/compile/output.dvi Polynomial Integral Transform for Solving Differential Equations Benedict Barnes School of Agriculture and Social Sciences, Anglican University of College of Technology, Nkoranza Campus, Brong Ahafo, Ghana Department of Mathematics, KNUST, Kumasi, Ghana Abstract. In this paper, we propose Polynomial Integral Transform for solving differential equations. Unlike the Laplace Transform and others, the Polynomial Integral Transform solves differential equa- tions with a little computational effort as well as time. In addition, the Polynomial Integral Transform entails a polynomial function as its kernel, which ensures the rapid convergence of the solution to a differential equation. Thus, this method transforms a linear differential equation into an algebraic equation, from which the solution is obtained. Moreover, we show the applicabilities of the Polynomial Integral Transform and its properties. 2010 Mathematics Subject Classifications: 44B24; 44B21 Key Words and Phrases: polynomial integral transform, polynomial function, kernel, differential equa- tions 1. Introduction Most of the problems encountered in science, engineering and physics involve rates of change. Initial condition of the dependent variable is usually measured at a point. Finding solutions to these problems are often either difficult or not feasible at all. There are many approaches to search for solution to the differential equation. Special substitution techniques have been adopted in finding solution to differential equation with variable coefficients. The Cauchy-Euler method transforms a linear differential equation into an algebraic equation with the use of appropriate substitution technique. Other methods such as methods of undeter- mined coefficients, variation of parameters are limited in usage, see [8]. In addition, these classical methods for search of solutions to the differential equations are tedious and cumber- some as one has to look for the appropriate substitution expression. Thus, there is no single substitution expression for a single type of differential equation. Email addresses: ewiekwamina@gmail.com, fkofi33@yahoomail.com http://www.ejpam.com 140 c© 2016 EJPAM All rights reserved. EUROPEAN JOURNAL OF PURE AND APPLIED MATHEMATICS Vol. 9, No. 2, 2016, 140-151 ISSN 1307-5543 – www.ejpam.com B. Barnes / Eur. J. Pure Appl. Math, 9 (2016), 140-151 141 Currently, integral transform method is the concern of mathematicians and scientists in general. Since the introduction of the Laplace integral transform, a number of integral trans- forms have been proposed for solving differential equations. An alternative integral transform, Laplace substitution method, for the construction of solutions of the partial differential equa- tions was observed by [11]. The Sumudu integral transform F(u) = s[ f (t); u] = 1 u ∫ ∞ 0 e− t u f (t)d t, u ∈ (−τ,τ). was proposed by [15] and applied to some controlled problems in engineering. This method faces the similar challenges as the Laplace integral transform. Thus, Sumudu integral trans- form solves a linear differential equation with constant coefficients. The author in [4], ob- served some properties of the Sumudu integral transform. Several studies have made use of the Sumudu integral transform to obtain the solutions of the differential equations. For ex- ample, see research papers by [5, 6, 9]. Recently, in [10], they compared both the Laplace integral transform and Sumudu integral transforms. In [12], the authors introduced the Natural integral transform and applied it to obtain solutions of the differential equations. Unlike the Laplace and Sumudu integral transforms, the Natural integral transform entails K(u, v, x) = e− vx u , where u and v are parameters, as its kernel, which transforms a linear differential equation into an algebraic