Abstract
The autocorrelation function of a stochastic process is one of the essential mathematical tools in the description of variability that is successfully applied in many scientific fields such as signal processing or financial time series analysis and forecasting. The aim of the paper is to provide an analysis of fuzzy transform of higher degree applied to stochastic processes with a focus on its autocorrelation function. We introduce a higher-degree fuzzy transform of a stochastic process and investigate its basic properties. Further, we introduce a higher-degree fuzzy transform of bivariate functions in the tensor product of polynomial spaces and demonstrate its approximation ability. In addition, we show that the bivariate higher-degree fuzzy transform of multiplicative separable functions is a product of univariate fuzzy transform of the respective functions, which is a desirable property in the processing of functions of higher dimensions. The obtained results are used to demonstrate an interesting identity between the autocorrelation function of the fuzzy transform of a stochastic process and the bivariate fuzzy transform of the autocorrelation function of the stochastic process. This identity provides two ways to determine the autocorrelation function of the fuzzy transform of a stochastic process, which can be useful, especially in a situation where its direct calculation becomes complex or even impossible, for example, the calculation of a sample autocorrelation function.
Similar content being viewed by others
Notes
More precisely, in the theory of stochastic processes we distinguish two main approaches to the definition of stationarity of a stochastic process. One approach referring as the strict-sense stationarity requires that the joined distribution function of a stochastic process does not depend on time shift. Nevertheless, for many applications, the strict-sense stationarity is too restrictive, and only a weaker form, referred as the weak-sense stationarity, for stochastic processes is considered requiring that the mean function is constant and the autocorrelation function is independent of time shift (see Sect. 2.3). For more information about stationary stochastic processes, we refer to Jazwinski (2007), Papoulis and Pillai (2002), Yaglom (1962).
For now, let us forget that the mean function is constant and the process is weak-sense stationary in its pure mathematical definition.
Note that the F-transform was originally introduced as a fuzzy transform of zero degree, where the F-transform components are real numbers, therefore, the fuzzy partition, usually determined by a generating function, bandwidth and later also by a shift and central node (see Definition 2), was only one parameter of F-transform.
We consider a “reasonably smooth” function to be a function that has derivatives up to a certain degree at each point of its domain.
For a reader, who is not familiar with the basic concepts of the theory of stochastic processes, we refer to Sect. 2.3 on p. 6.
See Sect. 3.2 on p. 10 for the definition.
Note that a bivariate polynomial P of degree n with complex coefficients has the form \(P(t,s)=\sum _{0\le i+j\le n} a_{ij}t^is^j\), while a bivariate polynomial Q from the tensor product of polynomial spaces of degree n over the field of complex numbers has the form \(Q(t,s)=\sum _{0\le i,j\le n} a_{ij}t^is^j\), where \(a_{ij}\) denotes complex coefficients.
See the last paragraph of Sect. 3.2.
A bivariate function f(x, y), \(x,y\in {\mathbb {R}}\), is said to be multiplicative separable if there are two univariate functions \(f_1(x)\) and \(f_2(y)\) such that \(f(x,y)=f_1(x)\cdot f_2(y)\) for any \(x,y\in {\mathbb {R}}\). An example of a multiplicative separable function is \(f(x,y)=2\sin (x)\cos (y)\).
Note that \(\psi \) is a non-negative function that is positive on a set of positive Lebesgue measure, in our case, it is easy to see that \(E=\{\mathbf {t}\mid \psi (\mathbf {t})>0\}\) is equal to \(\prod _{i=1}^n (a_i,b_i)\) for suitable \(a_i,b_i\in {\mathbb {R}}\), \(i=1,\ldots ,n\), which is a Lebesgue measurable set. In literature (see, e.g., Cruz-Uribe et al. 2011), such functions are often referred to as a weight or weighted function.
