Processing math: 100%
Research article Special Issues

New reproducing kernel functions in the reproducing kernel Sobolev spaces

  • In this paper we construct some new reproducing kernel functions in the reproducing kernel Sobolev space. These functions are new in the literature. We can solve many problems by these functions in the reproducing kernel Sobolev spaces.

    Citation: Ali Akgül, Esra Karatas Akgül, Sahin Korhan. New reproducing kernel functions in the reproducing kernel Sobolev spaces[J]. AIMS Mathematics, 2020, 5(1): 482-496. doi: 10.3934/math.2020032

    Related Papers:

    [1] Li He . Composition operators on Hardy-Sobolev spaces with bounded reproducing kernels. AIMS Mathematics, 2023, 8(2): 2708-2719. doi: 10.3934/math.2023142
    [2] F. Z. Geng . Piecewise reproducing kernel-based symmetric collocation approach for linear stationary singularly perturbed problems. AIMS Mathematics, 2020, 5(6): 6020-6029. doi: 10.3934/math.2020385
    [3] Min Wu, Jiali Zhou, Chaoyue Guan, Jing Niu . A numerical method using Legendre polynomials for solving two-point interface problems. AIMS Mathematics, 2025, 10(4): 7891-7905. doi: 10.3934/math.2025362
    [4] Shiyv Wang, Xueqin Lv, Songyan He . The reproducing kernel method for nonlinear fourth-order BVPs. AIMS Mathematics, 2023, 8(11): 25371-25381. doi: 10.3934/math.20231294
    [5] Mohammad Reza Foroutan, Mir Sajjad Hashemi, Leila Gholizadeh, Ali Akgül, Fahd Jarad . A new application of the Legendre reproducing kernel method. AIMS Mathematics, 2022, 7(6): 10651-10670. doi: 10.3934/math.2022594
    [6] Xiuying Li, Yang Gao, Boying Wu . Approximate solutions of Atangana-Baleanu variable order fractional problems. AIMS Mathematics, 2020, 5(3): 2285-2294. doi: 10.3934/math.2020151
    [7] Siyu Tian, Boyu Liu, Wenyan Wang . A novel numerical scheme for reproducing kernel space of 2D fractional diffusion equations. AIMS Mathematics, 2023, 8(12): 29058-29072. doi: 10.3934/math.20231488
    [8] Fethi Bouzeffour . Inversion formulas for space-fractional Bessel heat diffusion through Tikhonov regularization. AIMS Mathematics, 2024, 9(8): 20826-20842. doi: 10.3934/math.20241013
    [9] Yassamine Chellouf, Banan Maayah, Shaher Momani, Ahmad Alawneh, Salam Alnabulsi . Numerical solution of fractional differential equations with temporal two-point BVPs using reproducing kernal Hilbert space method. AIMS Mathematics, 2021, 6(4): 3465-3485. doi: 10.3934/math.2021207
    [10] Boyu Liu, Wenyan Wang . An efficient numerical scheme in reproducing kernel space for space fractional partial differential equations. AIMS Mathematics, 2024, 9(11): 33286-33300. doi: 10.3934/math.20241588
  • In this paper we construct some new reproducing kernel functions in the reproducing kernel Sobolev space. These functions are new in the literature. We can solve many problems by these functions in the reproducing kernel Sobolev spaces.


    The reproducing-kernel Hilbert space construction associates a positive definite kernel with a Hilbert space of functions often referred to as the native space of the kernel. This construction can be used to deal with the problem of reconstructing an unknown function which lies in the reproducing kernel Hilbert space from a given multivariate data sample in an “optimal” way [1].

    One of the most investigated spaces in mathematical analysis are Sobolev spaces. The knowledge of the reproducing kernels is very useful in the analysis of many computational problems. It is usually enough to analyse reproducing kernels instead of the corresponding Hilbert spaces. It is therefore somehow surprising that it is difficult to find in the literature explicit formulas for the reproducing kernels of the Sobolev spaces [2].

    Reproducing kernel space is a special Hilbert space. There are many works on the solution of the nonlinear problems with reproducing kernel method [3]. The concept of reproducing kernel can be traced back to the paper of Zaremba [4] in 1908. It was given for investigating the boundary value problems of the harmonic functions. In the early development stage of the reproducing kernel theory, most of the works were implemented by Bergman [5]. Bergman has investigated the corresponding kernels of the harmonic functions with one or several variables, and the corresponding kernel of the analytic function in squared metric, and implemented them in the research of the boundary value problem of the elliptic partial differential equation.

