Skip to main content

Uniformly convergent extended cubic B-spline collocation method for two parameters singularly perturbed time-delayed convection-diffusion problems


This work proposes a uniformly convergent numerical scheme to solve singularly perturbed parabolic problems of large time delay with two small parameters. The approach uses implicit Euler and the exponentially fitted extended cubic B-spline for time and space derivatives respectively. Extended cubic B-splines have advantages over classical B-splines. This is because for a given value of the free parameter \( \lambda \) the solution obtained by the extended B-spline is better than the solution obtained by the classical B-spline. To confirm the correspondence of the numerical methods with the theoretical results, numerical examples are presented. The present numerical technique converges uniformly, leading to the current study of being more efficient.

Peer Review reports


Consider the two-parameter singularly perturbed one-dimensional parabolic time delay convection-diffusion initial-boundary value problem defined as

$$\begin{aligned} \left\{ \begin{aligned}&\left( \frac{\partial }{\partial t} +L_{\varepsilon ,\mu }\right) u(x,t) =H(x,t), (x,t)\in {D} \\&u(x,t)={\phi _{b}(x,t)}, \; (x,t)\in \Gamma _{b}={{[0,1]}\times {[-\tau ,0]}},\\&u(0,t)={\phi _{l}(t)}, \; {\Gamma _{l}}=\lbrace {(0,t):{{0}\le {t}\le {T}}}\rbrace ,\\&u(1,t)={\phi _{r}(t)}, \; {\Gamma _{r}}=\lbrace {(1,t):{{0}\le {t}\le {T}}}\rbrace . \end{aligned} \right. \end{aligned}$$

where \(D = \Omega _{x}\times {(0, T]}\), \(\Omega _{x}= (0,1)\), \({0 <\varepsilon \le 1}, 0 \le \mu \le 1\), \(H(x,t)=c(x,t)u(x,t-\tau )+f(x,t)\) and \({\tau >0}\) represents the delay parameter and a(xt), b(xt), c(xt), f(xt), \(\phi _{b}(x,t), \phi _{l}(t)\) and \(\phi _{r}(t)\) are sufficiently smooth, bounded functions on \( \overline{D} =\left[ 0,1\right] \times \left[ 0,T\right] ,\) that satisfy

$$\begin{aligned} {a(x,t)}\ge {\alpha }> {0}, {b(x,t)}\ge {\beta }> {0}, {c(x,t)}\ge {\vartheta }> {0}, \gamma =\min _{{\bar{D}}}\left( \frac{b}{a} \right) . \end{aligned}$$

The operator \( L_{\varepsilon ,\mu } \) given as

$$\begin{aligned} L_{\varepsilon ,\mu }{u(x,t)}\equiv -\varepsilon u_{xx}-\mu a(x,t)u_{x}+b(x,t)u. \end{aligned}$$

The existence of function approximations has been the subject of extensive research [1,2,3,4,5,6,7,8,9,10,11,12,13,14]. Here, the existence and uniqueness of a solution of (1) can be established under the assumption that the data are Holder continuous and sufficient smoothness of initial-boundary data on \( \Gamma =\Gamma _{b}\cup \Gamma _{l}\cup \Gamma _{r} \) and compatibility conditions at the corner points \((0,0),(1,0),(0,-\tau )\) and \((1,-\tau )\), and delay terms [15].

$$\begin{aligned} \left\{ \begin{aligned}&\phi _{b}{(0,0)}=\phi _{l}{(0)},\\&{\phi _{b}{(1,0)}={\phi _{r}{(0)}}}, \end{aligned} \right. \end{aligned}$$

so that a unique solution exists and is sufficiently smooth for the model problem (1). For \(\varepsilon \rightarrow 0 \) and \( \mu =1,\) numerical methods available in [16, 17] for the problem given by Eqs.(1) whose solution exhibits an exponential boundary layer of width \( O\left( \varepsilon \right) \) in the left boundary layer \(\Gamma _{l}\). As the parameters \( \varepsilon \rightarrow 0\) and \( \mu \rightarrow 0\), the solution develops boundary layers at \( x = 0 \) and \( x = 1 \). The parameters and the ratio \( \varepsilon /\mu ^{2} \) affect the boundary layer’s width. We look at Eq. (1) above with \( \varepsilon /\mu ^{2} \rightarrow 0\) as \( \mu \rightarrow 0 \) and \( \mu ^{2}/\varepsilon \rightarrow 0 \) as \( \varepsilon \rightarrow 0 \). As a result, the uniformly convergent numerical treatment presented in this study is independent of the two parameters \( \varepsilon \) and \( \mu \).