equation. The duality of both the Natural and Laplace integral transforms was studied by [7]. Using the Natural integral transform, [2] sought the solution of differ- ential equation on the spaces of generalized functions. Also, in [1], the author extended the applications of the Hartley transform of differential equation on the space of the generalized functions. In order to ensure the rapid convergence of solution of the differential equation, the Fresnel integral transform with variables in the Boehmains space has been obtained. For example, see a research paper by [3]. On the contrary, the integral transform method for the fractional difference equation has been obtained. The authors in [14], implemented the S-transforms for solving such problems in engineering. We outline of this paper is as follows. In section 1, we give the introduction to integral transform methods. In this section, we discuss the integral transform methods for solving differential equations. In section 2, we present the definition and also, give the proof for the Polynomial Integral Transform. Using the Polynomial Integral Transform, we show that the solution of the differential equation converges for x ∈ [1,∞). The properties of the Polynomial Integral Transform is contained in section 3. In section 4, we apply the Polynomial Integral Transform to derivatives, some ordinary differential equations and partial differential equation. Section 5 contains the conclusion of this paper. B. Barnes / Eur. J. Pure Appl. Math, 9 (2016), 140-151 142 2. The Polynomial Integral Transform In the previous section, we observed that the discussed integral transforms are either pro- totypes or have almost the same applicabilities as the Laplace integral transform. In addition, almost all of these integral transforms use exponential function of parameter(s) as their ker- nels. Using the exponential function kernel does not only requires complex mathematical structures but also takes a long time before the solution is obtained. One of the interesting papers which has drawn much attention in the 21st century is the one given by [13]. Using the Mellin-Barnes integrals poses the similar challenges as the Laplace integral transform and its prototypes. An integral transform which uses polynomial function as its kernel requires a few time for computation as well as the convergence of the solution of the differential equation. In this paper, we introduce a Polynomial Integral Transform to solve differential equations in Hilbert space. This method entails a polynomial function as its kernel to transform the differential equation into the algebraic equation. The algebraic equation is then solved to obtain the solution of the differential equation. We state the Polynomial Integral Transform theorem for the ordinary differential equation. Theorem 1 (A Polynomial Integral Transform). Let f (x) be a function defined for x ≥ 1. Then the integral B( f (x)) = F(s) = ∫ ∞ 1 f (ln x).x−s−1d x , is the Polynomial Integral Transform of f (x) for x ∈ [1,∞), provided the integral converges. Proof. We consider the homogeneous Cauchy-Euler equation of the form: an xn dn y d xn + an−1 xn−1 dn−1 y d xn−1 + . . .+ a1 x d y d x + a0 y = 0. with the corresponding distinct roots y(x) = c1es1 ln x + c2es2 ln x + . . .+ cnesn ln x , where c1, c2, . . . , cn are constants. Also, consider a constant linear differential equation an dn y d tn + an−1 dn−1 y d tn−1 + . . .+ a1 d y d t + a0 y = 0, with a solution y(x) = c1es1 t + c2es1 t + . . .+ cnesn t . We can see that equation (3) has an integral transform B( f (t)) = F(s) = ∫ ∞ 0 f (t)e−st d t. B. Barnes / Eur. J. Pure Appl. Math, 9 (2016), 140-151 143 Again, we see from equations (2) and (4) that t = ln x Substituting equation (6) into equation (5), we obtain B( f (x)) =F(s) = ∫ ∞ 1 f (ln x). 