence that we are far from lengthy predictions in physics, biology, and engineering and also even more far from phenomena studied in health and social sciences. However, the parametric investigation suggests that the precision in calculations and the utilized methods are vastly significant, as the extrapolation horizons achieved by the proposed numerical scheme are about an order of magnitude higher than those in the existing literature, highlighting the potentiality of predictions.

5. Materials and Methods

Less

All the software operations ran on an Intel i7-6700 CPU @3.40GHz with 32GB memory and SSD hard disk, and the execution time was accordingly tracked. The errors of the derivatives in the relevant literature are reported in the Supplementary Data File S1, and they highlight the immense accuracy of numerical differentiation achieved in this work.

References

Less

Oreskes

, Shrader-Frechette

, and Belitz

, “Verification, validation, and confirmation of numerical models in the earth sciences,” Science, vol. 263, no. 5147, pp. 641–646, 1994

Ben Taieb

, Bontempi

, Atiya

A. F.

, and Sorjamaa

, “A review and comparison of strategies for multi-step ahead time series forecasting based on the NN5 forecasting competition,” Expert Systems with Applications, vol. 39, no. 8, pp. 7067–7083, 2012

Demanet

, and Townsend

Stable Extrapolation of Analytic Functions, 2016, https://arxiv.org/abs/1605.09601.

, “Scopus,” , http://www.scopus.com/scopus/homeurl.

Plevris

, Bakas

, Markeset

, and Bellos

, “Literature review of masonry structures under earthquake excitation utilizing machine learning algorithms,” in Proceedings of the 6th International Conference on Computational Methods in Structural Dynamics and Earthquake Engineering, COMPDYN 2017, Greece, June 2017, pp. 2685–2694

Sapankevych

, and Sankar

, “Time series prediction using support vector machines: a survey,” IEEE Computational Intelligence Magazine, vol. 4, no. 2, pp. 24–38, 2009

Ahmad

, Kalra

, and Stephen

, “Estimating soil moisture using remote sensing data: a machine learning approach,” Advances in Water Resources, vol. 33, no. 1, pp. 69–80, 2010

Yang

, Huang

, King

, and Lyu

M. R.

, “Localized support vector regression for time series prediction,” Neurocomputing, vol. 72, no. 10–12, pp. 2659–2669, 2009

Recknagel

, “Applications of machine learning to ecological modelling,” Ecological Modelling, vol. 146, no. 1-3, pp. 303–310, 2001

Crone

S. F.

, Hibon

, and Nikolopoulos

, “Advances in forecasting with neural networks? Empirical evidence from the NN3 competition on time series prediction,” International Journal of Forecasting, vol. 27, no. 3, pp. 635–660, 2011

Makridakis

, Chatfield

, Hibon

, Lawrence

, Mills

, Ord

, and Simmons

L. F.

, “The M2-competition: A real-time judgmentally based forecasting study,” International Journal of Forecasting, vol. 9, no. 1, pp. 5–22, 1993

Murray

J. F.

, Hughes

G. F.

, and Kreutz-Delgado

, “Machine learning methods for predicting failures in hard drives: a multiple-instance application,” Journal of Machine Learning Research, vol. 6, pp. 783–816, 2005

Richard

, Bermudez

J. C.

, and Honeine

, “Online prediction of time serise data with kernels,” IEEE Transactions on Signal Processing, vol. 57, no. 3, pp. 1058–1067, 2008

Hong

, Pinson

, and Fan

, “Global energy forecasting competition 2012,” International Journal of Forecasting, vol. 30, no. 2, pp. 357–363, 2014

Young

, “Technological growth curves. A competition of forecasting models,” Technological Forecasting & Social Change, vol. 44, no. 4, pp. 375–389, 1993

Chen

B.-J.

, Chang

M.-W.

, and Lin

C.-J.

, “Load forecasting using support vector machines: A study on EUNITE Competition 2001,” IEEE Transactions on Power Systems, vol. 19, no. 4, pp. 1821–1830, 2004

Makridakis

, Andersen

, Carbone

, Fildes

, Hibon

, Lewandowski

, Newton

, Parzen

, and Winkler

, “The accuracy of extrapolation (time series) methods: Results of a forecasting competition,” Journal of Forecasting, vol. 1, no. 2, pp. 111–153, 1982

, Shi

, Yong

, Gu

, and Song

, “A B-spline curve extension algorithm,” Science China Information Sciences, vol. 59, no. 3, 2016

Zhang

, and Little

, “Extensions of the penalized spline of propensity prediction method of imputation,” Biometrics: Journal of the International Biometric Society, vol. 65, no. 3, pp. 911–918, 2009

