Some comments on using fractional derivative operators in modeling non-local diffusion processes
Introduction
A non-local transport phenomenon can be defined as one where the transport flux at given point depends not only on values at that point but also on values at remote locations. For example, a non-local diffusion flux can be defined as a weighted sum of potential gradients evaluated at points throughout the domain. Due to their non-local nature, many works (e.g., see [1], [2], [3] have been directed at using fractional derivative operators to construct non-local transport models in physical settings. Recent work [4] studied the numerical solution of the diffusion equation where is the open interval , the operator is a fractional derivative of order , is an appropriately scaled diffusivity, and is a source/sink term. By numerically predicting the fate of an initial non-negative pulse on , the authors of [4] investigate three alternatives for the fractional operator , a Riemann–Liouville (RL) derivative, a Caputo (C) derivative, and a so called Patie–Simon (PS) derivative. A major and important contribution from the work of [4], is the development of a comprehensive and consistent treatment of boundary conditions for numerical solution of the fractional heat conduction equation. The work shows, under a variety of boundary and initial conditions, that numerical solutions using the RL and PS derivatives, while not-identical, are always conservative. This, however, is not the case for the C derivative, under some boundary and initial conditions this operator may not be conservative. The authors of [4] demonstrate this by considering a problem with fixed boundary values, , presenting a numerical solution, based on the C derivative, where an initial non-negative pulse produces later time solutions with negative values.
An alternative form for the fractional diffusion equation is where we interpret the right-hand side as the conserved statement the divergence of a flux . We note that in using the equation form suggested by [4], i.e., Eq. (1), conservation will only by ensured if . This is true in the case the operator is the RL derivative of order . For Caputo derivatives, however, the condition that , only holds if (see Remark 6.5 in [4] and Subsection 3.1. below). This explains, why, in some cases use of the order ( C fractional derivative in Eq. (1) is not conservative. Indeed, we note that a core advantage of the PS derivative, adopted by [4], is that, by its construction, and thus, when used in Eq. (1) conservation is enforced.
Our task here is to study the space fractional diffusion equation in , under a variety of flux definitions. In a departure from [4], we will base our analysis and solutions on the conserved form of the diffusion transport equation, Eq. (2). Our investigation generates a number of insights and results related to the solution of fractional diffusion equations in bounded domains and also highlights possible advantages in using the Caputo derivative in seeking discrete (numerical) solutions.
We start with a general conserved statement for a transient diffusion transport written in terms of a general flux . On specific definition of the flux we arrive at three alternative diffusion transport models, the classic Fourier second law (), a Caputo model (), and a Riemann–Liouville model ().
To lay a foundation for development of numerical solutions for the general conserved model we present some analytical results related to the well posedness of fractional flux models.
We gather here in Section 3 the currently available results, see [5], [6], [7], [8], on the well-posedness of (1) or (2) augmented with an array of boundary conditions and initial data. It turns out that we can consider the data from the following list: Dirichlet conditions, fixed flux, fixed slope conditions and mixed ones.
Another important issue is the validity of the maximum principle. It holds for some problems written in a conservative form. We present not only a rigorous argument, but also discuss the consequence of this principle on the invariance of Eq. (2) under a re-scaling; i.e., regardless of the measurement scale used the actual value of the potential (e.g., temperature) should always be the same. We show, by example, that Eq. (2) with the flux given in terms of the Riemann–Liouville derivative fails to satisfy this invariance. On the other hand this scale invariance and the maximum principle hold if we use the Caputo derivative in Eq. (2). We offer the explanation in Section 7 and further comments in Section 7.5.
A thorough discussion of the boundary value problem (2) is not complete without addressing the issue of the boundary conditions. This is done in Section 3, where we ask the question whether or not we can evaluate , the flux or the derivative of at the boundary of for a solution to (2). In particular, we note that the condition is automatically satisfied for sufficiently regular , see Proposition 3.1. We explain it in detail in Section 3.1. This section is a good place to discuss the correctness of the definition of and .
