Advancing circular economy principles through wild black soldier flies

Atinuke Chineme; Getachew Assefa; Irene M. Herremans; Barry Wylant; Marwa Shumo; Aliceanna Shoo; Mturi James; Frida Ngalesoni; Anthony Ndjovu; Steve Mbuligwe; Mike Yhedgo; Atinuke Chineme; Getachew Assefa; Irene M. Herremans; Barry Wylant; Marwa Shumo; Aliceanna Shoo; Mturi James; Frida Ngalesoni; Anthony Ndjovu; Steve Mbuligwe; Mike Yhedgo

doi:10.3934/environsci.2023047

AIMS Environmental Science

2023, Volume 10, Issue 6: 868-893. doi: 10.3934/environsci.2023047

Previous Article Next Article

Research article Special Issues

Advancing circular economy principles through wild black soldier flies

1.
School of Architecture, Planning, and Landscape, University of Calgary, Calgary, Canada
2.
Haskayne School of Business, University of Calgary, Calgary, Canada
3.
Center for Development Research (ZEF), University of Bonn, Germany
4.
AMREF Health Africa Tanzania
5.
School of Environmental Science and Technology, Ardhi University, Dar es Salaam, Tanzania
6.
Environmental Resources Consultancy, Dar es Salaam, Tanzania

Received: 29 August 2023 Revised: 28 November 2023 Accepted: 07 December 2023 Published: 18 December 2023

Biowaste management poses a significant and widespread challenge. However, its consideration as a resource has led to the emergence of innovative and sustainable biowaste management techniques. One such promising solution is the use of black soldier flies (BSF) in biowaste treatment. This technique offers various advantages, such as the transformation of biowaste into versatile products that can be used in agriculture, horticulture, aquaculture, animal husbandry, pharmaceuticals and energy production. Despite significant research on different aspects of the BSF biowaste treatment system, none have explored the application of circular economy principles in low-income settings using naturally occurring BSF, i.e., free-range BSF. This article addresses the gap utilizing a mixed-method approach through a case study to achieve two objectives: the localization of the circular economy through co-production with a community group and the viable production of black soldier fly larvae (BSFL) and compost to meet the community's needs. Through collaboration, a successful circular economy was established as biowaste was transformed into products and safely reintroduced into the local biosphere. Performance indices used included BSFL nutritional composition, harvest rates and heavy metal absence.

Through community involvement, circular economy principles were effectively implemented to redirect a retail market's fruit and vegetable waste from the landfill. The free-range open system produced 19.15 kg of BSFL, with 44.34% protein content, 20.6% crude fat and zero heavy metals. These outcomes align with existing research, indicating that a community-led open BSFL system can generate valuable products while fostering circular economy principles with minimal financial, technological, energy and water resources.

Keywords:

Citation: Atinuke Chineme, Getachew Assefa, Irene M. Herremans, Barry Wylant, Marwa Shumo, Aliceanna Shoo, Mturi James, Frida Ngalesoni, Anthony Ndjovu, Steve Mbuligwe, Mike Yhedgo. Advancing circular economy principles through wild black soldier flies[J]. AIMS Environmental Science, 2023, 10(6): 868-893. doi: 10.3934/environsci.2023047

Related Papers:

[1]	Jiangtao Dai, Ge Guo . A leader-following consensus of multi-agent systems with actuator saturation and semi-Markov switching topologies. Mathematical Biosciences and Engineering, 2024, 21(4): 4908-4926. doi: 10.3934/mbe.2024217
[2]	Duoduo Zhao, Fang Gao, Jinde Cao, Xiaoxin Li, Xiaoqin Ma . Mean-square consensus of a semi-Markov jump multi-agent system based on event-triggered stochastic sampling. Mathematical Biosciences and Engineering, 2023, 20(8): 14241-14259. doi: 10.3934/mbe.2023637
[3]	Na Zhang, Jianwei Xia, Tianjiao Liu, Chengyuan Yan, Xiao Wang . Dynamic event-triggered adaptive finite-time consensus control for multi-agent systems with time-varying actuator faults. Mathematical Biosciences and Engineering, 2023, 20(5): 7761-7783. doi: 10.3934/mbe.2023335
[4]	Xiaofeng Chen . Double-integrator consensus for a switching network without dwell time. Mathematical Biosciences and Engineering, 2023, 20(7): 11627-11643. doi: 10.3934/mbe.2023516
[5]	Wangming Lu, Zhiyong Yu, Zhanheng Chen, Haijun Jiang . Prescribed-time cluster practical consensus for nonlinear multi-agent systems based on event-triggered mechanism. Mathematical Biosciences and Engineering, 2024, 21(3): 4440-4462. doi: 10.3934/mbe.2024196
[6]	Qiushi Wang, Hongwei Ren, Zhiping Peng, Junlin Huang . Dynamic event-triggered consensus control for nonlinear multi-agent systems under DoS attacks. Mathematical Biosciences and Engineering, 2024, 21(2): 3304-3318. doi: 10.3934/mbe.2024146
[7]	Yan-Xiao Liu , Ching-Nung Yang, Qin-Dong Sun . Enhance embedding capacity for switch map based multi-group EMD data hiding. Mathematical Biosciences and Engineering, 2019, 16(5): 3382-3392. doi: 10.3934/mbe.2019169
[8]	Yuting Zhu, Wenyu Zhang, Junjie Hou, Hainan Wang, Tingting Wang, Haining Wang . The large-scale group consensus multi-attribute decision-making method based on probabilistic dual hesitant fuzzy sets. Mathematical Biosciences and Engineering, 2024, 21(3): 3944-3966. doi: 10.3934/mbe.2024175
[9]	Mingxia Gu, Zhiyong Yu, Haijun Jiang, Da Huang . Distributed consensus of discrete time-varying linear multi-agent systems with event-triggered intermittent control. Mathematical Biosciences and Engineering, 2024, 21(1): 415-443. doi: 10.3934/mbe.2024019
[10]	Meiqiao Wang, Wuquan Li . Distributed adaptive control for nonlinear multi-agent systems with nonlinear parametric uncertainties. Mathematical Biosciences and Engineering, 2023, 20(7): 12908-12922. doi: 10.3934/mbe.2023576

Abstract

1. Introduction

Assuming that the underlying asset process follows a geometric Brownian motion, the Black-Scholes model was firstly proposed in 1973, where the option value satisfies a partial difference equation and depends only on the risk-free interest rate and the volatility of asset price ^[1]. In order to fit the empirical facts of practical financial markets, more extended models were introduced and studied, including jump-diffusion models ^[2,3], stochastic volatility models ^[4], fractional differential models ^[5,6,7] and regime-switching models ^[8].

The idea of switching regimes is prevalently applied in order to allow Lévy processes to switch in a finite state space by a Markov chain. In option pricing models, the parameters, such as interest rate, drift and volatility, are allowed to take diverse values in a finite number of regimes ^[9]. For instance, the option models based on exponential Lévy processes under switching regimes were proposed and widely discussed to capture the sudden state movement from the bull market to bear market, and deal with the non-stationary behavior ^[10,11,12]. Since the partial integro-differential equation (PIDE) derived from the regime-switching exponential Lévy processes is difficult to be solved in closed form, it is essential to develop effective numerical methods.

Recently, numerical solution of fractional option models under switching regimes attract much attention from the community of financial engineering. Cartea and del-Castillo-Negrete ^[13] proposed a first-order shifted Grünwald difference formula for the option pricing models, including the finite moment log stable (FMLS) model ^[6], CGMY model ^[7] and KoBoL model ^[5]. A first-order penalty method for fractional regime-switching American option pricing models, was constructed in ^[14]. Further, an implicit-explicit preconditioned direct method was developed in ^[12] for fractional regime-switching models which was of first order in spatial approximation.

In this paper, we consider a second-order numerical scheme for fractional regime-switching option pricing models, based on the weighted and shifted Grünwald difference (WSGD) formula and Crank-Nicolson scheme. Theoretical analysis on the stability and second-order convergence of the numerical scheme is studied in detail. A second-order ADI method is proposed to accelerate the computation with a preconditioned direct solver for the discrete linear system. Numerical experiments on both fractional PDE and multi-regime FMLS and CGMY models are presented to show the convergence and efficiency of the proposed approach.

The structure of this paper is organized as follows: A second-order numerical scheme for fractional regime-switching option pricing models is presented in Section 2. Numerical analysis of stability and the second-order convergence are shown in Section 3. In Section 4, we introduce the ADI method with preconditioned direct solver for the discrete linear system. Numerical experiments in Section 5 demonstrated the convergence and efficiency of the proposed method. Finally, conclusions are drawn in Section 6.

2. A second-order finite difference method

Under the risk-neutral measure, assume that the stock price $S_t$ follows a geometric Lévy process

$\mathrm{d}\left(\ln S_t\right) = (r-\nu) \mathrm{d} t+\mathrm{d} L_t,$

where $r$ is the risk-free rate, $\nu$ is a convexity adjustment and $\mathrm{d} L_t$ is the increment of a Lévy process under the equivalent martingale measure ^[15]. Below, we discuss the general fractional regime-switching option model derivated by three particular Lévy processes: LS, CGMY and KoBoL, see ^[13] for more details.

Let $V_s(x, t)$ be the value of an European option in state $s$ , the fractional regime-switching option model is defined by

$\begin{equation} \begin{aligned} \frac { \partial V_s(x,t) } { \partial t } +c_{s,1} \frac { \partial V_s(x,t) } { \partial x } +c_{s,2}{D_{+} ^ {\xi_s, \alpha_s }} V_s(x,t)&+c_{s,3}{D_{-} ^ {\lambda_s, \alpha_s }} V_s(x,t) \\ &-d_s V_s(x,t)+\sum\limits_{j = 1}^{\bar{S}} q_{s, j} V_j(x, t) = 0, \end{aligned} \end{equation}$

(2.1)

where $x = \ln S_t$ , $1 < \alpha_s < 2, s = 1, 2, \ldots, \bar{S}$ . The other parameters in Eq (2.1) depend on a certain state of a Markov process in the finite set $\{ 1, 2, \ldots, \bar{S} \}$ . The constants $q_{s, j}$ represent the elements of the state transition matrix of the Markov process, which satisfy the conditions $\sum_{j = 1}^{\bar{S}} q_{s, j} = 0$ and $q_{s, j} \geq 0, \forall j\neq s$ .

