Nonlinear Engineering 2018; 7(3): 207–227
Shikaa Samuel* and Vinod Gill Diffusion-Chemotaxis Model of Effects of Cortisol on Immune Response to Human Immunodeficiency virus
https://doi.org/10.1515/nleng-2017-0018 favour the persistence of virus. Also, HIV attack is mainly Received July 9, 2017; revised November 16, 2017; accepted Decem- on the command centre of the immune system called CD4+ ber 20, 2017. T-cells [3]. Though, it can also infect other immune cells Abstract: In quest to contain and subsequently eradica- with low level CD4 expression such as macrophages and tion Human Immunodeficiency virus (HIV) in the society, dendritic cells [4]. mathematical modelling remains an important research There are convincing evidences that, chronic or persis- tool. In this paper, we formulated a mathematical model tence stress impact negatively on the immune system. to study the effects of cortisol on immune response toHIV The stimulation of sympathetic fibres releases hormones capturing the roles played by dendritic cells, T helper cells, that bind to receptors on white blood cells [5–7]. Par- regulatory T cells and cytotoxic T cells in the virus repli- ticularly, the hypothalamic–pituitary–adrenal (HPA), cation dynamics. The primary source of concentration of the sympathetic–adrenal–medullary (SAM), and the cortisol in this work is through psychological stress. Nu- hypothalamic–pituitary–ovarian (HPO) axes secrete the merical experiments are performed to examine the effect adrenal hormones such as epinephrine, norepinephrine, of cortisol on selective inhibition of antigen presentation and cortisol which regulate the immune cells distribution activities and up-regulation of naive cytotoxic T cells acti- and function [8]. vation in the case of acute and persistent stressful condi- Cortisol inhibits the production of interleukin (IL)-12, inter- tions. feron (IFN)-gamma, IFN-alpha, and tumor-necrosis-factor (TNF)-alpha by antigen-presenting cells (APCs) and T Keywords: chemotaxis system, radial basis functions, cor- helper (Th)1 cells, but upregulates IL-4, IL-10, and IL-13 se- tisol creted by Th2 cells which results in a shift toward a Th2 im- MSC: 93A30, 92B08, 65P10, 65P35 mune response [9]. It also prevents proliferation of T-cells by rendering the interleukin-2 producer T-cells unrespon- sive to IL-1, and unable to produce the T-cell growth factor (IL-2) [10]. 1 Introduction Several mathematical models have been formulated to study the HIV growth dynamics [11–16], drug therapy [17– In last three and a half decades, there is all fronts battle 20] and the rate of generation of HIV variants that escape against human immunodeficiency virus (HIV) disease due immune responses [21–25]. Presently, just handful models to its devastating consequences on the infected individual of immune responses to HIV are found in literature [26, 27]. and the society. Researchers have been working hard to These models capture only the time evolution while little unravel the mysteries behind the virus pathogenesis, im- or no attention is given to spatial distribution of interact- mune response and other possible parameters that may ing cells and proteins. In this paper, we proposed a mathe- influence the progress of the disease but it is still notcom- matical model of immune response to HIV under the influ- pletely understood [1, 2]. For