WorksPhysics and Astronomy
Theory and Application of the Fractional-order Delta Function Associated with the Inverse Laplace Transform of the Mittag-Leffler Function
Abstract
This paper is devoted to the study of the -Wright function () which is the inverse Laplace transform of the single-parameter Mittag-Leffler (ML) function (). Because can be viewed as the fractional-order generalization of the exponential function for , to which it reduces for , i.e. , its inverse Laplace transform, being , can be viewed as a generalized fractional-order Dirac delta function. At the limiting case of the -Wright function reduces to . We investigate numerically the behavior of this fractional-order delta function as well as it integral, the fractional-order unit-step function. Subsequently, we validate our results with experimental data for the charging of a supercapacitive device.
Introduction
The Dirac delta function or impulse function is a generalized function that has a wide range of applications in various fields of sciences and engineering1. Conceptually, it represents a function singular at a point on the real axis and zero elsewhere, and the integral of any continuous function when multiplied by the Dirac delta function gives the value of the function at that specific point. Its generalization may provide a convenient way to work around its singular nature, which is the main purpose of this work.
Our analysis in this work originates from the fact that the delayed Dirac delta function is the inverse Laplace transform of the exponential decaying function . From there, a fractional generalization of the delta function associated with the inverse Laplace transform of the Mittag-Leffler (ML) function arises in terms of the -Wright function 2–4, given that the ML function can be viewed as a generalization of the exponential function5. A schematic illustration of the Laplace transform pair and , and its (fractional) generalization and is depicted in Figure 1, and summarizes the principal result of this work.
Figure 1. A schematic representation of the Laplace transform (LT) pair given in 5 and their (fractional) generalization given in 16
The rest of the paper is organized as follows. In section II we present an overview of the standard delta function and its main properties, the ML function and its inverse Laplace transform, as well the representation of these functions in terms of the Fox -function. We also provide parametric graphical plots showing how the -Wright function tends to the standard delta Dirac function. In section III we study theoretically and verify experimentally (on a supercapacitor) a simple example of a fractional-order differential equation involving the -Wright function used to charge this capacitive device in the form of an applied current. This resembles finding the impulse response of the device using a fractional-order delta function.
Theory
First, we start by setting the basic definitions and notations to be used throughout the paper. The Laplace transform of a sufficiently well-behaved function (
is defined by:
and the inverse Laplace transform of is defined by:
The conditions of validity and the main rules are given in any textbook on advanced mathematics[6]. By using the sign to denote the juxtaposition of the function with its Laplace transform , a Laplace transform pair is represented by:
The Dirac delta function and the Laplace transform pair
If we consider the function in 3 to be , where
then we have the Laplace transform pair (Fig. 1):
The (one-dimensional) Dirac delta function is known to obey several fundamental properties. For being a well-behaved function we have[7]:
and if for then we have the normalization condition:
Also, if we chose the function we obtain the Fourier transform of as:
and by inversion we obtain the following representation of the delta function:
The delta function can also be expressed as the derivative of the Heaviside unit step function (denoted as , or ), i.e.:
where[7]
Finally, convolving with a continuous function gives:
and by virtue of the Laplace convolution theorem:
The delta function is the ”identity” with respect to the convolution integral.
The fractional delta function and the Laplace transform pair
Now we consider the single-parameter (or standard) Mittag-Leffler (ML) function denoted by with , , which is defined by (i) the power series representation2:
where the Gamma function is given by , or (ii) by the integral representation2
where indicates the Hankel path of integration. The ML function is a direct (fractional) generalization of the exponential series (Eq. (4))4; for it interpolates between and . A detailed account of its basic properties can be found for instance in the third volume of Batemann Manuscript Project editted by Erdélyi et al.[8]. The ML function has been generalized to other forms with two[9], three[10] and four[11] parameters but we focus here on its single-parameter version. The ML and related functions usually appear naturally in the solution of fractional order integro-differential equations describing anomalous relaxation, kinetics, diffusion and reaction[12] [5]. From Mainardi2 [Appendix F, p. 237], the inverse Laplace transform of for is given in terms of the Wright function2 as:
The (classical) Wright function with , is defined by the series representation[13] [2]:
or by the integral:
We note that for an auxiliary function of the Wright type is defined as[14]:
where is usually referred to as the -Wright function or the Mainardi function[15] [16]. Its series and integral representations are4:
and
respectively. The relationship of the -Wright and ML functions through the Laplace transform can be easily followed for instance from their integral representations, i.e.4:
Like the ML function, the (classical) Wright function has also been extended to the generalized Wright function defining a series generalizing the hypergeometric series.[17, 18, 19] They are known to play fundamental roles in various applications of fractional calculus[20]. In particular, like the ML function is viewed as the natural (fractional) generalization of the exponential function, the Mainardi function can be viewed as a natural (fractional) generalization of the Gaussian function and of the Airy function given that3,4:
It has also been suggested as the generalized hyper-Airy function2,4.