Two-parameter time delayed singularly perturbed parabolic problems have not been studied as extensively as one-parameter problems. Such type of problems are widespread in many phenomena of real life problems (see, for example, [18,19,20]) described by boundary layer problems. For singularly perturbed one-parameter partial differential equations many works have been delivered numerically in recent years (see, for example, [16, 21,22,23,24,25,26,27,28,29,30,31,32,33]). Not much numerical investigations have been done on two-parameter time delayed singularly perturbed parabolic problems. The work on two-parameter time delayed singularly perturbed parabolic problems have been started by Govindarao et al. [34], where they considered an upwind difference scheme on the Shishkin type meshes. First-order in both space and time numerical method was established. Sumit et al. [35] extend the works, where they considered a hybrid scheme for space consisting of central difference, upwind and midpoint operators on layer adapted piecewise uniform Shishkin mesh. Almost second-order in space and first order in time numerical method was established. Negero [36,37,38,39,40] also considered the problem similar to Sumit et al. and proposed numerical methods based on fitted operator methods on a uniform mesh, which improved the rate of convergence. However, for the problem under study, there are no known fitted extended cubic B-spline numerical methods. Here, the paper focus on exponentially fitted extended cubic B-spline for spatial discretization and the implicit Euler method for time discretization on uniform meshes. This is the more accurate compared to existing methods for the problem addressed in this work.

The paper is arranged as follows. Section "Preliminaries" presents the bounds on the derivatives and exact solution of Eq. (1). The discrete scheme are discussed in Section "Discretization of the problem". Section "Convergence analysis" deals with convergence and stability of the proposed numerical scheme. Numerical results are given in Section "Numerical examples and results" to illustrate the theory. The paper concludes with a discussion of the results obtained.

Notations: In this paper, we denote a generic positive constant by C, independent of mesh parameters \( \mu \) and \( \varepsilon \). The supremum norm on a domain D is defined as

$$\begin{aligned} \Vert \hslash \Vert _{\bar{D}}=\sup _{\left( x,t\right) \in \bar{D}}|\hslash (x,t)|. \end{aligned}$$


Lemma 1

(Continuous maximum principle) Let z(xt) \(\in C^{2}\left( D\right) \cap C^{0}\left( \bar{D}\right) \), and assume that \(z(x,t)\ge 0\), \(\forall (x,t)\in {\Gamma } =\Gamma _{l}\cup \Gamma _{b}\cup \Gamma _{r}\). Then \( \left( \frac{\partial }{\partial t} +L_{\varepsilon ,\mu }\right) z(x,t) \ge 0 \) in D implies that \(z(x,t)\ge 0\), \(\forall (x,t)\in \bar{D}\).

Proof Let \(\left( \zeta ^{*},\nu ^{*}\right) \in D\) such that \(z\left( \zeta ^{*},\nu ^{*}\right) ={\min _{(x,t)\in \bar{D}}z\left( x,t \right) }<0.\) Then \(\left( \xi ^{*},\vartheta ^{*} \right) \notin \Gamma \). Since at the point \(\left( \xi ^{*},\vartheta ^{*} \right) \) function \(\pi \) attains minimum, then, we have \(z_{x}=z_{t}=0\) at \(\left( \zeta ^{*},\nu ^{*}\right) \) and \(z_{xx}\left( \zeta ^{*},\nu ^{*} \right) \ge 0\) and thus,

$$\begin{aligned}&\left( \frac{\partial }{\partial t} +L_{\varepsilon ,\mu }\right) z(\zeta ^{*},\nu ^{*}) =\frac{\partial z\left( \zeta ^{*},\nu ^{*}\right) }{\partial t} -\varepsilon \frac{\partial ^{2} z\left( \zeta ^{*},\nu ^{*}\right) }{\partial x^{2}}\\&\quad -\mu {a(\zeta ^{*},\nu ^{*})}\frac{\partial z\left( \zeta ^{*},\nu ^{*}\right) }{\partial x}+ {b(x,t)}z{(\zeta ^{*},\nu ^{*})}<0, \end{aligned}$$

which is a contradiction. This implies \(z(x,t)\ge 0\) \(\forall \) \((x,t)\in \bar{D}\). \(\square \)