The left and right Riemann-Liouville tempered fractional derivatives $D_{+}^{\xi_s, \alpha_s}$ and $D_{-}^{\lambda_s, \alpha_s}$ are defined by

$\begin{equation} \begin{aligned} D_{+}^{\xi_s, \alpha_s} V_s(x, t) & = \frac{e^{\xi_s x}}{\Gamma\left(2-\alpha_s\right)} \frac{\partial^2}{\partial x^2} \int_x^{\infty} \frac{e^{-\xi_s \zeta} V_s(\zeta, t)}{(\zeta-x)^{\alpha_s-1}} d \zeta, \\ D_{-}^{\lambda_s, \alpha_s} V_s(x, t) & = \frac{e^{-\lambda_s x}}{\Gamma\left(2-\alpha_s\right)} \frac{\partial^2}{\partial x^2} \int_{-\infty}^x \frac{e^{\lambda_s \zeta} V_s(\zeta, t)}{(x-\zeta)^{\alpha_s-1}} d \zeta, \end{aligned} \end{equation}$

(2.2)

where $\Gamma(\cdot)$ is the Gamma function.

In the CGMY model, the parameters in the model (2.1) are given by

$\begin{equation} \begin{aligned} &c_{s,1} = r-C \Gamma(\alpha_s)\left[(\xi_s-1)^{\alpha_s}-\xi_s^{\alpha_s}+(\lambda_s+1)^{\alpha_s}-\lambda_s^{\alpha_s}\right], \\ &c_{s,2} = c_{s,3} = C \Gamma(-\alpha_s) ,\quad d_s = r+C \Gamma(-\alpha_s)\left(\xi_s^{\alpha_s}+\lambda_s^{\alpha_s}\right), \\ &C > 0, \quad \lambda_s \geq 0, \quad \xi_s \geq 0. \end{aligned} \end{equation}$

(2.3)

The model (2.1) also covers FMLS and KoBoL models by different choices, and we refer the readers to ^[16] for more details.

The terminal and boundary conditions for call options are given by

$\begin{equation} \begin{array}{lll} V_s(x, T) = \max \left\{e^x-K, 0\right\}, & x_l \leq x \leq x_r, \\ V_s(x_l, t) = 0, & 0 \leq t < T, \\ V_s(x_r, t) = e^{x_r}-K e^{-r(T-t)}, & 0 \leq t < T, \end{array} \end{equation}$

(2.4)

where $K$ is the strike price.

Let $N$ and $M$ be the number of the uniform discrete points in the space and time direction, respectively, and let $h = \left(x _ { r } - x _ { l } \right) / (N+1)$ and $\tau = T / M$ be the corresponding step length. Define $t _ { m } = m \tau (m = 0, 1, 2, \cdots, M), x _ { n } = x_ { l } + n h (n = 0, 1, 2, \cdots, N+1)$ .

Assume that the function $V_s(x, t)$ is continuously differentiable and $\partial^2 V_s(x, t)/\partial x^2$ is integrable in the interval $[0, T)\times [x_l, x_r]$ , then for every $\alpha \; (0 < \alpha < 2)$ , the Riemann-Liouville derivative of $V_s(x, t)$ exists and coincides with the Grünwald-Letnikov type ^[17]. Hence, we can use the Grünwald difference approaches to approximate the tempered fractional derivatives in Eq (2.2) to avoid the strong singularity when $\zeta = x$ .

The first-order shifted Grünwald difference scheme ^[18] was first proposed to approximate the fractional derivatives and used to solve the fractional option pricing models ^[12,14]. Now we consider the second-order weighted and shifted Grünwald difference scheme ^[19] in the discrete process of Eq (2.1).

Based on the weighted and shifted Grünwald difference scheme, the fractional derivative can be approximated by

$\begin{equation} \begin{aligned} D_{+}^{\xi_s, \alpha_s} V_s(x_n, t_m) & = \frac { 1 } { h ^ { \alpha_s } } \sum\limits _ { k = 0 } ^ { N-n + 2 } \omega _ { k } ^ { \alpha_s } e^{-\xi_s(k-1)h}V_s \left( x _ { n +k - 1 },t_m \right) + \mathcal{O} \left( h ^ { 2 } \right),\\ D_{-}^{\lambda_s, \alpha_s} V_s(x_n, t_m) & = \frac { 1 } { h ^ { \alpha_s } } \sum\limits _ { k = 0 } ^ { n + 1 } \omega _ { k } ^ { \alpha_s } e^{-\lambda_s(k-1)h}V_s \left( x _ { n - k + 1 },t_m \right) + \mathcal{O} \left( h ^ { 2 } \right), \end{aligned} \end{equation}$

(2.5)

where

$\begin{equation} \left\{ \begin{array} { l } \begin{aligned} &{ \omega _ { 0 } ^ { \alpha_s } = \frac { \alpha_s } { 2 } g _ { 0 } ^ { \alpha_s } ,\quad \omega _ { k } ^ { \alpha_s } = \frac { \alpha_s } { 2 } g _ { k } ^ { \alpha_s } + \frac { 2 - \alpha_s } { 2 } g _ { k - 1 } ^ { \alpha_s } } ,\\ &{ g _ { 0 } ^ { \alpha_s } = 1 ,\quad g _ { k } ^ { \alpha_s } = \left( 1 - \frac { \alpha_s + 1 } { k } \right) g _ { k - 1 } ^ { \alpha_s } , \quad k = 1,2 , \cdots ,} \\ &{\sum\limits_{k = 0}^{\infty}g _ { k } ^ { \alpha_s } = 0,\quad g _ { 1 } ^ { \alpha_s } = -\alpha_s < 0\quad g _ { 2} ^ { \alpha_s } > g _ { 3} ^ { \alpha_s } > \ldots > 0,}\\ &\omega_0^{\alpha_s} = \frac{\alpha_s}{2}, \quad \omega_1^{\alpha_s} = \frac{2-\alpha_s-\alpha_s^2}{2} < 0, \quad \omega_2^{\alpha_s} = \frac{\alpha_s\left(\alpha_s^2+\alpha_s-4\right)}{4}. \end{aligned} \end{array} \right. \end{equation}$

(2.6)

Denote $V_{s, n}^m$ be the numerical solution at the discrete point $(x_n, t_m)$ of regime $s$ . Discretising the convection term and the time term in Eq (2.1) by central differences, and the Crank-Nicolson scheme respectively, and introducing the time-reverse transformation $t^* = T - t$ , dropping $*$ for simplicity, the following fully discrete scheme is derived:

$\begin{equation} \begin{aligned} \frac { V_{s,n}^{m+1}-V_{s,n}^{m} } { \tau } = \frac{1}{2}&\left[c_{s,1} \frac { V_{s,n+1}^{m+1}-V_{s,n-1}^{m+1} } { 2h } + \frac{c_{s,2}}{h^{\alpha_s}}\sum\limits _ { k = 0 } ^ { N-n + 2 } \omega _ { k } ^ { \alpha_s } e^{-\xi_s(k-1)h}V_{s , n +k - 1 }^{m+1}\right. \\ &+\frac{c_{s,3}}{h^{\alpha_s}}\sum\limits _ { k = 0 } ^ { n + 1 } \omega _ { k } ^ { \alpha_s } e^{-\lambda_s(k-1)h}V_{s, n - k + 1 }^{m+1} -d_s V_{s,n}^{m+1}+\sum\limits_{j = 1}^{\bar{S}} q_{s, j} V_{j,n}^{m+1}\\ &+c_{s,1} \frac { V_{s,n+1}^{m}-V_{s,n-1}^{m} } { 2h } + \frac{c_{s,2}}{h^{\alpha_s}}\sum\limits _ { k = 0 } ^ { N-n + 2 } \omega _ { k } ^ { \alpha_s } e^{-\xi_s(k-1)h}V_{s , n +k - 1 }^{m}\\ &\left. +\frac{c_{s,3}}{h^{\alpha_s}}\sum\limits _ { k = 0 } ^ { n + 1 } \omega _ { k } ^ { \alpha_s } e^{-\lambda_s(k-1)h}V_{s, n - k + 1 }^{m} -d_s V_{s,n}^{m}+\sum\limits_{j = 1}^{\bar{S}} q_{s, j} V_{j,n}^{m}\right],\\ &n = 1,2,\ldots,N,\quad m = 0,1,2,\ldots,M-1. \end{aligned} \end{equation}$

(2.7)

Denote $V^m = \left(V_{1, 1}^m, V_{1, 2}^m, \ldots, V_{1, N}^m, V_{2, 1}^m, \ldots, V_{2, N}^m, \ldots, V_{\bar{S}, N}^m\right)^{T}$ , $Q = \left[q_{s, j}\right]_{s, j = 1}^{\bar{S}}$ and $p^{m+\frac{1}{2}} = \frac{1}{2}\tau(p^{m+1}+p^{m})$ , where

$\begin{equation*} \begin{aligned} p^m& = (p_1^m,p_2^m,\ldots,p_{\bar{S}}^m)^{T},\\ p_s^m& = \frac{ c_{s,1}}{2h}p_{s,1}^m+\frac{ c_{s,2}}{h^{\alpha_s}}p_{s,2}^m+\frac{ c_{s,3}}{h^{\alpha_s}}p_{s,3}^m,\\ p_{s,1}^m& = (-V_{s,0}^m,0,\ldots,0,V_{s,N+1}^m)^{T},\\ p_{s,2}^m& = (\omega_0^{\alpha_s}e^{\xi_s h}V_{s,0}^m+\omega_{N+1}^{\alpha_s}e^{-\xi_s Nh}V_{s,N+1}^m,\ldots,\omega_2^{\alpha_s}e^{-\xi_s h}V_{s,N+1}^m)^{T},\\ p_{s,2}^m& = (\omega_2^{\alpha_s}e^{-\lambda_s h}V_{s,0}^m,\ldots,\omega_{N+1}^{\alpha_s}e^{-\lambda_s Nh}V_{s,0}^m+\omega_0^{\alpha_s} e^{\lambda_s h}V_{s,N+1}^m)^{T}. \end{aligned} \end{equation*}$