Recall that the support of a real-valued function \(\psi \) is defined as the subset of the domain containing those elements which are not mapped to zero or the topological closure of this set, if a topological structure is assumed. In this paper, we consider the closed supports, i.e., \(\mathrm {Supp}(\psi )=\mathop {\mathrm {cl}}{\{\mathbf {t}\mid \mathbf {t}\in {\mathbb {R}}^n,\, \psi (\mathbf {t})\ne 0\}}\), where \(\mathrm {cl}\) denotes the standard topological closure on \({\mathbb {R}}^n\). Then, the support of \(\psi \) is the n-orthotope \(\prod _{i=1}^n[t_{k_i}-h_i,t_{k_i}+h_i]\), where \(t_{k_i}=k_i r_i\) and \(h_i\) is the ith component of the bandwidth \(\mathbf {h}\).
For more information on weighted function spaces, see Cruz-Uribe et al. (2011).
Note that the Cauchy–Schwarz inequality expressed using the inner product has the form \(|\langle \psi ,\zeta \rangle |^2\le \langle \psi ,\psi \rangle \langle \zeta ,\zeta \rangle \).
Obviously, we have \(|\alpha +\beta |^2=(\alpha +\beta )({\bar{\alpha }}+{\bar{\beta }})=\alpha {\bar{\alpha }}+\alpha {\bar{\beta }}+{\bar{\alpha }}\beta +\beta {\bar{\beta }}\) and \(|\alpha -\beta |^2=(\alpha -\beta )({\bar{\alpha }}-{\bar{\beta }})=\alpha {\bar{\alpha }}-\alpha {\bar{\beta }}-{\bar{\alpha }}\beta +\beta {\bar{\beta }}\ge 0\). From the second inequality, we find that \(\alpha {\bar{\alpha }}+\beta {\bar{\beta }}\ge \alpha {\bar{\beta }}+{\bar{\alpha }}\beta \), from which we obtain \(|\alpha +\beta |^2\le 2\alpha {\bar{\alpha }}+2\beta {\bar{\beta }}=2|\alpha |^2+2|\beta |^2.\)
Note that \( \int _c^d\int _c^d f(t)f(s)\, \mathrm{d}t \mathrm{d}s=\int _c^d f(t)\,\mathrm{d}t\cdot \int _c^d f(s)\,\mathrm{d}s\), which implies the Riemann integrability of f(t)f(s) on \([c,d]\times [c,d]\). Moreover, it is well known that the product of two Riemann integrable functions defined on a rectangle \([a,b]\times [c,d]\) is again a Riemann integrable function (see, e.g., Theorem 6.13 in Rudin 1976). In our case, we consider the product of functions \(\varGamma _\xi (t,s)\) and \(F(t,s)=f(t)f(s)\).
The first attempt to introduce the \(\hbox {F}^m\)-transform of a stochastic process can be found in Holčapek et al. (2013b), nevertheless, no proper definition has been given here.
A complex-valued discrete time stochastic process \(\xi \) is a discrete function that assigns a complex-valued random variable \(\xi (t)\) for any time \(t\in {\mathbb {Z}}\). The concept of weak-sense stationarity for discrete time stochastic processes is defined in the same way as for continuous time stochastic processes, where \({\mathbb {R}}\) is replaced by \({\mathbb {Z}}\) in the definition.
Note that \(\varGamma _\xi (t,s)={\mathbb {E}}[\xi (t)\xi (s)]\), \(\varGamma _\xi ^*(\tau )=\varGamma _\xi (0,\tau )\) and \(\varGamma ^*_\xi (\tau )=\varGamma _\xi ^*(-\tau )\) for any \(t,s,\tau \in {\mathbb {R}}\).
For \(n,m\in {\mathbb {N}}\), we use \(\mathrm {mod}(n,m)\) to denote the remainder after dividing n by m. For example, \(\mathrm {mod}(11,4)=3\), because \(10=4\cdot 2+3\) and \(0\le 3<4\) (the remainder 3 has to be less than divisor 4).
This follows from \(\int _a^b\int _c^d |f(t,s)|^2 \,\mathrm{d}t\mathrm{d}s=\int _a^b |f_1(t)|^2\,\mathrm{d}t\cdot \int _c^d |f_2(s)|^2\,\mathrm{d}s\).