Following this theory, we develop an explicit control volume discretization for the general conserved model, identifying a range of possible boundary condition treatments, the operation of which are essentially invariant to the precise definition of the flux used. A numerical analysis investigating the consistency and stability of these schemes is conducted. Then, taking a similar approach to the one adopted in [4] we study the fate of an initial non-negative pulse in with both the Caputo and Riemann–Liouville models. As expected these tests give identical results to those provided in [4] and illustrate the predictive differences between using Caputo and Riemann–Liouville flux models. However, due to the fact that our analysis is based on a conserved treatment we are able to write down explicit discrete expressions for the Riemann–Liouville and Caputo fluxes. These expressions show that difference between the Caputo and Riemann–Liouville flux models reduce to the addition of a correction term to the Riemann–Liouville flux with the form of an apparent advection transport. This observation provides us with a direct and clear explanation of the predictive differences between Caputo and Riemann–Liouville fluxes. In addition it also allows us to identify situations where the Riemann–Liouville flux definition may lead to predictions which are physically and mathematically unsound. To conclude we suggest an alternative fractional flux law that essentially negates the differences between the Caputo and Riemann–Liouville flux models.
Section snippets
The conserved diffusion transport equation
The transient diffusion equation, on the interval , written in the conserved form is where is the diffusion flux.
We consider three separate definitions for this flux:
The Fourier Law Assuming a unit diffusivity , the classic flux definition is the Fourier law [9] stating that the flux is proportional to the potential gradient.
Riemann–Liouville flux The Fourier law is local, i.e., calculating the flux at a specified point only requires knowledge
On the definitions of fractional derivatives
The first issue to be clarified before tackling the differential equations is the well-posedness of the operators and for . It is convenient to use for this purpose the notion of the weak derivative and Sobolev spaces. The space , , consists of functions whose weak derivatives up to order are in , and , is the fractional Sobolev space, we note that for we have . A good source of information about the Sobolev spaces is [13].
If we
A discrete form of the conserved equation
We will approximate the conserved transport equation Eq. (2) using a control volume scheme [19], [20]. The first step is to lay down a grid of equally spaced () node points, see Fig. 1. Around the internal nodes , we construct control volumes of length , for a given node point located at , the and faces of the control volume located at positions and respectively. Half control volumes (length ) are associated with the end-point nodes
Specific discrete flux definitions
The convenience of Eq. (17), and its boundary treatments Eqs. (18) to (21), is that, due to the conservative form, its operation will be invariant under any appropriate flux definition. Thus making the numerical solution, simply requires the appropriate construction of the discrete fluxes at the control volume faces.
Numerical analysis considerations
Before we consider the physical consequences of using the various discrete flux models, suggested above, in our explicit time integration of the heat balance, Eq. (17), it is worthwhile presenting the key numerical analysis results associated with these schemes. The consistency, stability, and convergence when the Fourier flux (22) is used, are classic results in the literature, see e.g. [22] or [23], as such we will focus most attention on the discrete fractional fluxes, Caputo (C) (28) and
The decay property
A well recognized feature of the classic diffusion treatment, based on the Fourier law, see Eq. (4), is that, over time, it tends to smooth out the predicted profile of the conserved value. For example, if an initial pulse is provided within , with a no-flux condition at and a fixed value , at , a Fourier diffusion treatment will continue to transport the conserved quantity, to both the left or the right, until all the local slopes, at all points in the domain, are zero. In
A suggestion for a parsimonious fractional flux definition
From the above it is clear that the differences between conserved non-local flux models can be quite pronounced. This feature significantly undermines our ability to conduct rigorous analysis of the governing equations and to construct physical and consistent numerical solution treatments. One way in which the ambiguity between various non-local flux models might be removed, is to modify the current phenomenological model for the heat flux. Our proposal here is, in an finite interval ,
Conclusion
Our work here has focused on the analytical and numerical study of conserved statements of diffusion transport which employ a non-local flux-laws constructed in terms of fractional derivative operators—space fractional diffusion equations.