Lemma 2

[35] Let u(xt) be the solution of problems (1) and i, j are any non-negative integers satisfying \(0 \le i +3j \le 4\). Then,

$$\begin{aligned} \left\| \dfrac{\partial ^{i+j}u}{\partial x^{i}\partial t^{j}}\right\| _{\bar{D}} \le C\left\{ \begin{aligned}&\frac{1}{\left( \sqrt{\varepsilon }\right) ^{i}}, \text {if}~ \frac{{\mu }^{2}}{\varepsilon }\rightarrow 0 ~ \text {as}~ \varepsilon \rightarrow 0,\\&\left( \frac{\mu }{\varepsilon }\right) ^{i}\left( \frac{{\mu }^{2}}{\varepsilon }\right) ^{j}, \text {if}~\frac{{\varepsilon }}{\mu ^{2}} \rightarrow 0 ~ \text {as}~ \mu \rightarrow 0,\\ \end{aligned} \right. \end{aligned}$$

where C a positive constant independent of the parameters \( \varepsilon \) and \( \mu \).

Discretization of the problem

The time semi-discretization

For the time domain [0, T] equidistant mesh discretization with uniform step size \(\Delta t\) is used such that

$$\begin{aligned} {\varOmega }^{M}_{t}=\left\{ t_{m}=m\Delta t, m=0,1,...,M, \Delta t =T/M \right\} , \end{aligned}$$

where M is mesh elements used on the interval [0, T]. The mesh for \(\left[ -\tau ,T\right] \) is defined as

$$\begin{aligned} {\varOmega }^{s}_{t}=\left\{ t_{m}=m\Delta t, m=0,1,...,s,t_{s}=\tau , \Delta t =\tau /s \right\} . \end{aligned}$$

where s mesh elements used on the interval \(\left[ -\tau , 0\right] \). Here, semi-discretizing the given problems (1) by applying implicit Euler scheme written as

$$\begin{aligned} \left\{ \begin{aligned}&\frac{U^{m}\left( x\right) -U^{m-1}\left( x\right) }{\Delta t}-\varepsilon \left( U_{xx}\right) ^{m}\left( x\right) -\mu a^{m}\left( x\right) \left( U_{x}\right) ^{m}\left( x\right) +b^{m}\left( x\right) U^{m}\left( x\right) \\&\quad =H^{m}\left( x\right) ,\\&U^{m}\left( 0\right) =\phi _{l}\left( t_{m}\right) , 0\le m\le M, x\in \Omega _{x}, \\&U^{m}\left( 1\right) =\phi _{r}\left( t_{m}\right) , 0\le m\le M,x\in \Omega _{x}, \\&U^{m}\left( x\right) =\phi _{b}\left( x,t_{m}\right) , -s\le m\le -1, x\in \Omega _{x}, \end{aligned} \right. \end{aligned}$$

where \(H^{m}\left( x\right) =-c^{m}(x){ {U}^{m-s}\left( x\right) }+f^{m}\left( x\right) \), \(0\le m\le M,x\in \Omega _{x}\) and \(U^{m}(x)\) is the approximate solution of \(u(x,t_{m})\) at (m)th time level. The Eq. (3) can be rewritten as

$$\begin{aligned} \left\{ \begin{aligned}&\left( 1+\Delta t L_{\varepsilon ,\mu }^{\Delta t}\right) U^{m}(x)=H(x,t_{m}),\\&U^{m}(0)={\phi _{l}(t_{m})},\;m=0,...,M,\\&U^{m}(1)={\phi _{r}(t_{m})},\;m=0,...,M, \\&U^{m}(x)={\phi _{b}(x,t_{m})},\; x\in \left( 0,1\right) , -\left( s+1 \right) \le m \le -1, \end{aligned} \right. \end{aligned}$$


$$\begin{aligned}&L_{\varepsilon ,\mu }^{\Delta t}=-\varepsilon \left( U_{xx}\right) ^{m} \left( x\right) -\mu a^{m}\left( x\right) \left( U_{x}\right) ^{m}\left( x\right) + b^{m}\left( x \right) U^{m}\left( x\right) \\&H\left( x,t_{m}\right) = -\Delta t c^{m}(x){{U}^{m-s}\left( x\right) }+\Delta tf^{m}\left( x\right) + U^{m}\left( x\right) . \end{aligned}$$

