SOLUTION OF SINGULAR RICCATI DIFFERENTIAL EQUATIONS USING THE REPRODUCING KERNEL HILBERT SPACE METHO

Research Paper

E-ISSN No : 2454-9916 | Volume : 6 | Issue : 10 | Oct 2020

Mathematics

SOLUTION OF SINGULAR RICCATI DIFFERENTIAL EQUATIONS USING THE REPRODUCING KERNEL HILBERT SPACE METHOD

Neda Rahimi Department of Applied Mathematics, Hamedan Branch, Islamic Azad University, Hamedan, Iran. ABSTRACT This paper deals with the approximation the solution of singular Riccati differential equations using the reproducing kernel Hilbert space scheme. The exact solution u(r) is represented in the form of series in the spacew_2^2 [0,1]. In the mean time, the n-term approximate solution u(r) obtained and is proved to converge to the exact solution u(r). Some numerical examples have also been studied to demonstrate the accuracy of the present method. Numerical experiments are performed to confirm our theoretic findings. 2010 Mathematics Subject Classification: 34K28; 34K07; 34B10 KEYWORDS AND PHRASES: Singular Riccati differential equation, Reproducing kernel Hilbert space scheme, exact solution.

1. INTRODUCTION: In this work, we consider the following quadratic Riccati differential equation with singularity in reproducing kernel space P(r)u′(r) = Q(r)u2(r) + R(r)u(r) + S(r), u(0) = ξ,

0 ≤ r ≤ 1,

(1)

where P(r), Q(r), R(r) and S(r) coefficients are continuous real functions, perhaps P(0) = 0 or P(1) = 0 and 𝑢(𝑟) ∈ 𝑤22 [0,1] Many physical phenomena such as theory of thermal explosions, studies of electro-hydrodynamics, nuclear physics, gas dynamics, chemical reaction, studies of atomic structures and atomic calculations can be modeled by singular differential equations. Due to this reason, in the literature there exist numerous methods dedicated to the singular problems, for example see [1–3]. The theory of reproducing kernels, firstly has been presented based on the S. Zaremba works in [4] for solving boundary value problems with harmonic and bi-harmonic functions. In recent years, reproducing kernel methods emerged as a powerful framework in numerical analysis, differential, integral equations, probability and statistics [5–7]. Furthermore, extensive works have been proposed and discussed based on their theory for the numerical solutions of singular problems. For example, using operator transformation considered the solution of Cauchy singular integral equation by reproducing kernel Hilbert space (RKHS) scheme in [8]. Also, in [9] has been used from RKHS method to approximating the solution of the second kind weakly singular Volterra integral equation on graded mesh. In [10–17] have been solved various classes of singularly boundary value problems. In [18–20] the authors has been implemented the RKHS method for solving singularly perturbed differential-difference equations. A brief outline of this paper is as follows: In section 2, RKHS is introduced. In section 3, the analytical solution for (1) in the space 𝑤22 [0,1] is introduced. Also, an iterative method to solve (1) numerically in the space 𝑤22 [0,1] is described. In section 4 the numerical examples are discussed. Then we end with a brief conclusion in section 5. 2. REPRODUCING KERNEL SPACES: In order to solution of (1), we construct a reproducing kernel functions. Definition 2.1. ([21]) Let E be a nonempty abstract set and C be the set of complex numbers. A function K: E × E → C is a reproducing kernel of the Hilbert space H if

• for each t ∈ E, K(.,t) ∈ H, • For each t ∈ E and ψ ∈ H, ⟨ψ (.), K (., t)⟩ = ψ(t). Copyright© 2020, IERJ. This open-access article is published under the terms of the Creative Commons Attribution-NonCommercial 4.0 International License which permits Share (copy and redistribute the material in any medium or format) and Adapt (remix, transform, and build upon the material) under the Attribution-NonCommercial terms.