    Zorzi et al. [6] have investigated the harmonic analysis of kernel functions and the sparse plus low rank network identification [7]. The empirical Bayesian learning in AR graphical models has been studied in [8].

    The reproducing kernel Sobolev space method (RKSSM), which computes the numerical solution, is of great attention to many branches of applied sciences. Many studies have been dedicated to the application of the RKSSM [1,9]. For more details see [9,10,11,12,13,14,15].

    We organize the paper as: We present the main definitions in Section 2. We construct some important reproducing kernel Sobolev spaces in Section 3. We obtain very reproducing kernel functions in this section. We give the conclusion in Section 4.

    Definition 1. We define the reproducing kernel Sobolev space Am2[a,b] as [3]:

    Am2[a,b]={f|f(m1)is absolutely continuous function,f(m)L2[a,b],x[a,b]},

    We define the inner product and norm for this space as:

    f,gAm2=ba(mi=0f(i)(x)g(i)(x))dx

    and

    fAm2=f,fWm2.

    Definition 2. We have the reproducing property as [3]:

    f,RyAm2=f(y).

    We obtain very useful reproducing kernel functions in the reproducing kernel Sobolev spaces in this section. These kernel functions are new in the literature.

    We have the inner product as:

    u,RyS11[0,1]=10(u(x)Ry(x)+u(x)Ry(x))dx

    We use integration by parts and obtain:

    u,RyS11[0,1]=10u(x)Ry(x)dx+u(1)Ry(1)u(0)Ry(0)10u(x)Ry(x)dx=u(1)Ry(1)u(0)Ry(0)10u(x)(Ry(x)Ry(x))dx.

    If we choose

    1)Ry(0)=0,2)Ry(1)=0,

    then, we obtain

    Ry(x)Ry(x)=δ(xy).

    When xy, we get

    Ry(x)Ry(x)=0.
    λ21=0λ=1.

    Thus, we reach

    Ry(x)={c1ex+c2ex,xy,d1ex+d2ex,x>y.

    By the property of the Dirac-Delta function, we get

    3)Ry+(y)=Ry(y),4)Ry+(y)Ry(y)=1.

    We have

    Ry(x)={c1exc2ex,x<yd1exd2ex,x>y.
    Ry(0)=c1c2=0c1=c2
    Ry(1)=d1ed2e=0d2=e2d1
    Ry+(y)=Ry(y)d1ey+d2ey=c1ey+c2ey
    Ry+(y)Ry(y)=1d1eyd2eyc1ey+c2ey=1

    when we solve these equations, we get:

    c1=12e1y+ey1(ee1),
    c2=12e1y+ey1(ee1),
    d1=12ey1+ey1(ee1),
    d2=12e1+y+e1y(ee1).

    Thus, we obtain the reproducing kernel function as:

    Ry(x)={12(e1y+e1+y)(e1+x+e1x)e21,xy12(e1+y+e1y)(e1+x+e1x)e21,x>y.

    For simplify, we define

    a(x)=12(e1y+e1+y)(e1+x+e1x)e21,
    b(x)=12(e1+y+e1y)(e1+x+e1x)e21.

    We need to show

    u,RyS12=u(y).

    Therefore, we get

    u,RyS12=y0u(x)Ry(x)dx+1yu(x)Ry(x)dx+y0u(x)Ry(x)dx+1yu(x)Ry(x)dx.
    u,RyS12=y0u(x)a(x)dx+1yu(x)b(x)dx+y0u(x)a(x)dx+1yu(x)b(x)dx=y0u(x)12(e1y+e1+y)e21[e1+x+e1x]dx+1yu(x)12(e1+y+e1y)e21[e1+x+e1x]dx+y0u(x)2(e1y+e1+y)e21[e1+xe1x]dx+1yu(x)2(e1+y+e1y)e21[e1+xe1x]dx.
    u,RyS12=e1+y+e1y2(e21)[1y(u(x)(ex1+e1x)+u(x)(e1+xe1x))dx]+12e1y+ey1(e21)[y0(u(x)(ex+1+e1x)+u(x)(e1+xe1x))dx].