example, chronic immune ence of cortisol which captures the time evolution of the vi- activation and viral latency are among major factors that ral load, as in previous models, the potential effect of cell- to-cell transmission of HIV [28–33]. The rest of the paper is organized thus: model formulation is done in Section 2 while the analysis of the model is presented in Section 3. *Corresponding Author: Shikaa Samuel, Department of mathe- The numerical scheme is explained in Section 4 alongside matics, Amity University Rajasthan, Jaipur-302030, India, E-mail: with the results while Section 5 concludes the paper with [email protected] important remarks. Vinod Gill, Department of mathematics, Amity University Rajasthan, Jaipur-302030, India 208 | S. Samuel and V. Gill, Diffusion-Chemotaxis Model of Effects of Cortisol
2 The Mathematical Model the production of chemokines responsible for antigen pre- sentation process but upregulates the cytokines respon- sible for activation of naive cytotoxic T cells [41]. Exper- In order to examine the effects of cortisol on immune imental results show cortisol suppression of interleukin- response to HIV, the time evolution of viral load, den- 12 (IL-12) synthesis and an increase in IL-10 production dritic cells, T helper cells, regulatory T cells, cytotoxic T [42]. This model captures the two roles played by cortisol cells, intercellular messengers and concentration of cor- amongst several others. In equation (4) below, c denotes tisol are modelled as a system of nonlinear partial differ- 4 concentration of cortisol with diffusion, secretion, decay ential equations, over a square domain Ω. Here, dendritic and source terms. cells play two roles; viral particles carried by exposed den- ∂c dritic cells are taken to CD4 T-cells for antigen presenta- 4 = D Δc + a c − σ c + g(x, t), (4) ∂t 4 4 4 4 4 4 tion process and also transmit the virus to CD4 T-cells [34– ∂c c (x, y,0) = C ,inΩ, 4 =0on∂Ω. 37]. The symbols representing the state variables are de- 4 40 ∂n scribed thus: c1 : Concentration chemokine secreted by Next, the population of dendritic cells is divided into dendritic cells; c2 : Concentration of cytokine secreted by three subpopulation. Equation (5) models rate of change T helper cells signalling attack on virus infected immune of the density of the immature dendritic cells which have cells;c3 : Concentration of cytokine secreted by T helper a source term ρ1 while b1p1v describes a portion of den- cells signalling self-regulation; c4 : Concentration of cor- dritic cells that successfully bind viral particle and die at tisol in lymph node; p1 : Average population naive im- the rate γ1. Equation (6) describes the rate of change of mature dendritic cells; p2 : Average population partially the density of immature dendritic cells with viral parti- mature dendritic cells; p3 : Average population fully ma- cle ready for presentation in this case called partially ma- ture dendritic cells; q1 : Average population of naive reg- tured. They move randomly at the rate D5 while they also ulatory T-cells; q2 : Average population of activated regu- move towards naive CD4 T-cells at the rate of χ1(u1)= latory T-cells; q3 : Average population of naive cytotoxic d1 2 with dissociation constant d1. The rate at which (d1+u1) T-cells; q4 : Average population of activated cytotoxic T- they mature is