International Education & Research Journal [IERJ]

38

Research Paper

E-ISSN No : 2454-9916 | Volume : 6 | Issue : 10 | Oct 2020

In Definition 2.1, second condition called the reproducing property, the value of the function ψ at the point t is reproducing by the inner product of ψ(t) with K(.,t). Remark 2.1. A Hilbert space H of functions on a set E is called a RKHS if there exists a reproducing kernel K of H. That is, a Hilbert space which possesses a reproducing kernel is called the RKHS. Definition 2.2. The inner product space w22 [0,1] is defined as w22 [0,1] = {x(t)|x(t),x′(t), is absolutely continuous, x(0) = α and x′′(t) ∈ L2[0,1]}. The inner product w22 [0,1] is defined by < 𝑥 (𝑡), 𝑦(𝑡) >

𝑤22 [0,1]

1

= 𝑥(0)𝑦(0) + 𝑥(1)𝑦(1) + ∫ 𝑥 " (𝑡)𝑦 " (𝑡)𝑑𝑡,

(2)

0

and the norm ‖𝑥‖𝑤22 [0,1] is denoted by ‖𝑥‖𝑤22 [0,1] = √< 𝑥, 𝑥 > 𝑤22 [0,1] , where 𝑥, 𝑦 ∈ 𝑤22 [0,1] Theorem 2.1. ( [22]) The space w22 [0,1] is a reproducing kernel space. That is, for any x(t) ∈ w22 [0,1] and each fixed t ∈ [0,1] there exists Rz(t) ∈ w22 [0,1] , such that < x(t), R z (t) > w22 [0,1] = x(z) . The reproducing kernel Rz(t) can be denoted by 𝑡

𝑅𝑧 (𝑡) = {6𝑧 6

((𝑧 − 1)𝑡 2 + 𝑧(𝑧 2 − 3𝑧 + 8)), 𝑡 ≤ 𝑧, (𝑧 2 (𝑡 − 1) + 𝑡(𝑡 2 − 3𝑡 + 8)) , 𝑡 > 𝑧.

(3)

Definition 2.3. The inner product space 𝑤22 [0,1] is defined as 𝑤22 [0,1] = {x(t)|x(t), is absolutely continuous, x(0) = α and x′(t) ∈ L2[0,1]}. The inner product 𝑤22 [0,1] is defined by 1

< 𝑥 (𝑡), 𝑦(𝑡) > 𝑤21 [0,1] = 𝑥(0)𝑦(0) + ∫0 𝑥 ′ (𝑡)𝑦 ′ (𝑡)𝑑𝑡,

(4)

and the norm ‖x‖w21 [0,1] is denoted by‖x‖w21 [0,1] = √< x, x > w21 [0,1] , where x, y ∈ w21 [0,1]. In [5] proved that 𝑤21 [0,1] is a complete reproducing kernel space and its reproducing kernel is 1 + 𝑡, 𝑅̅𝑧 (𝑡) = { 1 + 𝑧,

𝑡 ≤ 𝑧, 𝑡 > 𝑧,

(5)

3. THE ANALYTICAL SOLUTION OF (1) In this section, the solution of (1) is given in the reproducing kernel space 𝑤21 [0,1] In (1), it is clear that ℒ: 𝑤22 [0,1] → 𝑤21 [0,1] is a bounded linear operator. ̅ 𝑠 (𝑧)} ∞ of 𝑤22 [0,1] can Put χs(z) = Rzs(z) and ϕs(z) = L∗χs(z), where L∗ is the adjoint operator of L. The orthonormal system {∅ 𝑠=1 ̅ 𝑠 (𝑧)} ∞ be derived from GramSchmidt orthogonalization process of {∅ 𝑠=1 𝑠

̅ 𝑠 (𝑧) = ∑ ∝𝑠𝑖 ∅𝑖 (𝑧), (∝𝑠𝑠 > 0, 𝑠 = 1,2, . . . ) ∅ 𝑖=1

Theorem 3.1. ([22])Let {zs }

∞ ̅s (z)} ∞ is the complete system of w22 [0,1] and ϕs(z) = LtRz(t)|t=zs. is dense on [0,1], then {∅ s=1 s=1