Lemma 3

(Semi-discrete maximum principle) Assume that \(\Pi ^{m+1}\left( x\right) \in C^{2,1}\left( \bar{D}\right) \) such that \(\Pi ^{m+1}\left( 0\right) \ge 0\) and \(\Pi ^{m+1}\left( 1\right) \ge 0\). Then, \(\left( 1+\Delta t {\pounds }_{\varepsilon ,\mu }^{\Delta t}\right) \Pi ^{m+1}\left( x\right) \ge 0 \), \(\forall x\in D\), implies that \(\Pi ^{m+1}(x)\ge 0\), \(\forall x\in \bar{D}\).


Assume \(y^{*} \in {\bar{D}}\) such that \(\Pi ^{m+1}\left( y^{*} \right) ={\min _{(x)\in {\bar{D}}}\Pi ^{m+1}\left( x\right) }\) and suppose \(\Pi ^{m+1}\left( y^{*} \right) <0\). Now, it is clear that \(y^{*}\notin \left\{ 0,1\right\} \), which implies that \(y^{*} \in \left( 0,1\right) \). Therefore, we have \(\frac{d}{dx}\left( \Pi ^{m+1}\left( y^{*}\right) \right) =0\) and \(\frac{d^{2}}{dx^{2}}\left( \Pi \left( y^{*}\right) \right) \ge 0\) and thus

$$\begin{aligned}&\left( 1+\Delta t {\pounds }_{\varepsilon ,\mu }^{\Delta t}\right) \Pi ^{m+1}\left( y^{*} \right) =-{\varepsilon }\Delta t \frac{d^{2}}{dx^{2}}\left( \Pi ^{m+1}\left( y^{*}\right) \right) \\&\quad -\mu \Delta t{a^{m+1}\left( y^{*}\right) }\frac{d}{dx} \Pi ^{m+1}\left( y^{*}\right) +\left( 1+\Delta t \right) \Pi ^{m+1}\left( y^{*}\right) <0, \end{aligned}$$

this contradicts assumption and \(\Pi ^{m+1}(y^{*})\ge 0\), which implies that \(\Pi ^{m+1}(x)\ge 0,\) \(\forall (x)\in \bar{D}\). \(\square \)

Let \(u(x,t_{m})\) be the exact and \(U^{m}(x)\) be the approximate solution of the problem in (1). The error estimates for the temporal semi-discretization (4) \( E_{m+1}=U^{m}(x)-u\left( x,t_{m}\right) \) satisfy the following Lemma.

Lemma 4

(Local error estimate) The local error estimate with the semi-discretized problem (4) is given by

$$\begin{aligned} \left\| E_{m+1}\right\| _{\infty } \le C\left( \Delta t \right) ^{2}. \end{aligned}$$


Applying Taylor’s series expansion to \(u\left( x,t_{m}\right) \) gives,

$$\begin{aligned} u(x,t_{m+1})=u(x,t_{m})+\Delta t u_{t}\left( x,t_{m} \right) +O\left( \left( \Delta t \right) ^{2} \right) . \end{aligned}$$

Substituting (5) into the continuous problems (1) gives,

$$\begin{aligned} \frac{u(x,t_{m+1})-u(x,t_{m})}{\Delta t}&=u_{t}\left( x,t_{m} \right) +O\left( \left( \Delta t \right) ^{2} \right) \\&= \varepsilon u_{xx}\left( x,t_{m+1} \right) +\mu a(x,t_{m+1})u_{x}(x,t_{m+1})\\&-b(x,t_{m+1})u\left( x,t_{m+1}\right) -c(x,t_{m+1})u\left( x,t_{-s+m}\right) +f(x,t_{m+1})\\&+O\left( \left( \Delta t \right) ^{2}\right) . \end{aligned}$$