    We use integration by parts and obtain

    u,RyS12=12e1+y+e1y2(e21)[1yu(x)(ex1+e1x)dx+(e0e0)u(1)(ey1e1y)u(y)1yu(x)(ex1+e1x)dx]+12(e1y+e1+y)(e21)[y0u(x)(ex+1+e1x)dx+u(y)(e1+ye1y)u(0)(e1e1)y0u(x)(ex+1+e1x)dx]
    u,RyS12=12e1+y+e1y(e21)u(y)(e1yey1)+12e1y+ey1(e21)u(y)(e1+ye1y)=12u(y)(e21)[(e1yey1)(e1+y+e1y)+(e1+ye1y)(e1y+ey1)]
    u,RyS12=12(e21)u(y)[e2+e22ye2y1+e2+e2ye22y1]=u(y)2(e21)(2e22)=u(y).

    This completes the proof.

    We have

    u,RyS22=10[u(x)Ry(x)+u(x)Ry(x)+u(x)Ry(x)]dx=10u(x)Ry(x)dx+10u(x)Ry(x)dx+10u(x)Ry(x)dx

    After integration by parts, we get

    u,RyS22=10u(x)Ry(x)dx+u(1)Ry(1)u(0)Ry(0)10u(x)Ry(x)dx+u(1)Ry(1)u(0)Ry(0)u(1)Ry(1)+u(0)Ry(0)+10u(x)R(4)y(x)dx.

    Then, we obtain

    u,RyS22=u(1)Ry(1)u(0)Ry(0)+u(1)Ry(1)u(0)Ry(0)u(1)Ry(1)+u(0)Ry(0)+10u(x)[Ry(x)Ry(x)+R(4)y(x)]dx.

    We have

    1)Ry(1)Ry(1)=0,2)Ry(0)+Ry(0)=0,3)Ry(1)=0,4)Ry(0)=0.

    Therefore, we get

    u,RyS22=10u(x)[Ry(x)Ry(x)+R(4)y(x)]dx.

    By reproducing property, we have

    Ry(x)Ry(x)+R(4)y(x)=δ(xy).

    When xy, we get

    Ry(x)Ry(x)+R(4)y(x)=0.

    Then, we have

    1λ2+λ4=0.
    λ1=3212I,λ2=32+12I,λ3=32+12I,λ4=3212I.
    Ry(x)={c1e3x2cos(x2)+c2e3x2sin(x2)+c3e3x2cos(x2)+c4e3x2sin(x2),xyd1e3x2cos(x2)+d2e3x2sin(x2)+d3e3x2cos(x2)+d4e3x2sin(x2),x>y

    By properties of the Dirac-Delta function, we have

    5)Ry+(y)=Ry(y),6)Ry+(y)=Ry(y),7)Ry+(y)=Ry(y),8)Ry+(y)Ry(y)=1.

    Solving these eight equations give,

    c1=16(4sin(12y)e123ye12338cos(12y)e123ye1234cos(12y)e123ye123)3(4sin(12y)2e1234sin(12y)2e123+4cos(12y)2e1234cos(12y)2e123)
    c2=16(4cos(12y)e123ye1233+8sin(12y)e123ye123+4sin(12y)e123ye123)3(sin(12y)2+cos(12y)2)(4e1234e123)
    c3=16(4sin(12y)e123ye12334cos(12y)e123ye1238cos(12y)e123ye123)3(sin(12y)2+cos(12y)2)(4e1234e123)
    c4=16(4cos(12y)e123ye1233+4sin(12y)e123ye123+8sin(12y)e123ye123)3(sin(12y)2+cos(12y)2)(4e1234e123)
    d1=16(4sin(12y)e123ye12338cos(12y)e123ye1234cos(12y)e123ye123)3(sin(12y)2+cos(12y)2)(4e1234e123)
    d2=16(4cos(12y)e123ye1233+8sin(12y)e123ye123+4sin(12y)e123ye123)3(sin(12y)2+cos(12y)2)(4e1234e123)
    d3=16(4sin(12y)e123ye12334cos(12y)e123ye1238cos(12y)e123ye123)3(sin(12y)2+cos(12y)2)(4e1234e123)
    d4=16(4cos(12y)e123ye1233+4sin(12y)e123ye123+8sin(12y)e123ye123)3(sin(12y)2+cos(12y)2)(4e1234e123)