The key contributions include:
- •
A presentation and discussion of available results on well posedness of fractional diffusion equations.
- •
A conserved, control volume discrete solution of fractional diffusion equations, with the development of associated boundary
Acknowledgments
The authors thank BIRS in Banff for creating a stimulating environment at the workshop Advanced Developments for Surface and Interface Dynamics—Analysis and Computation, where this paper was initiated. VV notes the support of the James L. Record Chair, from the Department of Civil, Environmental and Geo- Engineering, University of Minnesota. A part of the research for this paper was done at the University of Warsaw, where PR and VV enjoyed a partial support of the National Science Centre, Poland
References (26)
- et al.
Boundary conditions for fractional diffusion
J. Comput. Appl. Math.
(2018) - et al.
Fractional partial differential equations with boundary conditions
J. Differential Equations
(2018) An analytic semigroup generated by a fractional differential operator
J. Math. Anal. Appl.
(2020)- et al.
The random Walk’s guide to anomalous diffusion: a fractional dynamics approach
Phys. Rev.
(2000) - et al.
Lipschitz regularity of solutions for mixed integro-differential equations
J. Differential Equations
(2012) - et al.
Stability and convergence of a finite volume method for the space fractional advection–dispersion equation
J. Comput. Appl. Math.
(2014) - et al.
Finite difference approximations for two-sided space-fractional partial differential equations
Appl. Numer. Math.
(2006) - et al.
Space-fractional advectione dispersion equations with variable parameters: Diverse formulas, numerical solutions and application to the macro dispersion experiment site data
Water Resour. Res.
(2007) - et al.
Fractional advection dispersion equations for modeling transport at the earth surface
J. Geophys. Res.
(2009) Anomalous heat transfer: Examples, fundamentals, and fractional calculus models
Adv. Heat Transf.
(2018)
Space–time fractional Dirichlet problems
Math. Nachr.
On viscosity solutions of space-fractional diffusion equations of Caputo type
SIAM J. Math. Anal.
Analytical Theory of Heat
Cited by (8)
Fractional models of falling object with linear and quadratic frictional forces considering Caputo derivative
2023, Chaos, Solitons and FractalsCitation Excerpt :Numerical and analytical solutions were constructed for the fractional neutron point kinetic equations modeling a nuclear reactor in [20]. Comparisons of the Caputo and Riemann–Liouville fractional models in [21] proved that Riemann–Liouville results may be physically and mathematically unsound for the diffusion transport equation. The stochastic resonance event in the signal-to-noise ratio using a fractional nonlinear model was studied in [22].
Fractional-derivative models for non-Fickian transport in a single fracture and its extension
2020, Journal of HydrologyCitation Excerpt :Zhang et al. (2013a) demonstrated that the FDE with the space-fractional derivative (2) can efficiently fit the observed early arrivals of tracers transport in a single fracture. It is also noteworthy that using the Caputo derivative with order between 1 and 2 in a diffusion equation can generate negative values (Baeumer et al., 2018a) (note that Namba et al. (2021) demonstrated that, when the diffusion term was modeled as the integer divergence of the Caputo fractional gradient with an order between 0 and 1, the model retains both sound mass balance and mathematical properties), while the Riemman-Liouville space fractional derivative is needed to define a well-defined stochastic process especially in bounded domains (Baeumer et al., 2018b), such as non-Fickian diffusion in the fracture. Therefore, two different types of fractional derivatives are needed in the FDE to capture mixed non-Fickian transport in the single fracture-matrix system.
Special solutions to the space fractional diffusion problem
2022, Fractional Calculus and Applied AnalysisA ONE-PHASE SPACE-FRACTIONAL STEFAN PROBLEM WITH NO LIQUID INITIAL DOMAIN
2022, SIAM Journal on Mathematical Analysis