Clearly \(E_{m+1}(x)\) satisfies the semi-discrete operator

$$\begin{aligned} \left( 1+ \Delta t{L}_{\varepsilon ,\mu }^{\Delta t}\right) {E}_{m+1}(x)=O\left( \left( \Delta t \right) ^{2}\right) , \end{aligned}$$

with the conditions:

$$\begin{aligned} E_{m+1}(0)=E_{m+1}(1)=0. \end{aligned}$$

Thus using maximum principle given at Lemma 3 we have

$$\begin{aligned} \left\| E_{m+1}\right\| _{\infty } \le C\left( \Delta t \right) ^{2}. \end{aligned}$$

\(\square \)

Lemma 5

(Global error estimate.) The global error estimate \(TE_{m}\) in the temporal direction at \(t_{m}\) is given by

$$\begin{aligned} \left\| TE_{m}\right\| \le C\left( \Delta t \right) . \end{aligned}$$


The global error estimate at the \(\left( m \right) th \) time step is given by

$$\begin{aligned} \left\| TE_{m}\right\| _{\infty }&=\left\| \sum _{k=1}^{m}{e_{k}}\right\| _{\infty }, m\le \dfrac{T}{\Delta t} \\&\le \left\| e_{1}\right\| _{\infty } +\left\| e_{2}\right\| _{\infty } +...+\left\| e_{m}\right\| _{\infty }. \end{aligned}$$

Using local error estimates given in Lemma 4,

$$\begin{aligned}&\le C_{1}\left( (m) \Delta t \right) \left( \Delta t \right) \\&\le C_{1} T\left( \Delta t \right) , \text {since }~m\left( \Delta t \right) \le T\\&\le C \left( \Delta t \right) ,C=C_{1} T, \end{aligned}$$

where C is constant independent of \(\varepsilon \), \(\mu \) and \(\Delta t\). \(\square \)

Lemma 6

[41] The solution \(U^{m}(x)\) of semi-discretized scheme (4) and its derivatives satisfies

$$\begin{aligned} \vert { \frac{d^{i}U^{m}(x)}{dx^{i}}}\vert \le C\left( 1+\omega _{1}^{-i}e^{-\nu \omega _{1} x }+\omega _{2}^{-i}e^{-\nu \omega _{2} \left( 1-x \right) } \right) , \text {for}~0\le i\le 4, \end{aligned}$$

where \( \nu \) is any real constant number, \(\lambda _{1}\left( x\right) \) and \(\lambda _{2}\left( x\right) \) are two real solutions of (4) such that \(\lambda _{1}\left( x\right) <0\) and \(\lambda _{2}\left( x\right) >0\) and by assumption \(\omega _{1}=-\max _{x\in \left[ 0,1 \right] }\lambda _{1}\left( x\right) \) and \(\omega _{2}=\min _{x\in \left[ 0,1 \right] }\lambda _{2}\left( x\right) \).

Discrete extended cubic B-splines construction

The spatial domain \(\left[ 0,1\right] \) is discretized into N equal number of mesh elements each of length \( h=N^{-1} \). This gives the spatial mesh

$$\begin{aligned} \Omega _{x}^{N}=\left\{ x_{n}=nh, n=1, 2, . . ., N, {x}_{0}=0,{x}_{N}=1\right\} , \end{aligned}$$

where \(x_{n}\) is mesh points. The extended cubic B-spline basis of degree 4, \( K_{n}\left( x,\lambda \right) , \) is defined as the form

$$\begin{aligned} K_{n}\left( x ,\lambda \right) = \frac{1}{24h^{4}} \left\{ \begin{aligned}&4h\left( 1-\lambda \right) \left( x-x_{n-2} \right) ^{3}+3\lambda \left( x-x_{n-2} \right) ^{4}, x\in \left[ x_{n-2},x_{n-1} \right] ,\\&\left( 4-\lambda \right) h^{4}+12h^{3}\left( x-x_{n-1} \right) \\&+6h^{2}\left( 2+\lambda \right) \left( x-x_{n-1} \right) ^{2} -12h\left( x-x_{n-1} \right) ^{3} \\ {}&-3\lambda \left( x-x_{n-1} \right) ^{4}, x\in \left[ x_{n-1},x_{n} \right] ,\\&\left( 4-\lambda \right) h^{4}+12h^{3}\left( x_{n_{n+1}}-x \right) \\&+ 6h^{2}\left( 2+\lambda \right) \left( x_{n_{n+1}}-x \right) ^{2} -12h\left( x_{n_{n+1}}-x \right) ^{3} \\&-3\lambda \left( x_{n_{n+1}}-x \right) ^{4}, x\in \left[ x_{n},x_{n+1} \right] ,\\&4h\left( 1-\lambda \right) \left( x_{n+2}-x \right) ^{3}+3\lambda \left( x_{n+2}-x\right) ^{4}, x\in \left[ x_{n+1},x_{n+2} \right] ,\\&0, ~~~~\text {otherwise}. \end{aligned} \right. \end{aligned}$$
Table 1 Values of \( K_{n}\left( x \right) \) and its first two derivatives at the nodal points