Again, Figure 1 shows a schematic representation of given in5, and their generalization given in16.
Representation in terms of the Fox H-function
The exponential function, the ML function and the Wright (or -Wright) function can all be represented in terms of the Fox -function as[21]:
respectively. The Fox’s -function[23] is defined in terms of Mellin-Barnes type integral as[21]:
with given by the ratio of products of Gamma functions:
are integers satisfying , , and , , or with , . The contour of integration is a suitable contour separating the poles , , of the gamma functions from the poles , of the gamma functions , that is . An empty product in 31, if it occurs, is taken to be one. The -function contains a vast number of elementary and special functions as special cases. Detailed and comprehensive accounts of the matter are available in Mathai, Saxena, and Haubold[21], Mathai and Saxena[24], and Kilbas and Saigo[25]
Numerical simulations
Plots of using its Fox’s -function representation (Eq. 29, case with and ), i.e.:
for and for different values of ranging from 0 to 0.99 are shown in Figure 2. We used the computational software Mathematica ver. 13 for plotting. One can clearly see the transition from the exponential decay for , to for , to the peaking (non-ideal) Dirac delta function (shifted to the right by one) as . Specifically, for the function is monotonically decreasing as is increased, while for the function exhibits a maximum whose position tends to as . This can be clearly seen in Figure 2(c), where we used sets of values of closer to one. Here we had to limit the range of simulation to short intervals around due to computational efficiency reasons for the estimation of the -function. The convergence of the results becomes more difficult for larger values of . Otherwise, similar results were reported by Mainardi and Tomirotti¹⁴ albeit obtained by means of numerical matching between the series and the saddle-point representations of the function.
Figure 2. Plots of (Eq. 32) for and for different values of from 0 to 0.99
In the same way as the Heaviside step function is the integral of the Dirac delta function (Eq. 10), the integral of the function (Eq. 32) is given by:
which can be viewed as a generalized Heaviside function that we denote by . Figure 3 presents the graphical features of for the same values of used in Figure 2, i.e. from 0 to 0.99. The plots show clearly the evolution of the function towards the quasi-piecewise unit step as the value of approaches one (Figure 3(c)).
Figure 3. Plots of (Eq. 33) for and for different values of from 0 to 0.99
Application
Fractional-order impulse response
In this section we give a simple example for the application of the generalization to . We consider the case of a first-order differential equation with initial value as follows:
which can represent for example the dynamics of an ideal capacitor in response to an impulse current excitation. The solution for (assuming ) is given in terms of the step function as:
We rewrite (34) wherein (i) the first-order derivative is replaced with where is the Caputo differential operator defined as:
where , , and (ii) the right-hand side is replaced with the -Wright function . The fractional derivative in the Caputo sense is frequently encountered in certain boundary value problems arising in the theory of viscoelasticity, hereditary solid mechanics, and supercapacitor modeling. We now have:
This equation can describe the dynamics of a non-ideal fractional-order capacitor (electric double-layer capacitors or supercapacitors) known to exhibit distributed time constants or frequency dispersion of capacitance[26, 27, 28]. It can also be used to model the transient photovoltage/photocurrent in solar cells in response to fast and small perturbation of incident light[29]. Applying the Laplace transform to both sides of Eq. (37) (with ), knowing that:
gives:
Then is found by inverse Laplace transform of Eq. (39) using formula 2.21 in Mathai, Saxena, and Haubold[21]:
which gives:
For and , reduces to the Heaviside step function .
Plots of Eq. (41) for and different values of between 0.90 and 0.99 are shown in Figure 4. Again, we had to limit the range for for computational reasons. Otherwise, it is clear from Figure 4(a) with , and as it should be, that there is a gradual tendency of towards the unit step function as gets closer to one. Otherwise, as the value of is decreased (Figure 4(b)–Figure 4(c)) we see more prominent peaking features centered at followed by a plateauing tendency as is increased.
Figure 4. Plots of Eq. (41) for and for different values of from 0.90 to 0.99 for (a) , (b) , and (c)
Experimental results
A GHC NanoForce supercapacitor, rated 1 F, 2.7 V, was characterized in response to current excitations of the form of with (Eq. (37)) on a Biologic VSP-300 electrochemical station. The spectral impedance of the device was first measured at open-circuit condition from 10 mHz till 1 MHz (not shown here), and fitted to a fractional-order capacitor of impedance function:
The fitting parameters were found to be F s with indicating a slight deviation of the device from the behavior of ideal capacitors.
In any case, Eq. (37) can be used to describe the time-domain voltage-current relationship in this device, i.e.:
The initial condition was experimentally imposed by first discharging the device long enough into a resistor until its voltage reached the value of (the lower limit of the measuring instrument).
The plots of current inputs for the three different values of (), and the resulting voltage developed across the device are shown in Figure 5. The results demonstrate very close agreement with the theory (compare with Figure 4) with a peaking response that become narrower as is increased towards one, followed then by a plateau at in all cases. We should mention here that the device has a series resistance of about , which may affect the extent of the voltage reach when the (fractional) impulse excitations are applied, as well as the value of the step response at which the device finally settles.