1 x e−s ln x d x B( f (x)) = ∫ ∞ 1 f (ln x).x−(s+1)d x is the polynomial integral transform of f (x) for x ∈ [1,∞), provided the integral converges. 2.1. The Convergence of the Polynomial Integral Transform In this subsection, we show that the Polynomial Integral Transform converges for variable defined in [1,∞). By Taylor series expansion, we obtain eln x−s−1 =1+ ln x−s−1 + (ln x−(s+1))2 2! + (ln x−(s+1))3 3! + (ln x−(s+1))4 4! + . . .+ ∞ ∑ n=0 (ln x−(s+1))n n! + . . . eln x−s−1 = ∞ ∑ n=0 (ln x−(s+1))n n! By the D’Lambert Ratio test, we obtain lim n→∞ |( ∞ ∑ n=0 (ln x−(s+1))n+1 (n+ 1)! ÷ ∞ ∑ n=0 (ln x−(s+1))n n! )| ⇒0. ln x−(s+1) ⇒0. Then B( f (x)) = sup 1≤x<∞ ∫ ∞ 1 | f (ln x).x−s−1|d x B( f (x))≤ sup 1≤x<∞ ∫ ∞ 1 | f (ln x)||x−s−1|d x B( f (x))≤M ∫ ∞ 1 | f (ln x)|d x , where M > 0. It implies that the polynomial integral transform converges uniformly for a given s. The func- tion f (x) must be piecewise continuous. Thus, f (x) has at most a finite number of disconti- nuities on any interval 1≤ x ≤ A, and the limit of f (x) exist at every point of discontinuity. B. Barnes / Eur. J. Pure Appl. Math, 9 (2016), 140-151 144 2.2. Existence of the Polynomial Integral Transform In this subsection, we show that the Polynomial Integral Transform exists for x ∈ [1,∞). To see this, we state the existence theorem for the Polynomial Integral Transform. Theorem 2. Let f (x) be a piecewise continuous function on [1,∞) and of exponential order, then the polynomial integral transform exists. Proof. By the definition of polynomial integral transform, we obtain I = ∫ ∞ 1 f (x).x−(s+1)d x I = ∫ A 1 f (x).x−(s+1)d x + ∫ ∞ A f (x).x−(s+1)d x I =I1 + I2, where I1 = ∫ A 1 f (x).x−(s+1)d x and I2 = ∫ ∞ A f (x).x−(s+1)d x . The integral I1 exists since f (x) is piecewise continuous. Taking I2 = ∫ ∞ A f (x).x−(s+1)d x I2 = ∫ ∞ A f (x).x−(s+1)d x ≤ M ∫ ∞ A eαx .x−(s+1)d x . By the Taylor series expansion, we obtain eαx ≈ ∞ ∑ n=0 α n xn n! . Substituting the expression for eαx in equation (7), we obtain I2 ≈M ∞ ∑ n=0 α n n! ∫ ∞ A x−(s+1−n)d x I2 ≈M ∞ ∑ n=0 α n n! ∫ ∞ 1 x−(s+1−n)d x I2 =M ∞ ∑ n=0 Mαn n!(s− n) , s > n B. Barnes / Eur. J. Pure Appl. Math, 9 (2016), 140-151 145 ∫ ∞ 1 f (x).x−(s+1)d x =M ∞ ∑ n=0 Mαn n!(s− n) , s > n This completes the proof. 3. Properties of the Polynomial Integral Transform In this section, we give the properties of the Polynomial Integral Transform. Theorem 3. The Polynomial Integral Transform is a linear operator. Proof. Suppose that f (x) and g(x) are functions and α1 and α2 are real constants. B(α1 f (x) +α2 g(x)) = ∫ ∞ 1 (α1 f (ln x) +α2 g(ln x)).x−s−1d x B(α1 f (x) +α2 g(x)) =α1 ∫ ∞ 1 f (ln x).x−s−1d x +α2 ∫ ∞ 1 g(ln x).x−s−1d x B(α1 f (x) +α2 g(x)) =α1B( f (x)) +α2B(g(x)) Theorem 4. The Inverse Polynomial Integral Transform is a also linear operator. Proof. Taking the inverse integral transform of the both sides of the above equation, we obtain α1 f (x) +α2 g(x) =B−1(α1( f (x)) +α2B(g(x))) α1 f (x) +α2 g(x) =α1B−1(( f (x))) +α2B−1(L(g(x))) α1 f (x) +α2 g(x) =α1B−1(F(s)) +α2B−1(G(s)) α1 f (x) +α2 g(x) =B−1(α1F(s) +α2G(s)), where B( f (x)) = F(s) and B(g(x)) = G(s), respectively. Theorem 5 (First shifting theorem). If B( f (x)) = B(s), then B(eax f (x)) = B(s− a), for s > 1. Proof. Let B(eax f (x)) = ∫ ∞ 1 ea ln x f (ln x).x−s−1d x B(eax f (x)) = ∫ ∞ 1 xa f (ln x).x−s−1d x B(eax f (x)) = ∫ ∞ 1 f (ln x).x−(s−a+1)d x B. Barnes / Eur. J. Pure Appl. Math, 9 (2016), 140-151 146 B(eax f (x)) =B(s− a). Theorem 6 (Second shifting theorem). Let Hc(x) = ¨ 0 0≤ x < c 1 x ≥ c be a unit step function. Then B(Hc f (x − c)) = F(s− c) Proof. By applying the Polynomial Integral Transform, we obtain B(Hc(x) f (x − c)) = ∫ ∞ 1 Hc(ln x) f (ln(x − c)).x−s−1d x B(Hc(x) f (x − c)) = lim t→∞ ∫ t 1 1. f (ln(x − c)).x−s−1d x B(Hc(x) f (x − c)) = lim t→∞ ∫ t 1 f (ln(x − c)).x−s−1d x We set u= x − c and substituting u into right hand side of the above equation, we obtain B(Hc(x) f (x − c)) = lim t→∞ ∫ t−c 1−c f (ln u).