b2 and also experience death at rate γ2. cells; u1 : Average population of naive T helper cells; u2 : Equation (7) stand for rate of change of the density of Average population of activated T helper cells; u3 : Aver- mature dendritic cells with effective chemotaxis χ2 (p2) = age population of latently infected T helper cells; u4 :Av- d2 2 with dissociation constant d2 and diffusion term erage population of actively infected T helper cells; v :Av- (d2+p2) D5Δp3, and natural death rate of γ3. Thus the following erage population density of HIV. equations: The immune cells release diverse intercellular messengers ∂p to communicate and direct the movement of both innate 1 = D Δp + ρ − b p v − γ p , (5) ∂t 5 1 1 1 1 1 1 and adaptive immune system [38, 39]. The immune system ∂p2 D ∇ ∇ ωb2p2 communication processes are very complex which com- = 5Δp2 − χ1(u1)p2 u1 + b1p1v − − γ2p2, ∂t ω + c4 prises of several chemical messengers secreted by different (6) immune cells for specific functions but for the sake of sim- ∂p3 D ∇ ∇ ωb2p2 plicity, we only consider c , c , c as described above. All = 5Δp3 + χ2(p2)p3 p2 + − γ3p3, (7) 1 2 3 ∂t ω + c4 the components have diffusion, secretion and decay terms 0 p1 (x, y,0) = p1, p2 (x, y,0) =0, p3 (x, y,0) =0inΩ, as given in equations (1)-(3) below ∂p ∂p ∂p 1 = 2 = 3 =0on∂Ω. ∂c1 ∂n ∂n ∂n = D1Δc1 + a1p3 − σ1c1, (1) ∂t Regulatory T cells down-regulate the proliferation of T ∂c2 helper cells to check excessive reaction of the immune sys- = D2Δc2 + a2u2 − σ2c2, (2) ∂t tem [43]. According to Sakaguchi et al [43] regulatory T ∂c 3 = D Δc + a u − σ c , (3) cells suppress the proliferation of naive T cells and their ∂t 3 3 3 2 3 3 differentiation to effector T cells in vivo. They can also sup- c1 (x, y,0) =0, c2 (x, y,0) =0, c3 (x, y,0) =0inΩ, press effector activities of differentiated CD4+ and CD8+ T ∂c ∂c ∂c 1 = 2 = 3 =0on∂Ω. cells and the function of natural killer cells, natural killer ∂n ∂n ∂n T cells, B cells, macrophages, osteoclasts, and dendritic There is strong evidence that adrenal hormones induced cells [44–47]. Once intercellular signal for this purpose is by stress, selectively inhibits or activates the response of received, the naive regulatory T cells are activated and di- immune cells [40]. Particularly, cortisol is known to inhibit rected towards the immune cells that need regulation with S. Samuel and V. Gill, Diffusion-Chemotaxis Model of Effects of Cortisol | 209
d3 sensitivity function χ3 (q4) = 2 with dissociation as in equation (13). Equation (14) models the subpopula- (d3+q4) constant d3. tion of latently infected T helper cells in which the term δ p + δ p + δ u + δ v u + u tells about infection ∂q ( 1 2 2 3 3 4 4 )( 1 2) 1 = D Δq + ρ − r c q − γ q , (8) ∂t 6 1 2 1 3 1 4 1 of both naive and activated T helper cells and subse- ∂q2 quently graduate into a group of actively infected T helper = D6Δq2 − ∇ χ3(q4)q2∇q4 + r1c3q1 − γ5q2, (9) ∂t cells at the rate k3 and die at the rate μ3. In equation (15), 0 q1 (x, y,0) = q1, q2 (x, y,0) =0inΩ, the actively infected T helper cells undergo lysing at the ∂q ∂q rate k4, they are also killed by the activated cytotoxic T 1 = 2 =0on∂Ω. ∂n ∂n cells at the rate r3 and die at rate μ4. Cytotoxic T cells kill the HIV infected immune cells to ∂u 1 = D Δu + ρ − k c u maintain the integrity of the immune system [48].The ac- ∂t 8 1 4 1 1 1 