Fan

, Zhang

C. M.

, and Li

J. J.

, “Extension algorithm for B-splines with GC²-continuous,” Chinese Journal of Computers, vol. 28, no. 6, pp. 933–938, 2005

Shetty

, and White

P. R.

, “Curvature-continuous extensions for rational B-spline curves and surfaces,” Computer-Aided Design, vol. 23, no. 7, pp. 484–491, 1991

Mhaskar

H. N.

, “Neural networks for optimal approximation of smooth and analytic functions,” Neural Computation, vol. 8, no. 1, pp. 164–177, 1996

Gonchar

A. A.

, and Rakhmanov

E. A.

, “Equilibrium distributions and the rate of rational approximation of analytic functions,” Mathematics of the USSR – Sbornik, vol. 62, no. 2, pp. 305–348, 1989

Prokhorov

V. A.

, “Rational approximation of analytic functions,” Sbornik: Mathematics, vol. 78, no. 1, pp. 139–164, 1994

Schwartz

, “High-accuracy approximation techniques for analytic functions,” Journal of Mathematical Physics, vol. 26, no. 3, pp. 411–415, 1985

Platte

R. B.

, Trefethen

L. N.

, and Kuijlaars

A. B. J.

, “Impossibility of fast stable approximation of analytic functions from equispaced samples,” SIAM Review, vol. 53, no. 2, pp. 308–318, 2011

Kargar

, Pascu

N. R.

, and Ebadian

, “Locally univalent approximations of analytic functions,” Journal of Mathematical Analysis and Applications, vol. 453, no. 2, pp. 1005–1021, 2017

CoşKun

, “Approximation of analytic functions of several variables by linear k-positive operators,” Turkish Journal of Mathematics, vol. 41, no. 2, pp. 426–435, 2017

Griebel

, and Oettershagen

, “On tensor product approximation of analytic functions,” Journal of Approximation Theory, vol. 207, pp. 348–379, 2016

Jung

S.-M.

, “Approximation of analytic functions by Hermite functions,” Bulletin des Sciences Mathematiques, vol. 133, no. 7, pp. 756–764, 2009

Bustamante

, and Lopes Lagomasino

, “Hermite-pade approximation for nikishin systems of analytic functions,” Sbornik Mathematics, vol. 77, no. 2, pp. 367–384, 1994

Jung

S.-M.

, “Approximation of analytic functions by Airy functions,” Integral Transforms and Special Functions, vol. 19, no. 11-12, pp. 885–891, 2008

Fornberg

, and Zuev

, “The Runge phenomenon and spatially variable shape parameters in RBF interpolation,” Computers & Mathematics with Applications, vol. 54, no. 3, pp. 379–398, 2007

Mai-Duy

, and Tran-Cong

, “Approximation of function and its derivatives using radial basis function networks,” Applied Mathematical Modelling, vol. 27, no. 3, pp. 197–220, 2003

Shu

, and Wu

Y. L.

, “Integrated radial basis functions-based differential quadrature method and its performance,” International Journal for Numerical Methods in Fluids, vol. 53, no. 6, pp. 969–984, 2007

Turchetti

, Vaienti

, and Zanlungo

, “Asymptotic distribution of global errors in the numerical computations of dynamical systems,” Physica A: Statistical Mechanics and its Applications, vol. 389, no. 21, pp. 4994–5006, 2010

Cheng

A. H.

, “Multiquadric and its shape parameter—a numerical investigation of error estimate, condition number, and round-off error by arbitrary precision computation,” Engineering Analysis with Boundary Elements, vol. 36, no. 2, pp. 220–239, 2012

Huang

C.-S.

, Lee

C.-F.

, and Cheng

A. H.-D.

, “Error estimate, optimal shape factor, and high precision computation of multiquadric collocation method,” Engineering Analysis with Boundary Elements, vol. 31, no. 7, pp. 614–623, 2007

Breiman

, “Random forests,” Machine Learning, vol. 45, no. 1, pp. 5–32, 2001

Rippa

, “An algorithm for selecting a good value for the parameter c in radial basis function interpolation,” Advances in Computational Mathematics, vol. 11, no. 2-3, pp. 193–210, 1999

Wang

J. G.

, and Liu

G. R.

, “On the optimal shape parameters of radial basis functions used for 2-D meshless methods,” Computer Methods in Applied Mechanics and Engineering, vol. 191, no. 23-24, pp. 2611–2630, 2002

Fasshauer

G. E.

, and Zhang

J. G.

, “On choosing “optimal” shape parameters for RBF approximation,” Numerical Algorithms, vol. 45, no. 1–4, pp. 345–368, 2007

Hou

, and Han

, “Constructive approximation to multivariate function by decay RBF neural network,” IEEE Transactions on Neural Networks and Learning Systems, vol. 21, no. 9, pp. 1517–1523, 2010

MATLAB, version R2017a, MathWorks Inc., Natick, Mass, USA, 2017

Tien

C. M. T.

, Thai-Quang

, Mai-Duy

, Tran

C.-D.