∞ Theorem 3.2. If {zs } is dense on [0,1] and the solution of (1) is unique, then the solution of (1) satisfies the form s=1 𝑠 ̅ 𝑣(𝑧) = ∑∞ 𝑠=1 ∑𝑖=1 ∝ 𝑠𝑖𝑓(𝑧𝑖 , 𝑣(𝑧𝑖 ))∅𝑠 (𝑧)

39

(6)

International Education & Research Journal [IERJ]

Research Paper

E-ISSN No : 2454-9916 | Volume : 6 | Issue : 10 | Oct 2020 ̅ 𝑠 (𝑧)} ∞ is the complete orthonormal basis of 𝑤22 [0,1]. Remark that from Proof. From Theorem 3.1, one can easily see that {∅ 𝑠=1 reproducing property ⟨ν(z),χs(z)⟩ = ν(zs) for each 𝑣(𝑧) ∈ 𝑤22 [0,1], so we have ∞

̅𝑠 (𝑧) > 𝑤22 [0,1]∅ ̅ 𝑠 (𝑧) 𝑣(𝑧) = ∑ < 𝑣(𝑧), ∅ 𝑠=1 ∞

𝑠

∗ ̅𝑠 (𝑧) = ∑ ∑ ∝ 𝑠𝑖 < 𝑣(𝑧), ℒ𝑥𝑠 (𝑧) > 𝑤22 [0,1]∅ 𝑠=1 𝑖=1 ∞

𝑠

̅𝑠 (𝑧) = ∑ ∑ ∝ 𝑠𝑖 < ℒ𝑣(𝑧), 𝑥𝑠 (𝑧) > 𝑤22 [0,1]∅ 𝑠=1 𝑖=1 ∞ 𝑠

̅ 𝑠 (𝑧) = ∑ ∑ ∝ 𝑠𝑖 < 𝑓(𝑧, , 𝑣(𝑧)), 𝑥𝑠 (𝑧) > 𝑤22 [0,1]∅ 𝑠=1 𝑖=1 ∞ 𝑠

̅𝑠 (𝑧), = ∑ ∑ ∝ 𝑠𝑖𝑓(𝑧𝑖 , 𝑣(𝑧𝑖 )), ∅ 𝑠=1 𝑖=1

and the proof of the theorem is complete. 3.3 The implementation scheme: Equation (6) can be denoted by

∞

̅ 𝑠 (𝑧) 𝑣(𝑧) = ∑ 𝐴𝑠 ∅ 𝑠=1

, Where 𝐴𝑠 = ∑𝑠𝑖=1 ∝𝑠𝑖 𝑓(𝑧𝑖 , 𝑣(𝑧𝑖 ). Let z1 = 0, it follows that f(z1,υ(z1)) is known. Considering the numerical computation, we put υ0(z1) = υ(z1) and define the n-term approximation to υ(z) by ̅ 𝑣𝑛 (𝑧) = ∑∞ 𝑠=1 𝐵𝑠 ∅𝑠 (𝑧) , where the coefficients Bs and ϕs(z) are given as

(7)

𝐵1 = ∝11 𝑓 ̅(𝑧1 , 𝑣0 (𝑧1 )), ̅1 (𝑧), 𝑣1 (𝑧) = 𝐵1 ∅ 2

𝐵1 = ∑ ∝2𝑖 𝑓(𝑧𝑖 , 𝑣𝑖−1 (𝑧𝑖 )) 𝑖=1

̅ 𝑠 (𝑧), 𝑣1 (𝑧) = ∑2𝑠=1 𝐵𝑠 ∅ ... 𝐵𝑛−1 = ∑2𝑖=𝑛−1 ∝2𝑖 𝑓(𝑧𝑖 , 𝑣𝑖−1 (𝑧𝑖 )) ̅𝑠 (𝑧). 𝑣𝑛 (𝑧) = ∑𝑛𝑠=1 𝐵𝑠 ∅ In the iterative process of (7), we can guarantee that the approximation υn(z) satisfies the initial condition u(0) = ξ. On the other hand, the approximate solution ) can be obtained by taking finitely terms in the series of υn(z) as follow, 𝑠 ̅ 𝑣𝑛𝑁 (𝑧) = ∑𝑁 𝑠=1 ∑𝑖=1 ∝𝑠𝑖 𝑓(𝑧𝑖 , 𝑣𝑛−1 (𝑧𝑖 ))∅𝑠 (𝑧)