(u+ c)−s−1du B(Hc(x) f (x − c)) = lim t→∞ ∫ t 1 f (ln(v − c)).v−s−1dv B(Hc(x) f (x − c)) =F(s− c), where v = u+ c. Theorem 7. If f (x) is a piecewise continuous function on [0,∞), but not of exponential order, then a polynomial integral transform B( f (x))→ 0 as s→∞. Proof. Let |B( f (x))|=| ∫ ∞ 1 f (ln x)x−s−1d x | B. Barnes / Eur. J. Pure Appl. Math, 9 (2016), 140-151 147 |B( f (x))| ≤ ∫ ∞ 1 | f (ln x)x−s−1|d x |B( f (x))|= ∫ ∞ 1 f (ln x)|x−(s+1)|d x . But, we observe that: |x−(s+1)| → 0 as s→∞. It follows that B( f (x))→ 0 as s→∞. 4. The Polynomial Integral Transform of Derivatives In this section, we give the Polynomial Integral Transform of derivatives of the function f (x) with respect to x . Theorem 8. If f , f ′, . . . f n−1 are continuous on [1,∞) and if f n(x) is piecewise continuous on [1,∞), then B( f (n)(x)) = snF(s)− sn−1 f (0)− sn−2 f ′(0)− . . .− f (n−1)(0), where F(s) = B( f (x)). Proof. Let B( f ′(x)) = ∫ ∞ 1 f ′(ln x).x−s−1d x Using integration by parts, we obtain L( f ′(x)) = lim t→∞ [ f (ln x)x−s]t1 + lim t→∞ ∫ t 1 f (ln x) 1 x x−sd x L( f ′(x)) =sF(s)− f (0). Proceeding a similar as above, we obtain B( f ′′(x)) = ∫ ∞ 1 f ′′(ln x).x−s−1d x B( f ′′(x)) =− f ′(0) + sL( f ′(x)) Substituting the expression of f ′(x) into the above equation, we obtain B( f ′′(x)) = s2F(s)− s f (0)− f ′(0). By induction, we obtain B( f (n)(x)) = snF(s)− sn−1 f (0)− sn−2 f ′(0)− . . .− f (n−1)(0), where F(s) = B( f (x)). B. Barnes / Eur. J. Pure Appl. Math, 9 (2016), 140-151 148 Corollary 1. Suppose f is a piecewise function and let F(s) by the Polynomial Integral Transform by equation (4). Then, we obtain L(xn f (x))(s) = (−1)nF (n)(s) Proof. By applying the Polynomial Integral Transform, we obtain F ′(s) = d ds ∫ ∞ 1 f (ln x).x−s−1d x F ′(s) = ∫ ∞ 1 f (ln x).x−1 ∂ ∂ s eln x−s d x F ′(s) = ∫ ∞ 1 f (ln x).x−1 ∂ ∂ s e−s ln x d x F ′(s) =− ∫ ∞ 1 ln x f (ln x).x−s−1d x −F ′(s) =L(x f (x)) Proceeding in a similar manner, we obtain F ′′(s) = d ds ∫ ∞ 1 ln x f (ln x).x−s−1d x F ′′(s) =L(x2 f (x)) ... (−1)nF (n)(s) =L(xn f (x))(s), where n= 1,2, . . .. 4.1. Applications of Polynomial Integral Transform to Linear Ordinary Differential Equation with Constant Coefficients We apply the Polynomial Integral Transform to obtain the solutions of the ordinary differ- ential equations as follows: Example 1. d2 y d x2 − d y d x − 6y(x) = 0, y(0) = 0, y ′(0) = −7 Using the polynomial integral transform, we obtain B(y ′′(x)− y ′(x)− 6y(x)) =B(0) y(x) =B−1( −1 (s− 3) + 2 (s+ 2) ) y(x) =− e3x + 2e−2x . B. Barnes / Eur. J. Pure Appl. Math, 9 (2016), 140-151 149 Example 2. d2 y d x2 − 3 d y d x + 2y(x) = x , y(0) = 0, y ′(0) = 1 Applying polynomial integral transform to the above equation, we obtain B(y ′′(x)−3y ′(x) + 2y(x)) = B(x) y(x) =B−1( 3 4 . 1 s + 1 4 . 1 s2 − 2. 1 (s− 1) + 5 4 . 1 (s− 2) ) y(x) = 3 4 + 1 2 x − 2ex + 5 4 e2x 4.2. A Polynomial Integral Transform in Two Variables In order to obtain analytic solution of the partial differential equations PDEs, we extend the polynomial integral transform to solve the functions in two dimensions as below: Theorem 9. Let f be a function defined for x , t ≥ 1. Then the integral Bx Bt( f (x , t); (p, s)) = F(p, s) = ∫ ∞ 1 ∫ ∞ 1 f (ln x , ln t).x−p−1 t−s−1d xd t, is the integral transform of f (x , t) for x , t ∈ [1,∞), provided the integral converges. Proof. It follows from Theorem 1. We then apply Polynomial Integral Transform to transform partial derivatives. By the def- inition of the Polynomial Integral Transform in two variables, we obtain following results: Bx Bt( ∂ f (x , t) ∂ x ; (p, s)) =pF(p, s)− F(0, s) Bx Bt( ∂ 2 f (x , t) ∂ x∂ t ; (p, s)) =psF(p, s)− pF(p, 0)− sF(0, s) + f (0,0) Bx Bt( ∂ 2 f (x , t) ∂ x2 ; (p, s)) =p2F(p, s)− pF(0, s)− ∂ F(0, s) ∂ x Bx Bt( ∂ 2 f (x , t) ∂ t2 ; (p, s)) =s2F(p, s)− sF(p, 0)− ∂ f (p, 0) ∂ t We consider a wave equation below: ∂ 2w ∂ t2 = ∂ 2w ∂ x2 x , t ≥ 0 w(0, t) =g(t), lim x→∞ w(x , t) = 0, x , t ≥ 0 w(x , 0) = ∂ w ∂ t (x , 0) = 0, REFERENCES 150 where w is the deflection of a string released from rest on the x-axis. Applying the polynomial integral transform in two variables, we obtain B(wt t) =B(wx x) ⇒Wx x − s2W =0 W (x , s) =A(s)esx + B(s)e−sx W (x , s) =G(s)e−sx w(x , t) =B−1(W (x , s)) w(x , t) =xu(t − 1)g(t − 1), where g(t) = B−1(G(s)). 