tivation of naive cytotoxic T cells is upregulated by corti- − (δ1p2 + δ2p3 + δ3u4 + δ4v) u1 − μ1u1, (12) sol concentration [f43]. In equation (10), ρ is the source 3 ∂u2 D c4 = 8Δu2 + k1c1u1 − (δ1p2 + δ2p3 + δ3u4 + δ4v) u2 term for naive cytotoxic T cells, r2 1+ ω c2q3 is the cor- ∂t tisol enhanced activation term and γ6q3is the portion of −μ2u2, (13) dead naive cytotoxic T cells. In equation (11) the activated ∂u3 = D8Δu3 + (δ1p2 + δ2p3 + δ3u4 + δ4v)(u1 + u2) cytotoxic T cells (q4) tend to move towards the infected T ∂t helper cells (u4) with a chemoattraction function χ4 (u4) = −k3u3 − μ3u3,(14) d4 2 with dissociation constant d4while k2q2q4 is the (d4+u4) ∂u4 = D8Δu4 + k3u3 − r3q4u4 − k4u4 − μ4u4, (15) regulation term; γ7q4 is the death term. To minimize the ∂t complexity of the model, it is assumed that, the HIV tar- 0 u1 (x, y,0) = u1, u2 (x, y,0) =0, u3 (x, y,0) =0, get cells are T helper cells while we neglect other immune u4 (x, y,0) =0inΩ, cells with low CD4 expression. ∂u ∂u ∂u ∂u 1 = 2 = 3 = 4 =0on∂Ω. ∂q c 3 = D Δq + ρ − r 1+ 4 c q − γ q , (10) ∂n ∂n ∂n ∂n ∂t 7 3 3 2 ω 2 3 6 3 Finally, the actively infected T helper cells are lysed to pro- ∂q4 c4 = D7Δq4 − ∇ χ4(u4)q4∇u4 + r2 1+ c2q3 duce N number of free virons per cell. The free virons are ∂t ω cleared from the system naturally at the rate μ5 and also −k2q2q4 − γ7q4, (11) by infecting the target cells at the rate δ4 as in equation q x, y,0 = q0, q x, y,0 =0inΩ, 3 ( ) 3 4 ( ) (16). ∂q3 ∂q4 = =0on∂Ω. ∂v ∂n ∂n = D Δv + Nk u − δ (u + u ) v − μ v, (16) ∂t 9 4 4 4 1 2 5 Human Immunodeficiency virus (HIV) attack is largely ∂v v (x, y,0) = v0 in Ω, =0on∂Ω. directed at the command centre of the immune system ∂n D [49–53]. In equations (12)-(15), 8 is common diffusion The parameter values of the model are presented in Table 1 coefficients of respective naïve, activated, latently in- below. fected and actively infected T helper cells for the sake of simplicity. During antigen presentation by the den- dritic cells, the naive T helper cells are activated at the rate k1. The infection of both naive and activated T 3 Kinetics System helper cells is considered in four ways: (i) during anti- gen presentation by the immature dendritic cells. (ii) Af- When the mobility of the cells is considered very small ter antigen presentation, mature dendritic cells migrate to be neglected, this results into the kinetic system of the with viral particles to lymph node crowded with CD4 T- model given below: cells. (iii) Free viral particles (iv) Cell-to-cell transfer of dc 1 = a p − σ c ,(17) virus by T helper cells. These four possible transmission dt 1 3 1 1 avenues are captured by (δ1p2 + δ2p3 + δ3u4 + δ4v) u1 dc2 = a2u2 − σ2c2, (18) and (δ1p2 + δ2p3 + δ3u4 + δ4v) u2. In equation (12) naive dt dc3 T helper cells have constant source ρ4 and death rate = a u − σ c , (19) dt 3 2 3 3 of μ1, the term k1c1u1 represents the proportion of ac- dc4 tivated T helper cells which experience death at rate μ = a4c4 − σ4c4 + g(t), (20) 2 dt 210 | S. Samuel and V. Gill, Diffusion-Chemotaxis Model of Effects of Cortisol
Table 1: Model Parameters.