References
Brémaud P (2014) Fourier analysis and stochastic processes. Springer, Cham
Brockwell PJ, Davis RA (2006) Time series: theory and methods. Springer, Berlin
Champeney DC, Champeney D (1987) A handbook of Fourier theorems. Cambridge University Press, Cambridge
Chen WK (2004) The electrical engineering handbook. Elsevier, Amsterdam
Crouzet JF (2012) Fuzzy projection versus inverse fuzzy transform as sampling/interpolation schemes. Fuzzy Sets Syst 193:108–121
Cruz-Uribe DV, Martell JM, Pérez C (2011) Weights, extrapolation and the theory of Rubio de Francia. Springer, Berlin
Debnath L, Mikusinski P (2005) Introduction to Hilbert spaces with applications. Elsevier, Amsterdam
Di Martino F, Sessa S (2020) Fuzzy transform for image fusion and edge detection. In: Fuzzy transforms for image processing and data analysis: core concepts, processes and applications. Springer, Cham, pp 61–79
Engleberg S (2007) Random signals and noise: a mathematical introduction. CRC Press, Boca Raton
Grami A (2015) Introduction to digital communications. Academic Press, Cambridge
Guérin CA (2000) Wavelet analysis and covariance structure of some classes of non-stationary processes. J Fourier Anal Appl 6:403–425
Hamilton JD (1994) Time series analysis. Princeton University Press, Princeton
Hodáková P (2015) Fuzzy (F-)transform of functions of two variables and its applications in image processing. Ph.D. thesis. University of Ostrava
Holčapek M, Tichý T (2015) Discrete multivariate F-transform of higher degree. In: Proceedings of IEEE international conference on fuzzy systems (FUZZ-IEEE). IEEE, pp 1–8
Holčapek M, Nguyen L (2018) Trend-cycle estimation using fuzzy transform of higher degree. Iran J Fuzzy Syst 15(7):23–54
Holčapek M, Novák V, Perfilieva I (2013) Analysis of stationary processes using fuzzy transform. In: Pasi G, Montero J, Ciucci D (eds) Proceedings of the 8th conference of the European society for fuzzy logic and technology (EUSFLAT 2013). Atlantis Press, pp 714–721
Holčapek M, Novák V, Perfilieva I (2013) Noise reduction in time series using F-transform. In: Proceedings of IEEE international conference on fuzzy systems, pp 1–8
Holčapek M, Perfilieva I, Novák V, Kreinovich V (2015) Necessary and sufficient conditions for generalized uniform fuzzy partitions. Fuzzy Sets Syst 277:97–121
Holčapek M, Nguyen L, Tichý T (2018) Polynomial alias higher degree fuzzy transform of complex-valued functions. Fuzzy Sets Syst 342:1–31
Hyndman RJ, Athanasopoulos G (2018) Forecasting: principles and practice, 2nd edn. Springer, Berlin
Jazwinski AH (2007) Stochastic processes and filtering theory. Dover Publications, Mineola, p 376
Kawata T (1965) Fourier analysis of nonstationary stochastic processes. Trans Am Math Soc 118(13):276–302
Lindgren G (2013) Stationary stochastic processes: theory and applications. CRC Press, Boca Raton
Masry E (1993) The wavelet transform of stochastic processes with stationary increments and its application to fractional Brownian motion. IEEE Trans Inf Theory 39(1):260–264
Mills TC, Markellos RN (2008) The econometric modelling of financial time series. Cambridge University Press, Cambridge
Nguyen L, Holčapek M (2018) Higher degree fuzzy transform: Application to stationary processes and noise reduction. In: Kacprzyk J, Szmidt E, Zadrožny S, Atanassov K, Krawczak M (eds) Advances in fuzzy logic and technology 2017. Proceedings of EUSFLAT 2017, IWIFSGN 2017. Springer, Cham, pp 1–12
Nguyen L, Novák V (2015) Filtering out high frequencies in time series using F-transform with respect to raised cosine generalized uniform fuzzy. In: Proceedings of the IEEE international conference on fuzzy systems 2015, Istanbul, Turkey