(8)

4. NUMERICAL SIMULATIONS In this section, the scheme in the paper will be applied to four numerical examples. All of the computations have been performed by using the Matlab R2010a. Results of each example are compared with exact solution. Problem 4.1. In equation (1), if P(r) = r, Q(r) = 1, R(r) = −√r and S(r) = −1 then true solution is

International Education & Research Journal [IERJ]

40

Research Paper

E-ISSN No : 2454-9916 | Volume : 6 | Issue : 10 | Oct 2020 𝑢(𝑟) =

𝑢(0) = 1, 𝑟𝑠 =

2√𝑟 − 3 −2𝑟 + 4√𝑟 − 3 𝑠−1 , 𝑠 = 1,2, … , 𝑁 𝑁−1

RKHS method, taking with the reproducing kernel function Rz(t) on [0,1], the numerical solution uN n (r) computed by (8). The numerical results at some selected gird points for N = 51 and n = 7 are given in Table 1.

Table 1: Numerical results for problem 4.1.

Problem 4.2. Consider the following singular equation (1 − 𝑟)𝑢′ (𝑟) = 𝑢2 (𝑟) + 𝑢(𝑟),

1

𝑢(0) = − , 2

𝑢(𝑟) = 𝑟𝑠 =

0 ≤ 𝑟 < 1 , with true solution 1 𝑟−2

𝑠−1 , 𝑠 = 1,2, … , 𝑁 𝑁−1

Using our method, taking and gird points N = 51 and n = 9, the numerical results are as given in Figure 1.

Figure 1: Comparisons of approximate solution with the exact solution (left) and the absolute errors with the relative errors of problem 4.2 (right). Problem 4.3. Consider the following singular equation

41

International Education & Research Journal [IERJ]

Research Paper

E-ISSN No : 2454-9916 | Volume : 6 | Issue : 10 | Oct 2020 𝑢 ′ (𝑟) − √𝑟𝑢2 (𝑟) + 𝑢(𝑟) =

𝑢(𝑟) √𝑟

,

𝑢(0) = 1,

0<𝑟≤1

2 1 + 2 (𝑟 − √𝑟) + 𝑒 −2√𝑟

. by RKHS method, taking 𝑟𝑠 = 9 are given in Table 2.

𝑠−1 𝑁−1

, 𝑠 = 1,2, … , 𝑁 𝑜𝑛 [0,1]. The numerical results at some selected gird points for N = 53 and n =

Table 2: Numerical results for problem 4.3.

Problem 4.4. Consider the following singular equation (1 − 𝑒 1−𝑟 )𝑢′ (𝑟) = 𝑢2 (𝑟), 𝑢(0) =

u(r) = Using our method, take 𝑟𝑠 =

𝑠−1 𝑁−1

1 2−𝑒

,

0≤𝑟<1

, with true solution

1

e𝑟 + 1 − e

, 𝑠 = 1,2, … , 𝑁 𝑁 = 35. and n = 7. The numerical results are given in Figure 2.

Figure 2: Comparisons of approximate solution with the exact solution (left) and the absolute errors with the relative errors of problem 4.4 (right). 5. CONCLUSIONS:

International Education & Research Journal [IERJ]