Symbol Description Value/Reference a1 Rate of secretion chemokine by mature dendritic cells for antigen presentation 0.04 pg/cell/day a2 Rate of secretion of cytokine by activated T helper cells signalling attack on virus infected immune cells 0.06 pg/cell/day [54] a3 Rate of secretion of cytokine by activated T helper cells signalling for self-regulation 0.06 pg/cell/day [55] a4 Secretion rate of cortisol 0.04 [56] b1 Rate of activation of dendritic cells 0.3/day [57] b2 Rate of maturation of dendritic cells 0.4 /day [57, 58] k1 Rate of activation of naive T helper cells 0.4/day [57] k2 Rate of down-regulation by regulatory T-cells 0.3/cell/day [58] −4 k3 Rate of activation of latently infected T helper cells into actively infected T helper cells 1.2×10 [59] k4 Rate of lysing actively infected T helper cells 0.25 /day [59] r1 Rate of activation of naive cytotoxic T-cells 0.3/day r2 Rate of activation of naive regulatory T-cells 0.4/day r3 Rate of killing of infected T helper cells by cytotoxic T-cells 0.00001 /day 3 δ1 Antigen presentation enhanced infection rate 0.01vironsmm /day[60] 3 δ2 Mature dendritic cells enhanced infection rate 0.02vironsmm /day[60] δ3 Cell-to-cell infection rate 0.3/day [59] 3 δ4 Free virus-to-cell infection rate 0.05 vironsmm /day[59] N Number of viruses produced by lysing an infected T helper cells 1500 virons/cell [60] ω Normal body concentration of cortisol 0.28mol/mm3 [56] 2 ρ1 Source term for naive immature dendritic cells 1.25 x 10 cells[60] 2 ρ2 Source term for naive regulatory T-cells 3.63x10 cells[60] 2 ρ3 Source term for naive cytotoxic T-cells 3.63x10 cells[60] 2 ρ4 Source term for naive T helper cells 2.1x10 cells[60] σ1 Decay rate of chemokine secreted by dendritic cells 2.0/day σ2 Decay rate of cytokine secreted by T helper cells signalling attack on virus infected immune cells 2.16/day [61, 62] σ3 Decay rate of cytokine secreted by T helper cells signalling self-regulation 3.70/day [55] σ4 Decay rate of cortisol 0.1/day [56] γ1 Death rate of immature dendritic cells 0.35/day [63] γ2 Death rate of partially mature dendritic cells 0.25/day [63] γ3 Death rate of fully mature dendritic cells 1.0 /day [63] γ4 Death rate of naive regulatory T-cells 0.3/day [64] γ5 Death rate of activated regulatory T-cells 0.3/day [58] γ6 Death rate of naive cytotoxic T-cells 0.3/day [58] γ7 Death rate of activated cytotoxic T-cells 0.3/day [58] μ1 Death rate of naive T helper cells 0.06/day μ2 Death rate of activated T helper cells 0.06/day μ3 Death rate of latently infected T helper cells 0.06/day μ4 Death rate of actively infected T helper cells 0.24/day μ5 Death rate of free HIV 3/day 2 D1 Diffusion rate of chemokine secreted by dendritic cells 0.0049 mm /day [65] 2 D2 Diffusion rate of cytokine secreted by T helper cells signalling attack on virus infected immune cells 0.0042mm /day [66] 2 D3 Diffusion rate of cytokine secreted by T helper cells signalling self-regulation 0.0042mm /day [67] 2 D4 Diffusion rate of cortisol in lymph node 0.0049 mm /day 2 D5 Diffusion rate of dendritic cells 0.0028 mm /day [68] 2 D6 Diffusion rate of regulatory T-cells 0.0021 mm /day [69] 2 D7 Diffusion rate of cytotoxic T-cells 0.0056 mm /day [69] 2 D8 Diffusion rate of T helper cells 0.0045 mm /day [70] 2 D9 Diffusion rate of HIV 0.0008mm /day 2 d1 Chemotaxis rate of partially mature dendritic cells 0.0028 mm /day [69] 2 d2 Chemotaxis rate of fully mature dendritic cells 0.0028 mm /day [69] 2 d3 Chemotaxis rate of activated regulatory T-cells 0.0045 mm /day [70] 2 d4 Chemotaxis rate of activated cytotoxic T-cells 0.0045 mm /day[70] S. Samuel and V. Gill, Diffusion-Chemotaxis Model of Effects of Cortisol | 211