Nguyen L, Novák V (2019) Forecasting seasonal time series based on fuzzy techniques. Fuzzy Sets Syst 361:114–129
Nguyen L, Holčapek M, Novák V (2017) Multivariate fuzzy transform of complex-valued functions determined by monomial basis. Soft Comput 21(13):3641–3658
Nguyen L, Novák V, Holčapek M (2020) Gold price: trend-cycle analysis using fuzzy technique. In: . Lesot M-J et al (eds) In proceedings of IPMU 2020, CCIS 1239. Springer, pp 254–266
Nguyen L, Novák V, Mirshahi S (2020) Trend-cycle estimation using fuzzy transform and its application for identifying of bull and bear phases in markets. J Intell Syst Account Finance Manag 27:111–124
Novák V (2018) Detection of structural breaks in time series using fuzzy techniques. Int J Fuzzy Logic Intell Syst 8(1):1–12
Novák V, Perfilieva I, Holčapek M, Kreinovich V (2014) Filtering out high frequencies using F-transform. Inf Sci 274:192–209
Papoulis A, Pillai SU (2002) Probability, random variables and stochastic processes, 4th edn. McGraw-Hill, New York
Perfilieva I (2004) Fuzzy transforms. In: Peters JF et al (eds) Transactions on rough sets II. Lecture notes in computer science, vol 3135. Springer, Berlin, pp 63–81
Perfilieva I (2006) Fuzzy transforms: theory and applications. Fuzzy Sets Syst 157(8):993–1023
Perfilieva I, Valášek R (2005) Fuzzy transforms in removing noise. In: Reusch B (ed) Computational intelligence, theory and applications. Advances in soft computing, vol 33. Springer, Berlin, Heidelberg, pp 221–230
Perfilieva I, Daňková M, Bede B (2011) Towards a higher degree F-transform. Fuzzy Sets Syst 180(1):3–19
Perfilieva I, Hodáková P, Hurtík P (2016) Differentiation by the F-transform and application to edge detection. Fuzzy Sets Syst 288:96–114
Perfilieva I, Holčapek M, Kreinovich V (2016) A new reconstruction from F-transform components. Fuzzy Sets Syst 288:3–25
Rudin W (1976) Principles of mathematical analysis. McGraw-Hill Education, London
Steeb WH, Hardy Y (2011) Matrix calculus and Kronecker product: a practical approach to linear and multilinear algebra, 2nd edn. World Scientific, Singapore
Tsay RS (2005) Analysis of financial time series. Wiley, Hoboken
Yaglom AM (1962) An introduction to the theory of stationary random functions. Revised English ed. Translated and edited by Richard A. Silverman, vol XIII. Prentice-Hall, Inc., Englewood Cliffs, NJ, p 235
Young N (1988) An introduction to Hilbert space. Cambridge University Press, Cambridge
Acknowledgements
We would like to thank the editors and anonymous referees for their valuable comments that significantly helped us to improve the paper. This work was supported by the ERDF/ESF project AI-Met4AI (No. CZ.02.1.01/0.0/0.0/17_049/0008414). The additional support was also provided by the Czech Science Foundation through the Project No. 18-13951S.
Author information
Authors and Affiliations
Corresponding author
Ethics declarations
Conflict of interest
The authors declare that they have no conflict of interest.
Human and animal rights
This article does not contain any studies with human participants or animals performed by the authors.
Additional information
Communicated by Vladik Kreinovich.
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Rights and permissions
About this article
Cite this article
Holčapek, M., Nguyen, L. Analysis of autocorrelation function of stochastic processes by F-transform of higher degree. Soft Comput 25, 7707–7730 (2021). https://doi.org/10.1007/s00500-020-05543-x
Published:
Issue Date:
DOI: https://doi.org/10.1007/s00500-020-05543-x