The matrix form of the numerical scheme (2.7) can be written as:

$\begin{equation} \left(I_{\bar{S}N}-\frac{1}{2}\tau (M_B+Q\otimes I_N) \right)V^{m+1} = \left(I_{\bar{S}N}+\frac{1}{2}\tau (M_B+Q\otimes I_N) \right)V^{m}+p^{m+\frac{1}{2}}, \end{equation}$

(2.8)

where

$\begin{equation} \begin{array}{l} M_B = \operatorname{diag}\left(T_1, T_2, \ldots, T_{\bar{S}}\right),\quad T_s = \frac{ c_{s, 1}}{2h} J+\frac{ c_{s, 2}}{h^{\alpha_s}} W_s+\frac{c_{s, 3}}{h^{\alpha_s}} G_s-d_s I_N, \end{array} \end{equation}$

(2.9)

with

$\begin{equation*} \begin{aligned} J& = \operatorname{tridiag}(-1,0,1),\\ W_s& = \left(\begin{array}{ccccc} \omega_1^{\alpha_s} & \omega_2^{\alpha_s}e^{-\xi_s h} & \cdots& \omega_{N-1}^{\alpha_s}e^{-\xi_s (N-2)h} & \omega_{N}^{\alpha_s}e^{-\xi_s (N-1)h} \\ \omega_0^{\alpha_s}e^{\xi_s h} & \omega_1^{\alpha_s} & \omega_2^{\alpha_s}e^{-\xi_s h} & \ddots & \omega_{N-1}^{\alpha_s}e^{-\xi_s (N-2)h} \\ 0 & \ddots & \ddots & \ddots & \vdots \\ \vdots & \ddots & \omega_0^{\alpha_s}e^{\xi_s h} & \omega_1^{\alpha_s} & \omega_2^{\alpha_s}e^{-\xi_s h} \\ 0 & \cdots & 0 & \omega_0^{\alpha_s}e^{\xi_s h} & \omega_1^{\alpha_s} \end{array}\right),\\ G_s& = \left(\begin{array}{ccccc} \omega_1^{\alpha_s} & \omega_0^{\alpha_s}e^{\lambda_s h} & 0 & \cdots & 0 \\ \omega_2^{\alpha_s}e^{-\lambda_s h} & \omega_1^{\alpha_s} & \omega_0^{\alpha_s}e^{\lambda_s h} & \ddots & \vdots \\ \vdots & \ddots & \ddots & \ddots & 0 \\ \omega_{N-1}^{\alpha_s}e^{-\lambda_s (N-2)h} & \ddots & \omega_2^{\alpha_s}e^{-\lambda_s h} & \omega_1^{\alpha_s} & \omega_0^{\alpha_s} e^{\lambda_s h}\\ \omega_{N}^{\alpha_s}e^{-\lambda_s (N-1)h} & \omega_{N-1}^{\alpha_s}e^{-\lambda_s (N-2)h} & \cdots & \omega_2^{\alpha_s}e^{-\lambda_s h} & \omega_1^{\alpha_s} \end{array}\right). \end{aligned} \end{equation*}$

3. Stability and convergence of the numerical scheme

In this section, the stability and convergence of the numerical scheme (2.8) are established.

A matrix is called positive definite if, and only, if its symmetric part is positive definite, that is, all the eigenvalues are positive.

Lemma 3.1. (Gerschgorin Disk Theorem) Suppose $A = \left[a_{i j}\right] \in \mathbb{C}^{n \times n}$ , let

$G_i(A) = \left\{z \in \mathbb{C}:\left|z-a_{i i}\right| \leqslant \sum\limits_{j \neq i}\left|a_{i j}\right|\right\}, \quad i = 1, \ldots, n,$

then

$\lambda(A) \subset G_1(A) \cup G_2(A) \cup \cdots \cup G_n(A).$

Theorem 3.1. (Stability) Assume that $1 < \alpha_s < 2$ and set

$\begin{equation} \eta_s (x): = (\alpha_s e^{h x }+2-\alpha_s)(1-e^{- h x})^{\alpha_s}. \end{equation}$

(3.1)

$\begin{equation} \frac{ c_{s, 2}}{2h^{\alpha_s}}\eta_s (\xi_s) +\frac{ c_{s, 3}}{2h^{\alpha_s}}\eta_s (\lambda_s) \leq d_s-\frac{1}{2} \left(q_{s,s}+\sum\limits_{k = 1,k\neq s}^{\bar{S}}q_{k,s}\right), \end{equation}$

(3.2)

for all $s = 1, 2, \ldots, \bar{S}$ , then the discretisation scheme (2.8) is stable.

Proof. Denote $B = -M_B-Q\otimes I_N$ and consider the matrix

$Z = \left(I+\frac{1}{2} \tau B\right)^{-1}\left(I-\frac{1}{2} \tau B\right).$

In order to show the stability of the scheme (2.8), it suffices to prove that the eigenvalues $\lambda_Z$ of the matrix $Z$ satisfy the estimate $\left|\lambda_Z\right| < 1$ . Or equivalently, that the eigenvalues $\lambda_B$ of matrix $B$ satisfy the estimate

$\begin{equation} \left|\frac{1-1 / 2 \tau \lambda_B}{1+1 / 2 \tau \lambda_B}\right| < 1. \end{equation}$

(3.3)

The inequality (3.3) means that any $\lambda_B$ has a positive real part $\Re(\lambda_B)$ . Therefore, the numerical scheme (2.8) is stable if the matrix $B$ is positive definite, i.e., its symmetric part $\mathcal{B}$ is positive definite.

Consider the symmetric Toeplitz matrix

$\begin{equation*} \mathcal{T}_s = -\frac{ c_{s, 2}}{2h^{\alpha_s}}( W_s+ W_s^T)-\frac{c_{s, 3}}{2h^{\alpha_s}}( G_s+G_s^T)+d_s I_N, \end{equation*}$

and block diagonal Toeplitz matrix

$\begin{equation*} \mathcal{T} = \operatorname{diag}\left(\mathcal{T}_1, \mathcal{T}_2, \ldots, \mathcal{T}_{\bar{S}}\right), \end{equation*}$

thus,

$\mathcal{B} = \mathcal{T}-\frac{Q+Q^T}{2}\otimes I_N.$

Note that $c_{s, 2}, c_{s, 3}$ and $d_s$ are nonnegative, we have

$[\mathcal{B}]_{l,l} = -\frac{ (c_{s, 2}+c_{s, 3})\omega_1^{\alpha_s}}{h^{\alpha_s}}+d_s-q_{s,s} > 0,$

where $1+(s-1)N\leq l \leq sN$ , $s = 1, 2, \ldots, \bar{S}$ .

Therefore, if the matrix $\mathcal{B}$ is strictly row diagonally dominant, then it is positive definite by Lemma 3.1, which means $\mathcal{B}$ satisfies the condition

$\begin{equation} [\mathcal{B}]_{l,l} > \sum\limits_{k = 1,k\neq l}^{\bar{S}N}\left| [\mathcal{B}]_{l,k} \right| \end{equation}$

(3.4)

for $1+(s-1)N\leq l \leq sN$ where $s = 1, 2, \ldots, \bar{S}$ .

It is clear that the $l$ th and the $(1+(2s-1)N-l)$ th rows are the same. Without loss of generality, we choose $1+(s-1)N \leq l \leq (s-1)N+ \lceil\frac{N}{2} \rceil$ , then

$\begin{equation*} \begin{aligned} \sum\limits_{k = 1,k\neq l}^{\bar{S}N}\left| [\mathcal{B}]_{l,k} \right| = \frac{ c_{s, 2}}{2h^{\alpha_s}}&\left( 2(\omega_0^{\alpha_s}e^{\xi_s h}+\omega_2^{\alpha_s}e^{-\xi_s h}+\cdots+\omega_{\ell}^{\alpha_s}e^{-\xi_s(\ell-1) h})\right.\\ &\left. +\omega_{\ell+1}^{\alpha_s}e^{-\xi_s\ell h}+\cdots+\omega_{N}^{\alpha_s}e^{-\xi_s(N-1) h} \right)\\ &+\frac{ c_{s, 3}}{2h^{\alpha_s}}\left( 2(\omega_0^{\alpha_s}e^{\lambda_s h}+\omega_2^{\alpha_s}e^{-\lambda_s h}+\cdots+\omega_{\ell}^{\alpha_s}e^{-\lambda_s(\ell-1) h})\right.\\ &\left. +\omega_{\ell+1}^{\alpha_s}e^{-\lambda_s\ell h}+\cdots+\omega_{N}^{\alpha_s}e^{-\lambda_s(N-1) h} \right)+\sum\limits_{k = 1,k\neq s}^{\bar{S}}\frac{q_{s,k}+q_{k,s}}{2}, \end{aligned} \end{equation*}$

where $\ell = l-(s-1)N$ . By rearranging the sequence $\{ g_k^{\alpha_s}\}$ from $\{ \omega_k^{\alpha_s}\}$ and according to the properties in Eq (2.6), we have

$\begin{equation*} \begin{aligned} &2(\omega_0^{\alpha_s}e^{\xi_s h}+\omega_2^{\alpha_s}e^{-\xi_s h}+\cdots+\omega_{\ell}^{\alpha_s}e^{-\xi_s(\ell-1) h}) +\omega_{\ell+1}^{\alpha_s}e^{-\xi_s\ell h}+\cdots+\omega_{N}^{\alpha_s}e^{-\xi_s(N-1) h}\\ = &2\sum\limits_{k = 0}^{\ell}\omega_k^{\alpha_s}e^{\xi_s(1-k)h} +\sum\limits_{k = \ell+1}^{N}\omega_k^{\alpha_s}e^{\xi_s(1-k)h}-2\omega_1^{\alpha_s}\\ < & (\alpha_s e^{\xi_s h}+2-\alpha_s)(1-e^{-\xi_s h})^{\alpha_s}-2\omega_1^{\alpha_s}, \end{aligned} \end{equation*}$

and

$\begin{equation*} \begin{aligned} &2(\omega_0^{\alpha_s}e^{\lambda_s h}+\omega_2^{\alpha_s}e^{-\lambda_s h}+\cdots+\omega_{\ell}^{\alpha_s}e^{-\lambda_s(\ell-1) h}) +\omega_{\ell+1}^{\alpha_s}e^{-\lambda_s\ell h}+\cdots+\omega_{N}^{\alpha_s}e^{-\lambda_s(N-1) h}\\ < & (\alpha_s e^{\lambda_s h}+2-\alpha_s)(1-e^{-\lambda_s h})^{\alpha_s}-2\omega_1^{\alpha_s}. \end{aligned} \end{equation*}$