An approximation extended cubic B-spline function, \( S(x,\lambda ) \) to the exact solution \(U\left( x,t_{m+1} \right) \) at \( (m+1) \)th time level is a linear combination of the extended cubic B-spline basis as

$$\begin{aligned} S(x,\lambda )=\sum _{n=-1}^{N+1}\zeta _{n}K_{n}\left( x,\lambda \right) , \end{aligned}$$

where \(\zeta _{n}\)’s are coefficients to be determined by collocation at each time level. Using the approximation given by (7) and Table 1 at nodal points \(x=x_{m}\) in (4) gives, The Eq. (3) can be rewritten as

$$\begin{aligned} \left\{ \begin{aligned}&\left( 1+\Delta t L_{\varepsilon ,\mu }^{\Delta t,h}\right) U^{m+1}(x_{n})=H\left( x_{n},t_{m}\right) ,\\&U^{m+1}(0)={\phi _{l}(t_{m+1})},\;m=0,...,M,\\&U^{m+1}(1)={\phi _{r}(t_{m+1})},\;m=0,...,M, \\&U^{m+1}(x_{n})={\phi _{b}(x_{n},t_{m+1})},\; x_{n}\in \left( 0,1\right) , -\left( s+1 \right) \le m \le -1, \end{aligned} \right. \end{aligned}$$


$$\begin{aligned} L_{\varepsilon ,\mu }^{\Delta t,h}&=-\sigma \left( \varepsilon ,\mu \right) \left( U_{xx}\right) ^{m+1} \left( x_{n}\right) -\mu a^{m+1}\left( x_{n}\right) \left( U_{x}\right) ^{m+1}\left( x_{n}\right) \\&+ b^{m+1}\left( x_{n}\right) U^{m+1}\left( x_{n}\right) ,\\ H\left( x_{n},t_{m}\right)&= -\Delta t c^{m+1}(x_{n}){{U}^{m+1-s}\left( x_{n}\right) }+\Delta tf^{m+1}\left( x_{n}\right) + U^{m}\left( x_{n}\right) . \end{aligned}$$

Putting the approximation (7) into collocation (8) the operator \(1+\Delta t L_{\varepsilon ,\mu }^{\Delta t,h} \) in (8) is given as

$$\begin{aligned} r_{n}^{-}\zeta _{n-1}+r_{n}^{c}\zeta _{n}+r_{n}^{+}\zeta _{n+1}=H\left( x_{n},t_{m}\right) , 0\le n\le N, \end{aligned}$$


$$\begin{aligned} \left\{ \begin{aligned} r_{n}^{-}&=-\sigma \left( \varepsilon ,\mu \right) \Delta t\frac{2+\lambda }{2h^{2}}+\mu \Delta t\frac{1}{2h}a^{m+1}\left( x_{n} \right) +\frac{4-\lambda }{24}\left( 1+ \Delta tb^{m+1}\left( x_{n}\right) \right) , \\ r_{n}^{c}&=\sigma \left( \varepsilon ,\mu \right) \Delta t\frac{2+\lambda }{h^{2}}+\frac{8+\lambda }{12}\Delta tb^{m+1}\left( x_{n} \right) , \\ r_{n}^{+}&=-\sigma \left( \varepsilon ,\mu \right) \Delta t\frac{2+\lambda }{2h^{2}}-\mu \Delta t\frac{1}{2h}a^{m+1}\left( x_{n} \right) +\frac{4-\lambda }{24}\left( 1+ \Delta tb^{m+1}\left( x_{n}\right) \right) , \\ \end{aligned} \right. \end{aligned}$$

where \(\sigma \left( \varepsilon ,\mu \right) =\varepsilon \frac{\rho \mu a_{m}}{2+\lambda }\coth {\left( \mu \frac{\rho a_{m}}{2}\right) }\).