    Then we find the reproducing kernel function for xy as:

    Ry(x)=161e31e123(cos(12y)e12(xy+1)3cos(12x)3+2cos(12y)e12(x+y1)3cos(12x)3+2cos(12y)e12(x+y1)3cos(12x)3+cos(12y)e12(xy+1)3cos(12x)3+sin(12y)e12(xy+1)3sin(12x)3+2sin(12y)e12(x+y1)3sin(12x)3+2sin(12y)e12(x+y1)3sin(12x)3+sin(12y)e12(xy+1)3sin(12x)33cos(12y)e12(xy+1)3sin(12x)+3cos(12y)e12(xy+1)3sin(12x)+3sin(12y)e12(xy+1)3cos(12x)3sin(12y)e12(xy+1)3cos(12x)).

    The reproducing kernel function for x>y is obtained as:

    Ry(x)=161e31e123(cos(12y)e12(xy1)3cos(12x)3+2cos(12y)e12(x+y1)3cos(12x)3+2cos(12y)e12(x+y1)3cos(12x)3+cos(12y)e12(xy1)3cos(12x)3+sin(12y)e12(xy1)3sin(12x)3+2sin(12y)e12(x+y1)3sin(12x)3+2sin(12y)e12(x+y1)3sin(12x)3+sin(12y)e12(xy1)3sin(12x)33cos(12y)e12(xy1)3sin(12x)+3cos(12y)e12(xy1)3sin(12x)+3sin(12y)e12(xy1)3cos(12x)3sin(12y)e12(xy1)3cos(12x)).

    We have

    u,ByS32[0,1]=10u(x)By(x)dx+10u(x)By(x)dx+10u(x)By(x)dx+10u(x)By(x)dx.

    After using integration by parts, we obtain

    u,ByS32[0,1]=10u(x)By(x)dx+u(1)By(1)u(0)By(0)10u(x)By(x)dx+u(1)By(1)u(0)By(0)u(1)By(1)+u(0)By(0)+10u(x)B(4)y(x)dx+u(1)By(1)u(0)By(0)u(1)B(4)y(1)+u(0)B(4)y(0)+u(1)B(5)y(1)u(0)B(5)y(0)10u(x)B(6)y(x)dx.

    Then, we get

    u,ByS32[0,1]=10u(x)[By(x)By(x)+B(4)y(x)B(6)y(x)]dx.

    If we have the following equations:

    1)By(0)+By(0)B(5)y(0)=0,2)By(0)+B(4)y(0)=0,3)By(0)=0,4)By(1)By(1)+B(5)y(1)=0,5)By(1)B(4)y(1)=0,6)By(1)=0.

    Then by reproducing property, we will get

    u,ByS32[0,1]=10u(x)[By(x)By(x)+B(4)y(x)B(6)y(x)]dx=u(y).

    Therefore, by Dirac-Delta function we get

    By(x)By(x)+B(4)y(x)B(6)y(x)=δ(xy).

    When xy, we have δ(xy)=0. Thus, we reach

    By(x)By(x)+B(4)y(x)B(6)y(x)=0.

    Then, we get

    1λ2+λ4λ6=0.

    If we solve the above equation, we will get

    λ1=1,λ2=1,λ3=22+i22,λ4=22i22,λ5=22i22,λ6=22+i22.

    Therefore, we obtain the reproducing kernel function By(x) as:

    By(x)={c1ex+c2ex+c3e22xcos(22x)+c4e22xsin(22x)+c5e22xcos(22x)+c6e22xsin(22x),xyd1ex+d2ex+d3e22xcos(22x)+d4e22xsin(22x)+d5e22xcos(22x)+d6e22xsin(22x),x>y.