Thus, the condition (3.4) becomes

$\begin{equation*} \begin{aligned} -\frac{ (c_{s, 2}+c_{s, 3})\omega_1^{\alpha_s}}{h^{\alpha_s}}+d_s-q_{s,s}\geq& \frac{ c_{s, 2}}{2h^{\alpha_s}}(\alpha_s e^{\xi_s h}+2-\alpha_s)(1-e^{-\xi_s h})^{\alpha_s}-\frac{ c_{s, 2}}{h^{\alpha_s}}\omega_1^{\alpha_s}\\ &+\frac{ c_{s, 3}}{2h^{\alpha_s}}(\alpha_s e^{\lambda_s h}+2-\alpha_s)(1-e^{-\lambda_s h})^{\alpha_s}-\frac{ c_{s, 3}}{h^{\alpha_s}}\omega_1^{\alpha_s}\\ &-\frac{q_{s,s}}{2}+\frac{1}{2}\sum\limits_{k = 1,k\neq s}^{\bar{S}}q_{k,s}, \end{aligned} \end{equation*}$

which can be written as

$\begin{equation*} \frac{ c_{s, 2}}{2h^{\alpha_s}}\eta_s (\xi_s) +\frac{ c_{s, 3}}{2h^{\alpha_s}}\eta_s (\lambda_s) \leq d_s-\frac{1}{2} \left(q_{s,s}+\sum\limits_{k = 1,k\neq s}^{\bar{S}}q_{k,s}\right), \end{equation*}$

where $\eta_s(x)$ is defined in Eq (3.1).

It is similar that the stability condition in Theorem 3.1 from ^[16] is given by

$\begin{equation} \frac{ c_{s, 2}\eta_s (\xi_s)}{h^{\alpha_s}(1+e^{\alpha_s(\lambda_s-\xi_s)h})} +\frac{ c_{s, 3}\eta_s (\lambda_s)}{h^{\alpha_s}(1+e^{\alpha_s(\xi_s-\lambda_s)h})} \leq d_s-\frac{1}{2} \left(q_{s,s}+\sum\limits_{k = 1,k\neq s}^{\bar{S}}q_{k,s}\right). \end{equation}$

(3.5)

Consider the specific parameters of the CGMY model from Eq (2.3), the stability condition (3.2) can be rewritten as

$C\Gamma(-\alpha_s)\left(\frac{\eta_s (\xi_s)+\eta_s (\lambda_s)}{2h^{\alpha_s}}-\xi_s^{\alpha_s}-\lambda_s^{\alpha_s}\right) \leq r-\frac{1}{2} \left(q_{s,s}+\sum\limits_{k = 1,k\neq s}^{\bar{S}}q_{k,s}\right),$

while condition (3.5) turns to

$C\Gamma(-\alpha_s)\left(\frac{ \eta_s (\xi_s)}{h^{\alpha_s}(1+e^{\alpha_s(\lambda_s-\xi_s)h})} +\frac{\eta_s (\lambda_s)}{h^{\alpha_s}(1+e^{\alpha_s(\xi_s-\lambda_s)h})} -\xi_s^{\alpha_s}-\lambda_s^{\alpha_s}\right) \leq r-\frac{1}{2} \left(q_{s,s}+\sum\limits_{k = 1,k\neq s}^{\bar{S}}q_{k,s}\right).$

It is can be seen that the condition (3.2) allows a wider range of the parameters in Eq (2.1), which can describe more state movement as the financial markets change.

Consider now the convergence of the scheme (2.8). Due to the non-smoothness of the initial and boundary conditions in Eq (2.4), Eq (2.1) does not have a solution in the classical form. As a result, we consider the generalized solution, which satisfies the fractional PDE almost everywhere in $(0, T)\times (x_l, x_r)$ . We define the viscosity solution of Eq (2.1) similar as Definition 2.4 in ^[20].

Theorem 3.2. (Convergence) Let ${V}_\ast^m$ be the viscosity solution of Eq (2.1). The scheme (2.8) is of second order convergence, i.e.,

$\begin{equation} \| {V}^m-V_\ast^m \| \leq C_0 (h^2+\tau^2) \end{equation}$

(3.6)

if the matrix $B = -M_B-Q\otimes I_N$ is positive definite, i.e.,

where the norm $\|v \| = \sqrt{h\sum_{i = 0}^{\bar{S}N}v_i^2}$ and $C_0$ is a positive constant.

Proof. The proof is similar as Theorem 3.2 in ^[16], and we omit the specific process here.

4. The ADI method

In this section, the ADI method will be used to solve scheme (2.7). The ADI method proposed by Peaceman and Rachford ^[21] in 1955 was widely used to solve two dimensional problems due to its computational effectiveness. Recently, the ADI method was also applied to solve two-asset option pricing problems under fractional models ^[22,23].

Denote

$\begin{equation*} \begin{aligned} \Delta_s V_{s,n}^{m}& = \sum\limits_{j = 1}^{\bar{S}} q_{s, j} V_{j,n}^{m} \end{aligned} \end{equation*}$

and

$\begin{equation*} \begin{aligned} \Delta_x^s V_{s,n}^{m} = c_{s,1} \frac { V_{s,n+1}^{m}-V_{s,n-1}^{m} } { 2h } &+ \frac{c_{s,2}}{h^{\alpha_s}}\sum\limits _ { k = 0 } ^ { N-n + 2 } \omega _ { k } ^ { \alpha_s } e^{-\xi_s(k-1)h}V_{s , n +k - 1 }^{m}\\ &+\frac{c_{s,3}}{h^{\alpha_s}}\sum\limits _ { k = 0 } ^ { n + 1 } \omega _ { k } ^ { \alpha_s } e^{-\lambda_s(k-1)h}V_{s, n - k + 1 }^{m} -d_s V_{s,n}^{m}, \end{aligned} \end{equation*}$

From Eq (2.7), it is easy to show

$\begin{equation} \begin{aligned} \left(1-\frac{\tau}{2}\Delta_x^s\right) \left(1-\frac{\tau}{2}\Delta_s\right)V_{s, n}^{m+1} = \left(1+\frac{\tau}{2}\Delta_x^s\right) \left(1+\frac{\tau}{2}\Delta_s\right) V_{s, n}^{m}+R, \end{aligned} \end{equation}$

(4.1)

where

$\begin{eqnarray} R = \frac{\tau^{2}}{4}\Delta_x^s\Delta_s\left(V_{s, n}^{m+1}-V_{s, n}^{m}\right). \end{eqnarray}$

(4.2)

When the time step $\tau > 0$ is sufficiently small, we omit the term $R$ and define the following ADI scheme similar to that of the Peaceman–Rachford type ^[21]:

$\begin{align} \left(1-\frac{\tau}{2}\Delta_x^s\right) \widehat{V}_{s, n} & = \left(1+\frac{\tau}{2}\Delta_s\right) V_{s, n}^{m} , \end{align}$

(4.3)

$\begin{align} \left(1-\frac{\tau}{2}\Delta_s\right) V_{s, n}^{m+1} & = \left(1+\frac{\tau}{2}\Delta_x^s\right) \widehat{V}_{s, n} . \end{align}$

(4.4)

The following theorem illustrates the convergence order of the ADI scheme (4.3) and (4.4).

Theorem 4.1. Assume that the exact solution of the fractional PDE in Eq (2.1) is unique, and its partial derivatives are in $\mathcal{L}^1(\mathbb{R})$ and vanish outside $[0, T)\times [x_l, x_r]$ . The ADI discretization for Eq (2.1) defined in Eqs (4.3) and (4.4) is also of second order convergence $\mathcal{O}(h^2+\tau^2)$ .

Proof. From Theorem 3.2, we have proved that the Crank–Nicolson method has convergence of order $\mathcal{O}(h^2+\tau^2)$ . In the ADI scheme, compared to the Crank–Nicolson scheme (2.7), the scheme (4.3) and (4.4) incurs an additional perturbation error $R$ defined in Eq (4.2).

Since

$\begin{eqnarray*} \begin{aligned} R = \frac{\tau^{2}}{4}\Delta_x^s\Delta_s\left(V_{s, n}^{m+1}-V_{s, n}^{m}\right) & = \frac{\tau^{3}}{4}\Delta_x^s\Delta_s\left(\left.\frac{\partial V_s}{\partial t}\right|_{(x_n,t_m)}+\mathcal{O}(\tau)\right)\\ & = \frac{\tau^{3}}{4}\Delta_s\left(\mathcal{L}_s\left.\frac{\partial V_s}{\partial t}\right|_{(x_n,t_m)}+\mathcal{O}(h^2+\tau)\right), \end{aligned} \end{eqnarray*}$

where

$\begin{equation*} \begin{aligned} \mathcal{L}_sV_s = c_{s,1} \frac { \partial V_s } { \partial x } +c_{s,2}{D_{+} ^ {\xi_s, \alpha_s }} V_s&+c_{s,3}{D_{-} ^ {\lambda_s, \alpha_s }} V_s -d_s V_s. \end{aligned} \end{equation*}$

When $\tau$ is sufficiently small, the perturbation error $R$ is a higher-order term than the other terms in Eq (4.1). Therefore, the convergence order of the ADI scheme (4.3) and (4.4) is $\mathcal{O}(h^2+\tau^2)$ .

In order to solve the ADI scheme (4.3)-(4.4), we need to define boundary conditions for $\widehat{V}_{s, n}$ , which is accomplished by subtracting Eq (4.4) from Eq (4.3)

$\begin{eqnarray} 2 \widehat{V}_{s,n} = \left(1+\frac{\tau}{2}\Delta_s\right) V_{s,n}^{m}+\left(1-\frac{\tau}{2}\Delta_s\right) V_{s,n}^{m+1}. \end{eqnarray}$

(4.5)

The corresponding algorithm is implemented as follows:

Algorithm 1 ADI method for scheme (4.3), (4.4)

1: Initialize

$V_{s,n}^0$ for

$s=1,2,\ldots,\bar{S}$ and

$n=1,2,\ldots,N$ using the payoff function.
2: for

$m=0,1,2,\ldots,M-1$ , do
3: For

$s=1,2,\ldots,\bar{S}$ , solve the following system for

$\widehat{V}_{s,*}={\left(\widehat{V}_{s,1},\widehat{V}_{s,2},\ldots,\widehat{V}_{s,N} \right)}^T$ .