Figure 5. Plots of input fractional-order delta current, with (Eq. (43)), and measured voltage on a GHC NanoForce supercapacitor
Conclusion
In this paper we studied the fractional generalization of the Laplace transform pair giving with focus on the behavior and application of (i) the fractional-order delta function and (ii) its integral giving the fractional-order unit step function. Both functions are crucial for the evaluation of the impulse response and step response of circuits and systems. We have experimentally demonstrated the current charging of a supercapacitor using the fractional delta function, and hence its fractional impulse response obtained in the form of the measured voltage. The results fully agree with the numerical modeling of the device. The results may find many other applications in problems associated with differential and integral equations of fractional order arising in different disciplines of physical sciences and engineering.
References
- [6]6 F. E. Harris, Mathematics for physical science and engineering: symbolic computing applications in Maple and Mathematica (Academic Press, 2014).
- [7]Hoskins, R. F., & Hoskins, R. F. (2009). Delta functions: an introduction to generalised functions.
- [8]8 A. Erdélyi, W. Magnus, F. Oberhettinger, and F. Tricomi, eds., Higher Transcendental Functions, Bateman Manuscript Project, Vol. 3 (McGraw-Hill New York, 1955).DOI
- [9]Wiman, A. (1905). Über den fundamentalsatz in der theorie der funktionen $e_{\alpha}(x)$. Acta Mathematica, 29.
- [10]10 T. R. Prabhakar, “A singular integral equation with a generalized Mittag Leffler function in the kernel,” Yokohama Math. J. 19, 7–15 (1971).
- [11]11 A. Shukla and J. Prajapati, “On a generalization of mittag-leffler function and its properties,” Journal of mathematical analysis and applications 336, 797–811 (2007).DOI
- [12]12 A. M. Mathai and H. J. Haubold, Special Functions for Applied Scientists (Springer New York, New York, NY, 2008).DOI
- [13]Mainardi, F. (1997). Fractional calculus: some basic problems in continuum and statistical mechanics.
- [14]14 F. Mainardi and M. Tomirotti, “Seismic pulse propagation with constant q and stable probability distributions,” Annali Di Geofisica 40, 1311–1328 (1997).arxiv.org/abs/1008.1341
- [15]15 R. Balescu, “V-langevin equations, continuous time random walks and fractional diffusion,” Chaos, Solitons & Fractals 34, 62–80 (2007).DOI
- [17]17 A. Erdélyi, W. Magnus, F. Oberhettinger, and F. Tricomi, eds., Higher Transcendental Functions, Bateman Manuscript Project, Vol. 1 (McGraw-Hill New York, 1953).DOI
- [18]Kilbas, A. A., Saxena, R., Saigo, M., & Trujillo, J. J. (2003). Generalized wright function as the h-function. Analytic Methods of Analysis and Differential Equations, 117–134.
- [19]19 M. El-Shahed and A. Salem, “An extension of wright function and its properties,” Journal of Mathematics 2015 (2015).DOI
- [20]20 Y. Luchko, “The wright function and its applications,” Handbook of Fractional Calculus with Applications 1, 241–268 (2019).DOI
- [21]21 A. M. Mathai, R. K. Saxena, and H. J. Haubold, The H-function: theory and applications (Springer Science & Business Media, 2009).
- [22]22 F. Mainardi and G. Pagnini, “The role of the fox–wright functions in fractional sub-diffusion of distributed order,” Journal of computational and applied mathematics 207, 245–257 (2007).arxiv.org/abs/0711.3779
- [23]23 C. Fox, “The $g$ and $h$ functions as symmetrical fourier kernels,” Transactions of the American Mathematical Society 98, 395–429 (1961).DOI
- [24]Mathai, A. M., Saxena, R. K., Saxena, R. K., & al., . et . (1978). The H-function with applications in statistics and other disciplines.
- [25]Saigo, M., Kilbas, A. A., & al., . et . (2004). H-transforms: Theory and applications.
- [26]26 A. Allagui and M. E. Fouda, “Inverse problem of reconstructing the capacitance of electric double-layer capacitors,” Electrochim. Acta, 138848 (2021).DOI
- [27]27 A. Allagui, A. S. Elwakil, and M. E. Fouda, “Revisiting the time-domain and frequency-domain definitions of capacitance,” IEEE Trans. Electron Devices 68 (2021).DOI
- [28]28 A. Allagui, D. Zhang, and A. S. Elwakil, “Short-term memory in electric double-layer capacitors,” Appl. Phys. Lett. 113, 253901–5 (2018).DOI
- [29]Zhang, D., Allagui, A., Elwakil, A. S., Nassef, A. M., Rezk, H., Chengi, J., & Choy, W. C. (2019). On the modeling of dispersive transient photocurrent response of organic solar cells. Org. Electron., 70, 42–47.