    If we solve the above equations, we will get the coefficients as:

    c1=14eye+e1eyeyey(ee1),
    c2=14eye+e1eyeyey(ee1),
    c3=14(sin(12y2)e12y2e122+cos(12y2)e122e12y2)2(sin(12y2)2+cos(12y2)2)(e122e122)e12y2e12y2
    c4=14(cos(12y2)e12y2e122sin(12y2)e122e12y2)2(sin(12y2)2+cos(12y2)2)(e122e122)e12y2e12y2
    c5=14(cos(12y2)e12y2e122sin(12y2)e122e12y2)2(sin(12y2)2+cos(12y2)2)(e122e122)e12y2e12y2
    c6=14(sin(12y2)e12y2e122+cos(12y2)e122e12y2)2(sin(12y2)2+cos(12y2)2)(e122e122)e12y2e12y2
    d1=14e1(ey+ey)eyey(ee1),
    d2=14e(ey+ey)eyey(ee1),
    d3=14(cos(12y2)e12y2e122+sin(12y2)e122e12y2)2(sin(12y2)2+cos(12y2)2)(e122e122)e12y2e12y2
    d4=14(sin(12y2)e12y2e122cos(12y2)e122e12y2)2(sin(12y2)2+cos(12y2)2)(e122e122)e12y2e12y2
    d5=14(sin(12y2)e12y2e122cos(12y2)e122e12y2)2(sin(12y2)2+cos(12y2)2)(e122e122)e12y2e12y2
    d6=14(cos(12y2)e12y2e122+sin(12y2)e122e12y2)2(sin(12y2)2+cos(12y2)2)(e122e122)e12y2e12y2

    Then, we will get the reproducing kernel function By(x) for xy as:

    By(x)=141(e21)(e21)e122+1(e112y2+12x2+122sin(12y2)cos(12x2)2+e112y212x2+122sin(12y2)sin(12x2)2+e1+12y2+12x2122sin(12y2)sin(12x2)2e1+12y212x2122sin(12y2)cos(12x2)2e112y2+12x2+122cos(12y2)sin(12x2)2+e112y212x2+122cos(12y2)cos(12x2)2+e1+12y2+12x2122cos(12y2)cos(12x2)2+e1+12y212x2122cos(12y2)sin(12x2)2e112y2+12x2+122sin(12y2)cos(12x2)2e112y212x2+122sin(12y2)sin(12x2)2e1+12y2+12x2122sin(12y2)sin(12x2)2+e1+12y212x2122sin(12y2)cos(12x2)2+e112y2+12x2+122cos(12y2)sin(12x2)2e112y212x2+122cos(12y2)cos(12x2)2e1+12y2+12x2122cos(12y2)cos(12x2)2e1+12y212x2122cos(12y2)sin(12x2)2e1y+x122+e1y+x+122e1yx122+e1yx+122e1+y+x122+e1+y+x+122e1+yx122+e1+yx+122).

    We obtain the reproducing kernel function By(x) for x>y as:

    By(x)=141(e21)(e21)e122+1(e1+12y212x2+122sin(12y2)cos(12x2)2+e112y2+12x2122sin(12y2)cos(12x2)2e1+12y212x2+122sin(12y2)cos(12x2)2e112y2+12x2122sin(12y2)cos(12x2)2e112y212x2+122sin(12y2)sin(12x2)2e1+12y2+12x2122sin(12y2)sin(12x2)2+e112y212x2+122sin(12y2)sin(12x2)2+e1+12y2+12x2122sin(12y2)sin(12x2)2e112y212x2+122cos(12y2)cos(12x2)2e1+12y2+12x2122cos(12y2)cos(12x2)2+e112y212x2+122cos(12y2)cos(12x2)2+e1+12y2+12x2122cos(12y2)cos(12x2)2e1+12y212x2+122cos(12y2)sin(12x2)2e112y2+12x2122cos(12y2)sin(12x2)2+e1+12y212x2+122cos(12y2)sin(12x2)2+e112y2+12x2122cos(12y2)sin(12x2)2+e1+yx122e1+yx+122+e1yx122e1yx+122+e1+y+x122e1+y+x+122+e1y+x122e1y+x+122).

    In this paper, we defined very useful reproducing kernel Sobolev spaces. We found very important reproducing kernel functions in these spaces. These kernel functions are very useful to solve many problems in the reproducing kernel Sobolev spaces.

    This work is supported by 2018-SIÜEǦT-044.

    The authors declare that there is no conflict of interest in this paper.