$\begin{eqnarray} \begin{aligned} \left(I_N-\frac{\tau}{2} T_{s}\right) \widehat{V}_{s,*}= V_{s,*}^{m}+\frac{\tau}{2} \sum_{j=1}^{\bar{S}} q_{s, j} V_{j,*}^{m}+\frac{\tau}{2}\hat{p}_s, \;\;\;\;\;\;\;\;\;\;(4.6)\end{aligned} \end{eqnarray}$
where

$T_{s}$ is defined in Eq (2.9) ,

${V}_{s,*}^n={\left({V}_{s,1}^m,{V}_{s,2}^m,\ldots,{V}_{s,N}^m \right)}^T$ ,

$\hat{p}_s=\frac{ c_{s,1}}{2h}\hat{p}_{s,1}+\frac{ c_{s,2}}{h^{\alpha_s}}\hat{p}_{s,2}+\frac{ c_{s,3}}{h^{\alpha_s}}\hat{p}_{s,3},$ with

$\begin{equation*} \begin{aligned} &\hat{p}_{s,1}=(-\widehat{V}_{s,0},0,\ldots,0,\widehat{V}_{s,N+1})^{T}, \\&\hat{p}_{s,2}=(\omega_0^{\alpha_s}e^{\xi_s h}\widehat{V}_{s,0}^m+\omega_{N+1}^{\alpha_s}e^{-\xi_s Nh}\widehat{V}_{s,N+1}^m,\ldots,\omega_2^{\alpha_s}e^{-\xi_s h}\widehat{V}_{s,N+1})^{T}, \\&\hat{p}_{s,2}=(\omega_2^{\alpha_s}e^{-\lambda_s h}\widehat{V}_{s,0},\ldots,\omega_{N+1}^{\alpha_s}e^{-\lambda_s Nh}\widehat{V}_{s,0}+\omega_0^{\alpha_s} e^{\lambda_s h}\widehat{V}_{s,N+1})^{T}. \end{aligned} \end{equation*}$
4: For

$n=1,2,\ldots,N$ , solve the following system for

${V}_{*,n}^{m+1}={\left({V}_{1,n}^{m+1},{V}_{2,n}^{m+1},\ldots,{V}_{\bar{S},n}^{m+1} \right)}^T$ .

$\begin{eqnarray} \begin{aligned} \left(I_{\bar{S}}-\frac{\tau}{2}Q\right) {V}_{*,n}^{m+1}= \left(1+\frac{\tau}{2}\Delta_x^s\right) \widehat{V}_{*,n}, \;\;\;\;\;\;\;\;(4.7)\end{aligned} \end{eqnarray}$
where

${V}_{*,n}^{m+1}={\left({V}_{*,n}^{m+1},{V}_{*,n}^{m+1},\ldots,{V}_{*,n}^{m+1} \right)}^T$ .
5: end for

| Show Table

DownLoad: CSV

In Algorithm 1, denote $\widetilde{T}_s = I_N-\frac{\tau}{2} T_{s}$ to be the coefficient matrix of the linear system (4.6), which has Toeplitz structure. Using the preconditioned direct method proposed in ^[12,24], the computation process of solving the linear equations with coefficient matrix $\widetilde{T}_s$ can be accelerated by fast Fourier transformations (FFT). The total computation cost to solve Eq (4.6) for each $s = 1, 2, \ldots, \bar{S}$ is $\mathcal{O}(N\log N)$ .

Since the order of the matrix $Q$ represents the number of the regime-switching states, which is far less than $N$ , the linear system (4.7) can be quickly solved.

For more modern ADI approaches, we refer the readers to ^[25,26], which will be our future work to study them under fractional option pricing problems.

5. Numerical experiments

In this section, numerical experiments on the fractional PDE, with known exact solution and European call options under multi-regime FMLS and CGMY models, are presented to show the convergence and efficiency of the proposed ADI approach.

Compared with the ADI method, GMRES and BiCGSTAB are used to solve the linear equation on every temporal layer, where the vector $V^{m-1}$ is taken as a initial guess and the iteration is terminated when the residual $r^{(k)}$ satisfies $\|r^{(k)} \|_2/\|r^{(0)} \|_2\leq 10^{-7}$ . All numerical experiments are carried out by Matlab R2020a.

Example 5.1. Consider the following FPDE problem with source term:

$\begin{equation} \left\{\begin{array}{l} \frac{\partial V_s(x, t)}{\partial t}-\frac{\partial V_s(x, t)}{\partial x}-D_{-}^{\lambda_s, \alpha_s} V_s(x, t)-\sum\limits_{j = 1}^{\bar{S}} q_{s, j} V_j(x, t) = f_s(x, t), \\ V_s(0, t) = 0, \quad 0 < t \leq 1, \\ V_s(1, t) = e^{-t-\lambda_s}, \quad 0 < t \leq 1 ,\\ V_s(x, 0) = e^{-\lambda_s x} x^{2+\alpha_s}, \quad 0 \leq x \leq 1, \end{array}\right. \end{equation}$

(5.1)

with

$f_s(x, t) = -e^{-t-\lambda_s x}\left(\frac{\Gamma\left(3+\alpha_s\right)}{\Gamma(3)} x^2+(1-\lambda_s)x^{2+\alpha_s}+(2+\alpha_s)x^{1+\alpha_s}\right) -\sum\limits_{j = 1}^{\bar{S}} q_{s, j} V_j(x, t),$

where the exact solution is $V_s(x, t) = e^{-t-\lambda_s x} x^{2+\alpha_s}$ .

The following two cases are considered as the settings in ^[12]:

(a) $\bar{S} = 2$ , $\alpha = (1.9, 1.6)$ , $\lambda = (0.92, 1.20)$ , $Q = \left(\begin{array}{cc} -6 & 6\\ 8 & -8 \end{array}\right)$ .

(b) $\bar{S} = 8,$ $\alpha = (1.6, 1.1, 1.9, 1.8, 1.8, 1.3, 1.6, 1.1),$ $\lambda = (2.04, 4.1, 3.6, 4.85, 2.66, 1.63, 0.53, 3.06),$

$Q = \left(\begin{array}{cccccccc} -25&1&10&5&2&2&2&3\\ 4&-38&10&10&2&4&5&3\\ 6&2&-39&4&10&5&5&7\\ 5&2&8&-32&2&10&2&3\\ 7&4&3&7&-38&2&6&9\\ 7&2&5&6&6&-39&3&10\\ 3&5&6&7&9&7&-45&8\\ 5&4&10&7&7&4&6&-43 \end{array}\right).$

In , , the surfaces of the numerical solution and error $|V^M-V_\ast^M|$ of case (a) are presented respectively, for $s = 1, 2$ , when $M = N = 1024$ .

Figure 1. The numerical solution and error of case (a) in Example 5.1 when

$s = 1$ .

DownLoad: Full-Size Img PowerPoint

Figure 2. The numerical solution and error of case (a) in Example 5.1 when

$s = 2$ .

DownLoad: Full-Size Img PowerPoint

Define the convergence order of the numerical scheme as

$\begin{eqnarray*} \text{Order}_m = \log_ {2}\frac{\|V^{m-1}-V_\ast^{m-1}\|}{\| V^{m}-{V_\ast}^m\|}, \end{eqnarray*}$

where $V_\ast^m$ is the exact solution on $t_m$ and $\|\cdot \|$ -norm is defined in Theorem 3.2.

In Tables 1, 2, the error and convergence order of Crank-Nicolson scheme and ADI scheme are listed for case (a) and case (b), respectively. We use "D-ADI" to represent the ADI Algorithm 1 with preconditioned direct method.

Table 1. Error and convergence order of three numerical schemes for case (a) in Example 5.1.

		GMRES		BiCGSTAB		D-ADI
Regime	$N=M$	$\\| V^{m}-{V_\ast}^m\\|$	Order	$\\| V^{m}-{V_\ast}^m\\|$	Order	$\\| V^{m}-{V_\ast}^m\\|$	Order
	$2^{4}$	2.3967E-04	—	2.3967E-04	—	1.9205E-04	—
	$2^{5}$	6.3382E-05	1.9189	6.3380E-05	1.9189	5.1484E-05	1.8993
	$2^{6}$	1.6324E-05	1.9571	1.6322E-05	1.9572	1.3344E-05	1.9479
$s=1$	$2^{7}$	4.1466E-06	1.9770	4.1434E-06	1.9779	3.3981E-06	1.9735
	$2^{8}$	1.0494E-06	1.9824	1.0450E-06	1.9873	8.5744E-07	1.9866
	$2^{9}$	2.6886E-07	1.9646	2.6360E-07	1.9871	2.1536E-07	1.9933
	$2^{10}$	7.4745E-08	1.8468	6.7470E-08	1.9660	5.3965E-08	1.9967
	$2^{4}$	2.6564E-04	—	2.6563E-04	—	3.2512E-04	—
	$2^{5}$	6.9890E-05	1.9263	6.9886E-05	1.9264	8.4830E-05	1.9383
	$2^{6}$	1.7969E-05	1.9596	1.7965E-05	1.9598	2.1702E-05	1.9667
$s=2$	$2^{7}$	4.5625E-06	1.9776	4.5583E-06	1.9786	5.4916E-06	1.9826
	$2^{8}$	1.1551E-06	1.9817	1.1500E-06	1.9869	1.3814E-06	1.9910
	$2^{9}$	2.9661E-07	1.9614	2.9041E-07	1.9854	3.4645E-07	1.9955
	$2^{10}$	8.3315E-08	1.8319	7.4617E-08	1.9605	8.6747E-08	1.9977

| Show Table

DownLoad: CSV

Table 2. Error and convergence order of three numerical schemes for case (b) in Example 5.1.