5. Conclusion We observed that the Polynomial Integral Transform solves differential equation with a few computations as well as time. Unlike the Laplace Integral Transform and others, the Polynomial Integral Transform involves a polynomial function as its kernel, which is easier and transforms complicated functions into algebraic equations. The solution of the differential equation is then obtained from the algebraic equation. Also, using the Polynomial Integral Transform, the convergence of the solution of the differential equation is faster as compared with the Laplace integral transform and others. We observed that the Polynomial Integral Transform is defined on the interval [1,∞). ACKNOWLEDGEMENTS The authors thank the readers of European Journal of Pure and Applied Mathematics, for making our journal successful. References [1] S. K. Q. Al-Omari. Notes for Hartley transforms of generalized functions. Italian Journal of pure and applied mathematics, 28:21–30, 2011. [2] S. K. Q. Al-Omari. On the applications of natural transforms. International journal of pure and applied mathematics, 85(4):729–744, 2013. [3] S. K. Q. Al-Omari and A. Kilicman. On diffraction Fresnel transform for Boehmians. Abstract and applied analysis, 2011, 2011. [4] M. A. Asiru. Further properties of the Sumudu transform and its applications. Inter- national journal of mathematical education in science and technology, 301(2):441–449, 2011. [5] A. Atangana and A. Kilicman. The use of Sumudu transform for solving certain nonlinear fractional heat-like equations. abstract and applied analysis, 2013, 2013. REFERENCES 151 [6] F. B. M. Belgacem and A. A. Karaballi. Sumudu transform fundamental properties inves- tigations and applications. Journal of applied mathematics and stochastic analysis, 2006, 2006. [7] F. B. M. Belgacem and R. Silambarasan. Advances in the natural transform. volume 1493. AIP conference proceedings, 2012. [8] W. E. Boyce and R. C. Diprima. Elementary differential equations and boundary value problems. John Wiley and Sons, Inc„ UK, 2001. [9] V. B. L. Chaurasia. Application of Sumudu transform in Schödinger equation occurring in quantum mechanics. Applied mathematical sciences,, 4(57):2843–2850, 2010. [10] H. Eltayeb and A. Kilicman. On double sumudu transform and double laplace transform. Malaysian journal of mathematical sciences, 4(1):17–30, 2010. [11] S. Handibag and B. D. Karande. Laplace substitution method for solving partial differ- ential equations involving mixed partial derivatives. International journal of pure and applied mathematics, 78(7):973–979, 2012. [12] Z. H. Khan and W. A. Khan. Natural transform-properties and applications. NUST journal of engineering sciences, 1(1):127–133, 2008. [13] F. Mainardi and G. Pagnini. Mellin-barnes integrals for stable distributions and their convolutions. Fractional calculus and applied mathematics: an international journal of theory and applications, 11(4), 2008. [14] J. J. Mohan and G. V. S. R. Deekshitulu. Solutions of fractional difference equations using s-transforms. Malaya journal of matematik, 3(1):1–13, 2013. [15] G. K. Watugala. Sumudu transform-a new integral transform to solve differential equa- tions and control engineering problems. Mathematical engineering in industry, 6(4):319– 329, 1993.