, and Tran-Cong

, “A three-point coupled compact integrated RBF scheme for second-order differential problems,” CMES: Computer Modeling in Engineering & Sciences, vol. 104, no. 6, pp. 425–469, 2015

Tien

C. M. T.

, Pham-Sy

, Mai-Duy

, Tran

C.-D.

, and Tran-Cong

, “A high-order coupled compact integrated RBF approximation based domain decomposition algorithm for second-order differential problems,” CMES: Computer Modeling in Engineering & Sciences, vol. 104, no. 4, pp. 251–304, 2015

Babouskos

N. G.

, and Katsikadelis

J. T.

, “Optimum design of thin plates via frequency optimization using BEM,” Archive of Applied Mechanics, vol. 85, no. 9-10, pp. 1175–1190, 2015

Yiotis

A. J.

, and Katsikadelis

J. T.

, “Buckling of cylindrical shell panels: a MAEM solution,” Archive of Applied Mechanics, vol. 85, no. 9-10, pp. 1545–1557, 2015

, and Zhang

, “Size-dependent pull-in phenomena in electrically actuated nanobeams incorporating surface energies,” Applied Mathematical Modelling: Simulation and Computation for Engineering and Environmental Systems, vol. 35, no. 2, pp. 941–951, 2011

Hong

, and Chen

Boundary Element Method, Taipei, Taiwan, 1993

Bailey

D. H.

, Jeyabalan

, and Li

X. S.

, “A comparison of three high-precision quadrature schemes,” Experimental Mathematics, vol. 14, no. 3, pp. 317–329, 2005

Mai-Duy

, Le

T. T.

, Tien

C. M.

, Ngo-Cong

, and Tran-Cong

, “Compact approximation stencils based on integrated flat radial basis functions,” Engineering Analysis with Boundary Elements, vol. 74, pp. 79–87, 2017

Makridakis

, and Bakas

, “Forecasting and uncertainty: a survey,” Risk and Decision Analysis, vol. 6, no. 1, pp. 37–64, 2016

Heyman

Google Books: a Complex and Controversial Experiment, New York Times, 2015

Appendix

Less

Year 2019 volume 2019 Issue 5

PDF

Cite this Article

BibTeX

Article Info

doi: 10.34133/2019/3903187

Receive Date：2018-11-08
Online Date：2026-10-27
Published：2019-05-20

Article Data

Affiliations

History

Received：2018-11-08
Accepted：2019-03-18

Affiliations

Intelligent Systems Lab & Civil Engineering Department, School of Architecture, Engineering, Land and Environmental Sciences, Neapolis University Pafos, 2 Danais Avenue, 8042 Paphos Cyprus

References

Share

https://castjournals.cast.org.cn/joweb/research/EN/10.34133/2019/3903187

Share to

Scan QR to access full text

Cite this article

BibTeX

Citations

表12种不同金属材料的力学参数

科 Family	属数 Number of genus	种数 Number of species	占总种数比例 Percentage of total species (%)	属 Genus	种数 Number of species	占总种数比例 Percentage of total species (%)
鹅膏菌科Amanitaceae	2	11	5.26	鹅膏菌属 Amanita	10	4.78
小菇科 Mycenaceae	2	12	5.74	丝盖伞属 Inocybe	5	2.39
多孔菌科 Polyporaceae	8	14	6.70	蜡蘑属 Laccaria	5	2.39
红菇科 Russulaceae	3	23	11.00	小皮伞属 Marasmius	6	2.87
				小菇属 Mycena	11	5.26
				光柄菇属 Pluteus	5	2.39
				红菇属 Russula	17	8.13
				栓菌属 Trametes	5	2.39

关闭全屏

BibTeX
EndNote
RefWorks
TxT

Figure 1 Given values of $f (x)$ (a) and interpolation within the edge interval (b).

Figure 2 Extrapolation of analytic functions $f (x)$ . The prediction starts after the vertical line.

Figure 3 Extrapolation of function $f (x, y) = s i n (e 2 x) + c o s (e 3 y / 2)$ (a) and normalized errors (b).

Articles: Latest Articles; Most Read; Collections

Updates: Events; News; Multimedia

About: About Us

Contact

No. 86 Xueyuan South Road, Haidian District, Beijing

100081

010-62662699(Operation and Technical Support)

qkjq@cast.org.cn

Copyright © 2025 China Association for Science and Technology. All rights reserved. For all open access content, the relevant licensing terms apply.
Sponsored by the Office of the Leading Group for Cybersecurity and Informatization of CAST, and supported by Science and Technology Review Publishing House