		GMRES		BiCGSTAB		D-ADI
Regime	$N=M$	$\\| V^{m}-{V_\ast}^m\\|$	Order	$\\| V^{m}-{V_\ast}^m\\|$	Order	$\\| V^{m}-{V_\ast}^m\\|$	Order
	$2^{7}$	2.3244E-06	—	2.3236E-06	—	2.7531E-05	—
$s=1$	$2^{8}$	5.8678E-07	1.9860	5.8543E-07	1.9888	6.8850E-06	1.9995
	$2^{9}$	1.4837E-07	1.9836	1.4697E-07	1.9939	1.7221E-06	1.9993
	$2^{7}$	2.6927E-06	—	2.6917E-06	—	3.7611E-05	—
$s=2$	$2^{8}$	6.7564E-07	1.9947	6.7401E-07	1.9977	9.3779E-06	2.0038
	$2^{9}$	1.7016E-07	1.9893	1.6848E-07	2.0002	2.3407E-06	2.0023
	$2^{7}$	2.0421E-06	—	2.0412E-06	—	3.2462E-05	—
$s=3$	$2^{8}$	5.1579E-07	1.9852	5.1445E-07	1.9883	8.1187E-06	1.9994
	$2^{9}$	1.3055E-07	1.9822	1.2917E-07	1.9937	2.0306E-06	1.9993
	$2^{7}$	2.2554E-06	—	2.2544E-06	—	4.1254E-05	—
$s=4$	$2^{8}$	5.6974E-07	1.9850	5.6817E-07	1.9883	1.0316E-05	1.9996
	$2^{9}$	1.4428E-07	1.9815	1.4266E-07	1.9937	2.5801E-06	1.9994
	$2^{7}$	2.1676E-06	—	2.1668E-06	—	2.8966E-05	—
$s=5$	$2^{8}$	5.4738E-07	1.9855	5.4605E-07	1.9885	7.2445E-06	1.9994
	$2^{9}$	1.3848E-07	1.9829	1.3710E-07	1.9938	1.8120E-06	1.9993
	$2^{7}$	2.6192E-06	—	2.6183E-06	—	3.0412E-05	—
$s=6$	$2^{8}$	6.5989E-07	1.9888	6.5841E-07	1.9916	7.5967E-06	2.0012
	$2^{9}$	1.6671E-07	1.9849	1.6517E-07	1.9951	1.8992E-06	2.0000
	$2^{7}$	2.7223E-06	—	2.7214E-06	—	2.5136E-05	—
$s=7$	$2^{8}$	6.8708E-07	1.9862	6.8567E-07	1.9888	6.2867E-06	1.9994
	$2^{9}$	1.7361E-07	1.9846	1.7213E-07	1.9940	1.5725E-06	1.9992
	$2^{7}$	2.6352E-06	—	2.6343E-06	—	3.4432E-05	—
$s=8$	$2^{8}$	6.6171E-07	1.9937	6.6016E-07	1.9965	8.5878E-06	2.0034
	$2^{9}$	1.6671E-07	1.9889	1.6510E-07	1.9995	2.1439E-06	2.0021

| Show Table

DownLoad: CSV

From Tables 1, 2, it is seen that both the Crank-Nicolson and ADI schemes have second-order convergence. It is also observed that the convergence of the ADI scheme is more stable. Since the order from Tables 1, 2 represents the convergence on each regime, it can further verify the theoretical analysis in Theorem 3.2.

In , , the average of the iteration number (denoted by "IT") and the total CPU time (in seconds, denoted by "CPU") of GMRES, BiCGSTAB and ADI methods are compared when the number of discrete points $N$ increases from $2^4$ to $2^9$ respectively.

Table 3. Iteration numbers and CPU time of different solvers for case (a) in Example 5.1.

	GMRES		BiCGSTAB		D-ADI
$N=M$	IT	CPU	IT	CPU	CPU
$2^{4}$	47	0.0345	22.94	0.0059	0.0242
$2^{5}$	77	0.0539	40.47	0.0063	0.0368
$2^{6}$	119	0.1674	62.45	0.0321	0.1222
$2^{7}$	182	0.7743	84.90	0.2811	0.4580
$2^{8}$	307	5.2065	112.75	2.4813	2.0023
$2^{9}$	520	56.9593	156.25	28.1146	10.3205
$2^{10}$	886	922.9244	218.91	390.7879	57.3228

| Show Table

DownLoad: CSV

Table 4. Iteration numbers and CPU time of different solvers for case (b) in Example 5.1.

	GMRES		BiCGSTAB		D-ADI
$N=M$	IT	CPU	IT	CPU	CPU
$2^{4}$	57	0.0410	27.63	0.0079	0.0329
$2^{5}$	91	0.0820	45.16	0.0173	0.0792
$2^{6}$	145	0.3451	72.38	0.1356	0.3364
$2^{7}$	218	2.1381	107.41	1.3419	1.2518
$2^{8}$	340	22.8717	151.21	15.3208	5.8594
$2^{9}$	591	345.5954	232.94	237.7708	32.4836

| Show Table

DownLoad: CSV

From Tables 3, 4, it is observed that both GMRES and BiCGSTAB require more iteration step than ADI method, and so does the CPU time. By comparing the CPU time of the three methods, it is obvious that the preconditioned direct ADI method is fast, and can significantly reduce the computation time.

Then, the proposed preconditioned direct ADI method is applied to deal with the multi-regime European option pricing model in Example 5.3. The parameters in this example change in different regimes, by which the sudden state movement and the non-stationary behavior of the market is described.

In order to verify the convergence order for non-smooth payoff function as initial conditions in Eq (2.4), we demonstrate the results of an option pricing problem in Example 5.2.

Example 5.2. Consider the multi-regime FMLS model for pricing European call option, where $x_l = \ln(0.1), x_r = \ln(100), K = 50, r = 0.05, T = 1$ . The regime-switching parameters are set as

$\bar{S} = 2,\quad \alpha = (1.9, 1.6),\quad \sigma = (0.25, 0.5),\quad Q = \left(\begin{array}{cc} -6&6\\ 8&-8 \end{array}\right).$

In , we list the error and convergence order of ADI scheme. The viscosity solution is approximated by the numerical solution using a dense mesh with $N = M = 2^{13}$ .

Table 5. Error and convergence order of the ADI scheme in Example 5.2.

Regime	$N=M$	$\\| V^{m}-{V_\ast}^m\\|$	Order	$\\| V^{m}-{V_\ast}^m\\|_\infty$	Order
	$2^{7}$	1.1242E-02	—	1.3525E-02	—
	$2^{8}$	2.1656E-03	2.3760	2.4681E-03	2.4542
$s=1$	$2^{9}$	5.2820E-04	2.0356	6.6201E-04	1.8985
	$2^{10}$	1.4432E-04	1.8719	1.5758E-04	2.0708
	$2^{11}$	3.0417E-05	2.2463	4.2304E-05	1.8972
	$2^{7}$	1.1462E-02	—	1.3801E-02	—
	$2^{8}$	2.2038E-03	2.3788	2.4871E-03	2.4722
$s=2$	$2^{9}$	5.3432E-04	2.0442	6.5044E-04	1.9350
	$2^{10}$	1.4726E-04	1.8593	1.7085E-04	1.9287
	$2^{11}$	3.0424E-05	2.2751	4.1473E-05	2.0425

| Show Table

DownLoad: CSV

From , it is observed that the ADI scheme can keep the second-order convergence under the non-smooth initial conditions. The convergence orders are not as steady as those in Example 5.1 perhaps because of the discontinuity at the strike price $K$ , which can be improved by the Padé schemes proposed in ^[27].

Example 5.3. Consider the multi-regime CGMY model for pricing European call option where $x_l = ln(0.1), x_r = \ln(200), K = 60, r = 0.05, T = 1, C = 0.1$ .

Consider the following two cases:

(a) $\bar{S} = 4, \quad \alpha = (1.5, 1.7, 1.3, 1.8), \quad \sigma = (0.25, 0.5, 0.75, 0.5), \quad \xi = (2, 1, 5, 1), \quad\lambda = (1, 3, 2, 4),$

$Q = \left(\begin{array}{cccc} -12&2&4&6\\ 8&-20&10&2\\ 5&4&-10&1\\ 2&4&8&-14 \end{array}\right).$

(b) $\bar{S} = 6$ , $\alpha = (1.5, 1.2, 1.7, 1.3, 1.6, 1.8)$ , $\sigma = (0.25, 0.5, 0.3, 0.75, 0.5, 0.2)$ , $\xi = (1, 2, 1, 2, 3, 3)$ , $\lambda = (3, 4, 3, 1, 2, 2)$ ,

$Q = \left(\begin{array}{cccccccc} -13&2&4&1&2&4\\ 3&-20&8&3&4&2\\ 2&1&-10&2&4&1\\ 2&4&2&-16&1&7\\ 1&1&3&1&-7&1\\ 4&4&1&1&2&-12 \end{array}\right).$

The option values of four regimes and six regimes are depicted, respectively, in when $M = N = 512$ , where the blue dashed line represents the payoff of the European call option and the other colored lines represent the option prices with different regimes at the value date.

Figure 3. The value of European call option under the multi-regime CGMY model and payoff function in Example 5.3.

DownLoad: Full-Size Img PowerPoint

6. Conclusions

In this paper, a second-order finite difference method is proposed to discretise a class of fractional regime-switching option pricing models. In addition, the sufficient conditions of stability and convergence of the numerical scheme are studied in detail. The ADI scheme with preconditioned direct method is considered to deal with the multi-regime structure to accelerate the computation. Numerical experiments verify the theoretical convergence order and show the efficacy of the proposed method.

Acknowledgments

The authors are very grateful to the referees for their constructive comments and valuable suggestions, which greatly improved the original manuscript of the paper. This work is supported by the National Natural Science Foundation of China (No. 11971354 and No. 12171366).

Conflict of interest

The authors declare there is no conflicts of interest.