dp1 * * γ1γ2γ3ξ2 + ωb2γ1 = ρ1 − b1p1v − γ1p1,(21)u4 = k3, v = ,(33) dt b1ρ1 − ωb1b2 − b1γ2γ3ξ2 dp2 ωb2p2 = b1p1v − − γ2p2,(22) g g dt ω + c4 where ξ = 1+ , ξ = ω + and σ > a . 1 ω(σ4−a4) 2 σ4−a4 4 4 dp3 ωb2p2 = − γ3p3,(23) dt ω + c4 dq 1 = ρ − r c q − γ q ,(24)3.2 Linearized System Eigenvalues dt 2 1 3 1 4 1 dq 2 = r c q − γ q ,(25)Here, the kinetic system (17)-(32) of the model is linearized dt 1 3 1 5 2 dq c using the Jacobian matrix method [71]. Simple algebra 3 = ρ − r 1+ 4 c q − γ q ,(26) dt 3 2 ω 2 3 6 3 yields the eigenvalues of the Jacobian matrix given below: dq c 4 = r 1+ 4 c q − k q q − γ q ,(27) dt 2 ω 2 3 2 2 4 7 4 λ1 =−σ1, λ2 =−σ2, λ3 =−σ3, λ4 = a4 − σ4, du ωb 1 λ =−b v* − γ , λ =− 2 − γ , λ =−γ , = ρ4 − k1c1u1 − (δ1p2 + δ2p3 + δ3u4 + δ4v) u1 − μ1u1, 5 1 1 6 * 2 7 3 dt ω + c4 (28) * λ8 =−r1c3 − γ4, λ9 =−γ5, du2 = k1c1u1 − (δ1p2 + δ2p3 + δ3u4 + δ4v) u2 − μ2u2, c* dt λ =−r 1+ 4 c* − γ , λ =−γ , 10 2 ω 2 6 11 7 (29) du λ =−k c* − δ p* + δ p* + δ u* + δ v* − μ , 3 = (δ p + δ p + δ u + δ v)(u + u ) − k u − μ u , 12 1 1 1 2 2 3 3 4 4 1 dt 1 2 2 3 3 4 4 1 2 3 3 3 3 * * * * (30) λ13 =− δ1p2 + δ2p3 + δ3u4 + δ4v − μ2,
du4 * = k u − r q u − k u − μ u , (31) λ14 =−k3 − μ3, λ15 =−r3q4 − k4 − μ4, dt 3 3 3 4 4 4 4 4 4 dv λ =−δ u* + u* − μ . (34) = Nk u − δ (u + u ) v − μ v,(32) 16 4 1 2 5 dt 4 4 4 1 2 5 0 c1 (0) =0, c2 (0) =0, c3 (0) =0, c4 (0) = c4, Since all the eigenvalues are negative, the kinetic system 0 is uniformly and asymptotically stable. p1 (0) = p1, p2 (,0) =0,p3 (0) =0, 0 0 q1 (0) = q1, q2 (0) =0,q3 (0) = q3, q4 (0) =0, 0 u1 (0) = u1, u2 (0) =0,u3 (0) =0, 0 4Model Solution u4 (0) =0, v (0) = v 4.1 Numerical Method 3.1 The Equilibrium Points Firstly, we observed that the system (1)-(16) can be grouped into diffusion ((1)-(5), (8), (10), (12)-(16)) and chemotaxis The spatially homogeneous equilibrium point is gotten by ((6), (7), (9), (11)) equations thereby expressing the system setting the kinetic parts of (17)-(32) to zero and solving, we in the general form of Keller–Segel equations [72] thus: have a b ω ρ − γ γ ρ − γ γ Diffusion Chemotaxis Kinetic Terms c* = 1 2 , c* = 3 6 7 , c* = 2 4 5 , ∂u 1 σ 2 r γ ξ 3 r γ = D Δu − ∇ (χu∇v) + f (u, v) , (35) 1 2 7 1 1 5 ∂t 1 * g (t) * ρ1 − ωb2 − γ2γ3ξ2 Diffusion Kinetic Term c4 = , p1 = , σ4 − a4 γ1 ∂v = D2Δv + f(u, v) , (36) * * * * ρ2 − γ4γ5 ∂t p2=γ3ξ2, p3 = ωb2, q1 = γ5, q2 = , γ5 In this work, we use Radial Basis Functions- * k2ρ2 − k2γ4γ5 * ρ3 − γ6γ7 q3 = + γ7, q4 = , Pseudospectral (RBF-PS) method [73] to solve the model γ5 γ7 numerically. Other numerical techniques such as the fi- * ρ4 − μ2u2 − (k3 + μ3) u3 u1 = , nite difference methods (FDMs), finite element methods μ1 (FEMs), and finite volume methods (FVMs) can also be * r1γ5 (ρ3 − γ6γ7) + (ρ2 − r2γ4γ5γ7) ξ1 u2 = , applied. The pseudospectral (PS) techniques are superior r1r2γ5γ7ξ1 to FDMs and FEMs because of its spectral accuracy i.e. PS * r3 (ρ3 − γ6γ7) − γ7 (k4 + μ4) u3 = , methods achieve a desired accuracy with lesser number of γ7 212 | S. Samuel and V. Gill, Diffusion-Chemotaxis Model of Effects of Cortisol spatial grid-points than FDMs and FEMs [74]. Analytical where c is the shape parameter. More on RBFs solution of methods such as homotopy methods proved to be useful PDEs can be found in [75–78]. in case of systems of few equations