42

Research Paper

E-ISSN No : 2454-9916 | Volume : 6 | Issue : 10 | Oct 2020

In this paper, the reproducing kernel Hilbert space scheme was implemented for solving a singular Riccati differential equations. This confirms the validity of the present method and it is efficient, accurate and reliable for singular Riccati differential equations. REFERENCES: I. K. Maleknejad, A. Arzhang, Numerical solution of the Fredholm singular integro-differential equation with Cauchy kernel by using Taylor-series expansion and Galerkin method, Applied Mathematics and Computation 182 (2006) 888-897. II. H. Brunner, A. Pedas, G. Vainikko, A spline collocation method for linear Volterra integro-differential equations with weakly singular kernels, BIT 41(5) (2001) 891-900. III. P. M. Lima, A. Bellour, M. V. Bulatov, Numerical solution of integrodifferential equations arising from singular boundary value problems, Applied Mathematics and Computation 336 (2018) 1-15. IV. N. Aronszajn, Theory of reproducing kernels, Transactions of the American Mathematical Society, 68 (1950) 337404. V. M.G. Cui, Y.Z. Lin, Nonlinear Numerical Analysis in Reproducing Kernel Space, Nova Science Publisher, New York, 2009. VI. D. Alpay, Reproducing Kernel Spaces and Applications, Birkh¨auser, Berlin, 2003. VII. A. Berlinet, C. Thomas-Agnan, Reproducing Kernel Hilbert Space in Probability and Statistics, Kluwer Academic Publishers, 2004. VIII. H. Beyrami, T. Lotfi, K. Mahdiani, A new efficient method with error analysis for solving the second kind Fredholm integral equation with Cauchy kernel, J. Comput. Appl. Math. 300 (2016) 385–399. IX. H. Beyrami, T. Lotfi, K. Mahdiani, Stability and error analysis of the reproducing kernel Hilbert space method for the solution of weakly singular Volterra integral equation on graded mesh, Applied Numerical Mathematics 120 (2017) 197-214. X. Y. Wang, T. Chaolu, Z. Chen, Using reproducing kernel for solving a class of singular weakly nonlinear boundary value problems, Int. J. Comput. Math. 87 (2010) 367-380. XI. F.Z. Geng, M.G. Cui, Solving singular nonlinear two-point boundary value problems in the reproducing kernel space, J. Korean Math. Soc. 45(3) (2008) 77-87. XII. M.G. Cui and F.Z. Geng, Solving singular two-point boundary value problem in reproducing kernel space, J. Comput. Appl. Math. 205 (2007) 6-15. XIII. M. Khaleghi, M.T. Moghaddam, E. Babolian, S. Abbasbandy, Solving a class of singular two-point boundary value problems using new effective reproducing kernel technique, Appl. Math. Comput. 331 (2018) 264-273. XIV. F. Geng, A novel method for solving a class of singularly perturbed boundary value problems based on reproducing kernel method, Appl. Math. Comput. 218 (2011) 4211-4215. XV. Y. Wang, L. Su, X. Cao, X. Li, Using reproducing kernel for solving a class of singularly perturbed problems, Comput. Math. Appl. 61 (2011) 421-430. XVI. F.Z. Geng, S.P. Qian, S. Li, A numerical method for singularly perturbed turning point problems with an interior layer, J. Comput. Appl. Math. 225 (2014) 97-105. XVII. F.Z. Geng, S.P. Qian, Reproducing kernel method for singularly perturbed turning point problems having twin boundary layers, Appl. Math. Lett. 26 (2013) 998-1004. XVIII. F.Z. Geng, S.P. Qian, M.G. Cui, Improved reproducing kernel method for singularly perturbed differential-difference equations with boundary layer behavior, Appl. Math. Comput. 252 (2015) 58-63. XIX. H. Sahihi, S, Abbasbandy, T. Allahviranloo, Reproducing kernel method for solving singularly perturbed differentialdifference equations with boundary layer behavior in Hilbert space, J. Comput. Appl. Math. 328 (2018) 30-43. XX. H. Sahihi, S, Abbasbandy, T. Allahviranloo, Computational method based on reproducing kernel for solving singularly perturbed differential-difference equations with a delay, Appl. Math. Comput. 361 (2019) 583-598. XXI. F. Geng, Solving singular second order three-point boundary value problems using reproducing kernel Hilbert space method, Appl. Math. Comput. 215 (2009) 2095-2102. XXII. M.G. Sakar, Iterative reproducing kernel hilbert spaces method for Riccati differential equations, J. Comput. Appl. Math. 309 (2017) 163-174.

43

International Education & Research Journal [IERJ]