References

[1]	Kaza S, Yao L, Bhada-Tata P, Van Woerden F. What a Waste 2.0: A Global Snapshot of Solid Waste Management to 2050. Washington, DC: The World Bank; 2018. Available from: https://datatopics.worldbank.org/what-a-waste/trends_in_solid_waste_management.html https://doi.org/10.1596/978-1-4648-1329-0
[2]	Jayasinghe R, Mushtaq U, Smythe TA, Baillie C, (2013) The Garbage Crisis: A Global Challenge for Engineers. San Rafael: Morgan & Claypool Publishers; 2013[cited 2021 Feb 17]. 3-5. https://www-morganclaypool-com.ezproxy.lib.ucalgary.ca/doi/pdfplus/10.2200/S00453ED1V01Y201301ETS018
[3]	Ajibade LT (2007) Indigenous Knowledge System of waste management in Nigeria. IJTK 6: 642-647.
[4]	Sridhar MKC, Hammed T (2014) Turning Waste to Wealth in Nigeria: An Overview. J Hum Ecol 46: 195–203. https://doi.org/10.1080/09709274.2014.11906720 doi: 10.1080/09709274.2014.11906720
[5]	Lohri CR, Diener S, Zabaleta I, Mertenat A, Zurbrügg C (2017) Treatment technologies for urban solid biowaste to create value products: a review with focus on low- and middle-income settings. Rev Environ Sci Biotechnol. 16: 81–130. https://doi.org/10.1007/s11157-017-9422-5 doi: 10.1007/s11157-017-9422-5
[6]	Asomani-Boateng R (2007) Closing the Loop: Community-Based Organic Solid Waste Recycling, Urban Gardening, and Land Use Planning in Ghana, West Africa. J Plan Educ Res 27: 132–45. https://doi.org/10.1177/0739456X07306392 doi: 10.1177/0739456X07306392
[7]	Asomani-Boateng R (2016) Local Networks: Commodity Queens and the Management of Organic Solid Waste in Indigenous Open-Air Markets in Accra, Ghana. J Plan Educ Res 36: 182–94. https://doi.org/10.1177/0739456X15604445 doi: 10.1177/0739456X15604445
[8]	Dortmans B, Diener S, Verstappen B, Zurbrügg C.EAWAG in Switzerland: Black Soldier Fly Biowaste Processing. A Step-by-Step Guide. EAWAG – Swiss Federal Institute of Aquatic Science and Technology, 2017. Available from: https://www.eawag.ch/fileadmin/Domain1/Abteilungen/sandec/publikationen/SWM/BSF/BSF_Biowaste_Processing_HR.pdf
[9]	Edmond C. World Economic Forum in USA: This is what the world's waste does to people in poorer countries. World Economic Forum, 2019. Available from: https://www.weforum.org/agenda/2019/05/this-is-what-the-world-s-waste-does-to-people-in-poorer-countries/
[10]	Gahana GC, Patil YB, Shibin KT, et al. (2018) Conceptual frameworks for the drivers and barriers of integrated sustainable solid waste management: A TISM approach. Manag Environ Qual 29: 516–46. http://dx.doi.org.ezproxy.lib.ucalgary.ca/10.1108/MEQ-10-2017-0117 doi: 10.1108/MEQ-10-2017-0117
[11]	Oteng-Ababio M, Melara Arguello JE, Gabbay O (2013) Solid waste management in African cities: Sorting the facts from the fads in Accra, Ghana. Habitat Int 39: 96–104. https://doi.org/10.1016/j.habitatint.2012.10.010 doi: 10.1016/j.habitatint.2012.10.010
[12]	Siragusa L, Arzyutov D (2020) Nothing goes to waste: sustainable practices of re-use among Indigenous groups in the Russian North. Curr Opin Environ Sust 43 :41–8. https://doi.org/10.1016/j.cosust.2020.02.001 doi: 10.1016/j.cosust.2020.02.001
[13]	Cappellozza S, Leonardi MG, Savoldelli S, et al. (2019) A First Attempt to Produce Proteins from Insects by Means of a Circular Economy. Animals 9:278. https://doi.org/10.3390/ani9050278 doi: 10.3390/ani9050278
[14]	Shumo M, Osuga IM, Khamis FM, et al. (2019) The nutritive value of black soldier fly larvae reared on common organic waste streams in Kenya. Sci Rep 9: 10110. https://doi.org/10.1038/s41598-019-46603-z doi: 10.1038/s41598-019-46603-z
[15]	da Silva GDP, Hesselberg T (2020) A Review of the Use of Black Soldier Fly Larvae, Hermetia illucens (Diptera: Stratiomyidae), to Compost Organic Waste in Tropical Regions. Neotrop Entomol. 49: 151–62. https://doi.org/10.1007/s13744-019-00719-z doi: 10.1007/s13744-019-00719-z
[16]	Singh A, Kumari K (2019) An inclusive approach for organic waste treatment and valorisation using Black Soldier Fly larvae: A review. J Environ Manag 251: 109569. https://doi.org/10.1016/j.jenvman.2019.109569 doi: 10.1016/j.jenvman.2019.109569
[17]	Nyakeri EM, Ogola HJ, Ayieko MA (2017) An Open System for Farming Black Soldier Fly Larvae as a Source of Proteins for Smallscale Poultry and Fish Production. J Insects Food Feed 3: 51–56. https://doi.org/10.3920/JIFF2016.0030 doi: 10.3920/JIFF2016.0030
[18]	Caplin R. Smith School of Enterprise and the Environment in University of Oxford; Business Models in Sustainability Transitions: A Study of Organic Waste Management and Black Soldier Fly Production in Kenya. Smith School of Enterprise and the Environment, Oxford, United Kingdom: Available from: https://www.linkedin.com/posts/ryancaplin_business-models-for-organic-waste-and-bsf-activity-6988508871459766272--5KL/
[19]	Ojha S, Bußler S, Schlüter OK (2020) Food waste valorisation and circular economy concepts in insect production and processing. Waste Manage 118: 600–9. https://doi.org/10.1016/j.wasman.2020.09.010 doi: 10.1016/j.wasman.2020.09.010
[20]	Gold M, Ireri D, Zurbrügg C, et al. (2021) Efficient and safe substrates for black soldier fly biowaste treatment along circular economy principles. Detritus 16: 31–40. https://doi.org/10.31025/2611-4135/2021.15116 doi: 10.31025/2611-4135/2021.15116
[21]	Paes LAB, Bezerra BS, Deus RM, et al. (2019) Organic solid waste management in a circular economy perspective – A systematic review and SWOT analysis. J Clean Prod 239: 118086. https://doi.org/10.1016/j.jclepro.2019.118086 doi: 10.1016/j.jclepro.2019.118086
[22]	Wang YS, Shelomi M (2017) Review of Black Soldier Fly (Hermetia illucens) as Animal Feed and Human Food. Foods 6:91. https://doi.org/10.3390/foods6100091 doi: 10.3390/foods6100091
[23]	Diener S, Zurbrügg C, Tockner K (2009) Conversion of organic material by black soldier fly larvae: establishing optimal feeding rates. Waste Manag Res. 27: 603–10. https://doi.org/10.1177/0734242X09103838 doi: 10.1177/0734242X09103838
[24]	Surendra KC, Tomberlin JK, van Huis A, et al. (2020) Rethinking organic wastes bioconversion: Evaluating the potential of the black soldier fly (Hermetia illucens (L.)) (Diptera: Stratiomyidae) (BSF). Waste Manage 117: 58–80. https://doi.org/10.1016/j.wasman.2020.07.050 doi: 10.1016/j.wasman.2020.07.050
[25]	Diener S, Gutiérrez FR, Zurbrügg C, et al.Are Larvae of the Black Soldier Fly– Hermatia Illucens – A Financially Viable Option For Organic Waste Management in Costa Rica? In: Proceedings Sardinia 2009, Twelfth International Waste Management and Landfill Symposium. S. Margherita di Pula, Cagliari, Italy: CISA Publisher, Italy; 2009.
[26]	Diener S, Zurbrügg C, Gutiérrez FR, et al. WasteSafe in Khulna, Bangladesh: Black Soldier Fly Larvae For Organic Waste Treatment – Prospects And Constraints.; WasteSafe Khulna, Bangladesh, 2011 Available from: https://www.eawag.ch/fileadmin/Domain1/Abteilungen/sandec/publikationen/SWM/BSF/Black_soldier_fly_larvae_for_organic_waste_treatment.pdf
[27]	Myers H, Tomberlin J, Lambert B, et al. (2008) Development of Black Soldier Fly (Diptera: Stratiomyidae) Larvae Fed Dairy Manure. Environ Entomol 37: 11–5. https://doi.org/10.1603/0046-225X(2008)37[11:DOBSFD]2.0.CO; 2 doi: 10.1093/ee/37.1.11
[28]	Isibika A, Simha P, Vinnerås B, et al. (2023) Food industry waste - An opportunity for black soldier fly larvae protein production in Tanzania. Sci Total Environ 858: 9. https://doi.org/10.1016/j.scitotenv.2022.159985 doi: 10.1016/j.scitotenv.2022.159985
[29]	Fuss M, Vasconcelos Barros RT, et al. (2018) Designing a framework for municipal solid waste management towards sustainability in emerging economy countries - An application to a case study in Belo Horizonte (Brazil). J Clean Prod 178: 655–64. https://doi.org/10.1016/j.jclepro.2018.01.051 doi: 10.1016/j.jclepro.2018.01.051
[30]	Yhdego M (1995) Urban solid waste management in Tanzania issues, concepts and challenges. Resources, Conserv Recy 14: 1–10. https://doi.org/10.1016/0921-3449(94)00017-Y doi: 10.1016/0921-3449(94)00017-Y
[31]	Ferronato N, Portillo MAG, Lizarazu GEG, et al. (2020) Formal and informal waste selective collection in developing megacities: Analysis of residents' involvement in Bolivia. Waste Manag Res. 39: 108–21. https://doi.org/10.1177/0734242X20936765 doi: 10.1177/0734242X20936765
[32]	Aleluia J, Ferrão P (2016) Characterization of urban waste management practices in developing Asian countries: A new analytical framework based on waste characteristics and urban dimension. Waste Manage 58: 415–29. https://doi.org/10.1016/j.wasman.2016.05.008 doi: 10.1016/j.wasman.2016.05.008
[33]	Ddiba D, Andersson K, Rosemarin A, et al. (2022) The circular economy potential of urban organic waste streams in low- and middle-income countries. Environ Dev Sustain. 24: 1116–44. https://doi.org/10.1007/s10668-021-01487-w doi: 10.1007/s10668-021-01487-w
[34]	Velenturf APM, Purnell P (2021) Principles for a sustainable circular economy. Sustain Prod Consump 27: 1437–57. https://doi.org/10.1016/j.spc.2021.02.018 doi: 10.1016/j.spc.2021.02.018
[35]	Suárez-Eiroa B, Fernández E, Méndez-Martínez G, et al. (2019) Operational principles of circular economy for sustainable development: Linking theory and practice. J Clean Prod 214: 952–61. https://doi.org/10.1016/j.jclepro.2018.12.271 doi: 10.1016/j.jclepro.2018.12.271
[36]	Wright CY, Godfrey L, Armiento G, et al. (2019) Circular economy and environmental health in low- and middle-income countries. Globalization Health 15: 65. https://doi.org/10.1186/s12992-019-0501-y doi: 10.1186/s12992-019-0501-y
[37]	Schroeder P, Dewick P, Kusi-Sarpong S, et al. (2018) Circular economy and power relations in global value chains: Tensions and trade-offs for lower income countries. Res Conserv Recy 136: 77–8. https://doi.org/10.1016/j.resconrec.2018.04.003 doi: 10.1016/j.resconrec.2018.04.003
[38]	Bringsken BM, Loureiro I, Ribeiro C, et al. (2018) Community Involvement Towards a Circular Economy: a Sociocultural Assessment of Projects and Interventions to Reduce Undifferentiated Waste. Eur J Sustain Dev 7: 496–506. https://doi.org/10.14207/ejsd.2018.v7n4p496 doi: 10.14207/ejsd.2018.v7n4p496
[39]	Padilla-Rivera A, do Carmo BBT, Arcese G, et al. (2021) Social circular economy indicators: Selection through fuzzy delphi method. Sustain Prod Consump 26: 101–110. https://doi.org/10.1016/j.spc.2020.09.015 doi: 10.1016/j.spc.2020.09.015
[40]	Chineme A, Assefa G, Herremans IM, et al. (2022) African Indigenous Female Entrepreneurs (IFÉs): A Closed-Looped Social Circular Economy Waste Management Model. Sustainability 14: 11628. https://doi.org/10.3390/su141811628 doi: 10.3390/su141811628
[41]	Ibáñez-Forés V, Bovea MD, Coutinho-Nóbrega C, et al. (2019) Assessing the social performance of municipal solid waste management systems in developing countries: Proposal of indicators and a case study. Ecol Indic 98: 164–178. https://doi.org/10.1016/j.ecolind.2018.10.031 doi: 10.1016/j.ecolind.2018.10.031
[42]	Verschuere B, Brandsen T, Pestoff V (2012) Co-production: The State of the Art in Research and the Future Agenda. VOLUNTAS: Int J Voluntary Nonprofit Org 23: 1083–1101. https://doi.org/10.1007/s11266-012-9307-8 doi: 10.1007/s11266-012-9307-8
[43]	Manuel-Navarrete D, Buzinde C, Swanson T (2021) Fostering horizontal knowledge co-production with Indigenous people by leveraging researchers' transdisciplinary intentions. Ecol Soc 26:22. https://doi.org/10.5751/ES-12265-260222 doi: 10.5751/ES-12265-260222
[44]	Medina M, WIDER in Helsinki, Finland: Solid wastes, Poverty and the Environment in Developing Country Cities: Challenges and Opportunities. WIDER Helsinki, Finland, 2010. Available from: http://hdl.handle.net/10419/54107
[45]	Pestoff V (2014) Collective Action and the Sustainability of Co-Production. Public Manag Rev 16: 383–401. https://doi.org/10.1080/14719037.2013.841460 doi: 10.1080/14719037.2013.841460
[46]	Amref Health Africa Tanzania. A Baseline Survey To Establish A Reference For A Functional Social Enterprise Based On The Current Practices, Demand And Attitudes Towards Solid Waste Management At Household Level In Ilala Municipal Council. 2018. Unpublished report by MK Center
[47]	Halog A, Anieke S (2021) A Review of Circular Economy Studies in Developed Countries and Its Potential Adoption in Developing Countries. Circ Econ Sust 1 :209–230. https://doi.org/10.1007/s43615-021-00017-0 doi: 10.1007/s43615-021-00017-0
[48]	Korhonen J, Honkasalo A, Seppälä J (2018) Circular Economy: The Concept and its Limitations. Ecol Econ 143: 37–46. https://doi.org/10.1016/j.ecolecon.2017.06.041 doi: 10.1016/j.ecolecon.2017.06.041
[49]	Ellen MacArthur Foundation. Towards the Circular Economy Vol. 1: An Economic and Business Rationale for an Accelerated Transition. Ellen MacArthur Foundation; 2013. Available from: https://emf.thirdlight.com/link/x8ay372a3r11-k6775n/@/preview/1?o
[50]	Robinson S. Social Circular Economy: Opportunities for People Planet and Profit. Social Circular Economy, 2017. Available from: http://www.socialcirculareconomy.com/news
[51]	Gharfalkar M, Court R, Campbell C, et al. (2015) Analysis of waste hierarchy in the European waste directive 2008/98/EC. Waste Manage 39: 305–313. https://doi.org/10.1016/j.wasman.2015.02.007 doi: 10.1016/j.wasman.2015.02.007
[52]	Oh J, Hettiarachchi H (2020) Collective Action in Waste Management: A Comparative Study of Recycling and Recovery Initiatives from Brazil, Indonesia, and Nigeria Using the Institutional Analysis and Development Framework. Recycling 5:4. https://doi.org/10.3390/recycling5010004 doi: 10.3390/recycling5010004
[53]	World Health Organization, United Nations City in Denmark Circular economy and health: opportunities and risks. World Health Organization. Regional Office for Europe, 2018. Available from: https://apps.who.int/iris/handle/10665/342218
[54]	Guo H, Jiang C, Zhang Z, et al. (2021) Material flow analysis and life cycle assessment of food waste bioconversion by black soldier fly larvae (Hermetia illucens L.). Sci Total Environ 750: 141656. https://doi.org/10.1016/j.scitotenv.2020.141656 doi: 10.1016/j.scitotenv.2020.141656
[55]	Salomone R, Saija G, Mondello G, et al. (2017) Environmental impact of food waste bioconversion by insects: Application of Life Cycle Assessment to process using Hermetia illucens. J Clean Prod 140: 890–905. https://doi.org/10.1016/j.jclepro.2016.06.154 doi: 10.1016/j.jclepro.2016.06.154
[56]	Sankara F, Sankara F, Pousga S, et al. (2023) Optimization of Production Methods for Black Soldier Fly Larvae (Hermetia illucens L.) in Burkina Faso. Insects 14: 776. https://doi.org/10.3390/insects14090776 doi: 10.3390/insects14090776
[57]	Chineme A, Shumo M, Assefa G, et al. (2023) Simple is Better When Appropriate: An Innovative Accessible Approach to Biowaste Treatment Using Wild Black Soldier Fly, In: Solid Waste Management - Recent Advances, New Trends and Applications[Working Title]: IntechOpen, . http://dx.doi.org/10.5772/intechopen.1002449.
[58]	Bowan PA, Ziblim SD. (2020) Households' Knowledge, Attitudes, and Practices towards Municipal Solid Waste Disposal. J Studies Soc Sci 19.
[59]	Hammed T, Sridhar M, Olaseha I (2011) Effect of Demographic Characteristics and Perceptions of Community Residents on Solid Waste Management Practices in Orita-Aperin, Ibadan, Nigeria. J Environ Syst 33: 187–199.
[60]	Pfiester M, Kaufman PE. University of Florida in USA: Drone fly, rat-tailed maggot - Eristalis tenax (Linnaeus). University of Florida, s 2009. Available from: https://entnemdept.ufl.edu/creatures/livestock/rat-tailed_maggot.htm
[61]	Ibrahim IA, Gad AM (1975) The Occurrence of Paedogenesis in Eristalis Larvae (Diptera: Syrphidae). J Med Entomol 12: 268. https://doi.org/10.1093/jmedent/12.2.268 doi: 10.1093/jmedent/12.2.268
[62]	Alvarez L. The Role of Black Soldier Fly, Hermetia illucens (L.) (Diptera: Stratiomyidae) in Sustainable Waste Management in Northern Climates[Ph.D thesis].[Windsor]: University of Windsor; 2012. 171 p.
[63]	Liu X, Chen X, Wang H, et al. (2017) Dynamic changes of nutrient composition throughout the entire life cycle of black soldier fly. PLOS ONE 2017 12: e0182601. https://doi.org/10.1371/journal.pone.0182601 doi: 10.1371/journal.pone.0182601