    [1] E. F. Gregory, Y. Qi, Reproducing kernels of Sobolev spaces via a green kernel approach with differential operators and boundary operators, Adv. Comput. Math., 38 (2013), 891-921. doi: 10.1007/s10444-011-9264-6
    [2] M. G. Sakar, Iterative reproducing kernel Hilbert spaces method for Riccati differential equations, J. Comput. Appl. Math., 309 (2017), 163-174. doi: 10.1016/j.cam.2016.06.029
    [3] M. Cui, Y. Lin, Nonlinear Numerical Analysis in the Reproducing Kernel Space, Nova Science Publishers Inc, New York, 2009.
    [4] S. Zaremba, Sur le calcul numérique des fonctions demandées dan le probléme de dirichlet et le probleme hydrodynamique, Bulletin International l'Académia des Sciences de Cracovie, 68 (1908), 125-195.
    [5] S. Bergman, The Kernel Function and Conformal Mapping, American Mathematical Society, New York, 1950.
    [6] M. Zorzi, A. Chiuso, The harmonic analysis of kernel functions, Automatica, 94 (2018), 125-137. doi: 10.1016/j.automatica.2018.04.015
    [7] M. Zorzi, A. Chiuso, Sparse plus low rank network identification: A nonparametric approach, Automatica, 76 (2017), 355-366. doi: 10.1016/j.automatica.2016.08.014
    [8] M. Zorzi, Empirical Bayesian learning in AR graphical models, Automatica, 109 (2019), 108516.
    [9] E. Novak, Reproducing Kernels of Sobolev Spaces on Rd and Applications to Embedding Constants and Tractability, Arxiv.
    [10] O. A. Arqub, Fitted reproducing kernel Hilbert space method for the solutions of some certain classes of time-fractional partial differential equations subject to initial and Neumann boundary conditions, Comput. Math. Appl., 73 (2017), 1243-1261. doi: 10.1016/j.camwa.2016.11.032
    [11] S. Momani, O. A. Arqub, T. Hayat, et al. A computational method for solving periodic boundary value problems for integro-differential equations of Fredholm-Voltera type, Appl. Math. Comput., 240 (2014), 229-239.
    [12] B. Azarnavid, K. Parand, An iterative reproducing kernel method in Hilbert space for the multipoint boundary value problems, J. Comput. Appl. Math., 328 (2018), 151-163. doi: 10.1016/j.cam.2017.07.015
    [13] A. Akgül, New reproducing kernel functions, Math. Probl. Eng., 2015 (2015), 1-10.
    [14] A. Akgül, On solutions of variable-order fractional differential equations, An International Journal of Optimization and Control: Theories & Applications (IJOCTA), 7 (2017), 112-116.
    [15] A. Akgül, E. K. Akgül, D. Baleanu, Numerical solutions of fractional differential equations of Lane-Emden type by an accurate technique, Adv. Differ. Equ., 2015 (2015), 220.
  • This article has been cited by:

    1. F. Z. Geng, Piecewise reproducing kernel-based symmetric collocation approach for linear stationary singularly perturbed problems, 2020, 5, 2473-6988, 6020, 10.3934/math.2020385
    2. Xiuying Li, Yang Gao, Boying Wu, Approximate solutions of Atangana-Baleanu variable order fractional problems, 2020, 5, 2473-6988, 2285, 10.3934/math.2020151
    3. Jim Agler, Zinaida A. Lykova, N.J. Young, A Hilbert space approach to singularities of functions, 2023, 284, 00221236, 109826, 10.1016/j.jfa.2022.109826
    4. Zhi-Yuan Li, Mei-Chun Wang, Yu-Lan Wang, Solving a class of variable order nonlinear fractional integral differential equations by using reproducing kernel function, 2022, 7, 2473-6988, 12935, 10.3934/math.2022716
    5. Tianyu Zhang, Noah Simon, Regression in tensor product spaces by the method of sieves, 2023, 17, 1935-7524, 10.1214/23-EJS2188
  • Reader Comments
  • © 2020 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)
通讯作者: 陈斌, bchen63@163.com
  • 1. 

    沈阳化工大学材料科学与工程学院 沈阳 110142

  1. 本站搜索
  2. 百度学术搜索
  3. 万方数据库搜索
  4. CNKI搜索

Metrics

Article views(4915) PDF downloads(579) Cited by(5)

Other Articles By Authors

/

DownLoad:  Full-Size Img  PowerPoint
Return
Return

Catalog