but become cumber- In view of (37),(38) and (43), the approximate solutions of some and sometimes impossible with a system of large (35) and (36) at every time step tn = nΔt with zero bound- { }N { }N number of equations. RBF-PS method is implemented by ary conditions at the collocation points xi i=1 and yi i=1 formation of differentiation matrix associated with spec- is leads to a system of vector equations tral method using RBF instead of orthogonal polynomial Un+1 = Un + Δt D Un + Un − X + f (Un , Vn) , (46) such as chebyshev or hermite. The time partial deriva- 1 xx yy n+1 n n n n n tives are usually discretized by appropriate time stepping V = V + Δt D2 Vxx + Vyy + f (U , V )) ,(47) scheme. Here, we use the forward finite difference scheme given as where we use .* to represent element-wise multiplication of matrices and ∂u u (x, y, Δt + t) − u (x, y, t) Un+1 − Un ≈ = ,(37) ∂t Δt Δt n n n n n X = χ V . * Ux . * Vx + Uy . * Vy ∂v v (x, y, Δt + t) − v (x, y, t) V n+1 − V n ≈ = . (38) + χ Vn . * Un . * Vn + Vn ∂t Δt Δt xx yy n n n n n In case of spatial derivatives,we use RBF Ψ ( x − x ) for a + U . * χx V . * Vx + χy V . * Vy , i ∈ 2 n n N n −1 n n −1 n set of data points xi R , i = 1, ..., N such that the inter- U = U (x , y ) , U = A A U , U = A A U , i i i=1 x x y y polations of u and v are given as n −1 n n −1 n Uxx = AxxA U , Uyy = AyyA U . N N
U (x) = α1i Ψ ( x − xi ) , V (x) = α2i Ψ ( x − xi ) , Consequently, using the same approach, the discretized i=1 i=1 (39) form of system (1)-(16) is given as where . is the Euclidean norm, α1i and α2i are the co- n+1 n n n n n C1 = C1 + Δt D1 C1xx + C1yy + a1P3 − σ1C1 , (48) efficients respectively. The required derivatives of RBF are Cn+1 = Cn + Δt D Cn + Cn + a Un − σ Cn ,(49) given by 2 2 2 2xx 2yy 2 2 2 2 n+1 n D n n n n N C3 = C3 + Δt 3 C3xx + C3yy + a3U2 − σ3C3 , (50) ∂2 Un (x) = αn Ψ ( x − x ) , n+1 n xx 1i ∂x2 i C4 = C4 i=1 +Δt D Cn + Cn + a Cn − σ Cn + g(x , y , tn) , N 4 4xx 4yy 4 4 4 4 i j ∂2 V n (x) = αn Ψ ( x − x ) , (40) (51) xx 2i ∂x2 i i=1 Pn+1 = Pn 1 1 N n n n n n n n ∂ +Δt D5 P1xx + P1yy + ρ1 − b1P1. * V − γ1P1 ,(52) Ux (x) = α1i Ψ ( x − xi ) , ∂x n+1 n i=1 P = P 2 2 N ωb Pn n n ∂ +Δt D Pn + Pn − Θ + b Pn . V n − 2 2 − γ Pn , Vx (x) = α2i Ψ ( x − xi ) .(41)5 2xx 2yy 1 1 1 * n 2 2 ∂x ω + C4 i=1 (53) Now, equations (39)–(41)can be written in matrix form as Pn+1 = Pn 3 3 U = Aα1, V = Aα2,(42) n n n ωb2P2 n +Δt D5 P + P − Θ2 + − γ3P , (54) Uxx = Axxα , Vxx = Axxα , Ux = Ax α , Vx = Ax α ,(43) 3xx 3yy n 3 1 2 1 2 ω + C4 T T n+1 n where U = [u1, u2, ..., uN ] , V = [v1, v2, ..., vN ] ,Ai,j = Q1 = Q1 2 Ψ x − x , A = ∂ Ψ x − x ,A = n n n n n i j xxi,j ∂x2 i j xi,j +Δt D6 Q1xx + Q1yy + ρ2 − r1C3. * Q1 − γ4Q1 , (55) ∂ Ψ x − x α = α , α , ..., α T, α = ∂x i j 1 [ 11 12 1N ] 2 Qn+1 Qn T 2 = 2 [α21, α22, ..., α2N ] while xj are centres. Here, we used +Δt D Qn + Qn − Θ + r Cn . * Qn − γ Qn , (56) the following RBFs 6 2xx 2yy 3 1 3 1 5 2 n+1 n n n Q3 = Q3 + Δt D7 Q3xx + Q3yy + ρ3 Ψ x , y = x − x 2 + y − y 2 + c2 Multiquadric(MQ) ( i i) ( i) ( i) Cn −r 1+ 4 . * Cn . * Qn − γ Qn , (57) (44) 2 ω 2 3 6 3 2 2 n+1 n n n Ψ x , y = exp −c x − x − c y − y Gaussian(GA) Q4 = Q4 + Δt D7 Q4xx + Q4yy − Θ4 ( i i) ( i) ( i) Cn (45) + r 1+ 4 . * Cn . * Qn − k Qn . * Qn − γ Qn , (58) 2 ω 2 3 2 2 4 7 4 S. Samuel and V. Gill, Diffusion-Chemotaxis Model of Effects of Cortisol | 213