This article has been cited by:

1.	Sai Zhang, Li Tang, Yan-Jun Liu, Formation deployment control of multi-agent systems modeled with PDE, 2022, 19, 1551-0018, 13541, 10.3934/mbe.2022632
2.	Jinxin Du, Lei Liu, Adaptive fuzzy fixed time formation control of state constrained nonlinear multi-agent systems against FDI attacks, 2024, 21, 1551-0018, 4724, 10.3934/mbe.2024207
3.	Xiaofeng Chen, Double-integrator consensus for a switching network without dwell time, 2023, 20, 1551-0018, 11627, 10.3934/mbe.2023516
4.	Yuhang Yao, Jiaxin Yuan, Tao Chen, Xiaole Yang, Hui Yang, Distributed convex optimization of bipartite containment control for high-order nonlinear uncertain multi-agent systems with state constraints, 2023, 20, 1551-0018, 17296, 10.3934/mbe.2023770
5.	Xinling Li, Xiaofeng Ma, Weipeng Tai, Jianping Zhou, Designing an event-triggered H∞ filter with possibly inconsistent modes for Markov jump systems, 2023, 139, 10512004, 104092, 10.1016/j.dsp.2023.104092
6.	Chunsong Han, Kang Wang, Tengyu Ma, Guangtao Ran, Juntao Pan, Adaptive Event‐Triggered Sliding‐Mode Fault‐Tolerant Consensus Control for Nonlinear Multi‐Agent Systems With Unknown Perturbations, 2025, 1049-8923, 10.1002/rnc.7968

Reader Comments

Your name:*

Email:*
© 2023 the Author(s), licensee AIMS Press. This is an open access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/4.0)