For the given boundary conditions we have

$$\begin{aligned} \left\{ \begin{aligned}&\frac{4-\lambda }{24}\zeta _{-1}+\frac{8+\lambda }{12}\zeta _{0}+\frac{4-\lambda }{24}\zeta _{1}=\phi _{l}\left( t_{m+1}\right) ,\\&\frac{4-\lambda }{24}\zeta _{N-1}+\frac{8+\lambda }{12}\zeta _{N}+\frac{4-\lambda }{24}\zeta _{N+1}=\phi _{r}\left( t_{m+1}\right) . \end{aligned} \right. \end{aligned}$$

The Eqs.(9, 10) gives to \(\left( N+3\right) \times \left( N+3\right) \) systems in \(\left( N+3\right) \) unknowns \( \zeta _{-1},\zeta _{0},\zeta _{1},...,\zeta _{N+1}\). From Eqs. (9, 10), eliminating \(\zeta _{-1}\) and \(\zeta _{N+1}\) results \(\left( N+1\right) \) system of equations in \(\left( N+1\right) \) unknowns \(\zeta _{0}, \zeta _{1},...,\zeta _{N}\) which can be written in a matrix form as

$$\begin{aligned} RV=Q, \end{aligned}$$


$$\begin{aligned} R=\left( \begin{array}{ccccccccc} -2\left( \frac{8+\lambda }{4-\lambda } \right) r_{0}^{-}+r_{0}^{c} &{}-r_{0}^{-}+r_{0}^{+} &{} 0 &{}\dots &{} \dots &{} \dots &{}\dots &{} 0\\ R_{1}\left( x_{1} \right) &{} R_{2}\left( x_{1} \right) &{}R_{3}\left( x_{1} \right) &{} 0 &{}0 &{} \dots &{} \dots &{}0\\ 0&{}R_{1}\left( x_{2} \right) &{} R_{2}\left( x_{2} \right) &{}R_{3}\left( x_{3} \right) &{} 0 &{} \dots &{} \dots &{} 0\\ \vdots &{}\ddots &{}\ddots &{} \ddots &{} \vdots &{}\vdots &{} \vdots &{} \vdots \\ 0&{}\dots &{}\dots &{}\dots &{}0&{} R_{1}\left( x_{N-1} \right) &{} R_{2}\left( x_{N-1} \right) &{}R_{3}\left( x_{N-1} \right) \\ 0&{} \dots &{} \dots &{} \dots &{} \dots &{} 0 &{} r_{N}^{-}-r_{N}^{+}&{}r_{N}^{c} -2\left( \frac{8+\lambda }{4-\lambda } \right) r_{N}^{+} \end{array}\right) , \end{aligned}$$

where \( R_{n}\left( x_{n} \right) ,n=1,2,...,N-1 \) are defined as

$$\begin{aligned} \left\{ \begin{aligned} R_{1}\left( x_{n} \right)&=-\sigma \left( \varepsilon ,\mu \right) \Delta t\frac{2+\lambda }{2h^{2}}+\mu \Delta t\frac{1}{2h}a^{m+1}\left( x_{n} \right) +\frac{4-\lambda }{24}\left( 1+ \Delta tb^{m+1}\left( x_{n}\right) \right) , \\ R_{2}\left( x_{n} \right)&=\sigma \left( \varepsilon ,\mu \right) \Delta t\frac{2+\lambda }{h^{2}}+\frac{8+\lambda }{12}\Delta tb^{m+1}\left( x_{n} \right) , \\ R_{3}\left( x_{n} \right)&=-\sigma \left( \varepsilon ,\mu \right) \Delta t\frac{2+\lambda }{2h^{2}}-\mu \Delta t\frac{1}{2h}a^{m+1}\left( x_{n} \right) +\frac{4-\lambda }{24}\left( 1+ \Delta tb^{m+1}\left( x_{n}\right) \right) , \end{aligned} \right. \end{aligned}$$

and column vectors V and Q are given as \( V=\left[ \zeta _{0}, \zeta _{1},..., \zeta _{N}\right] ^{T}\) and

$$\begin{aligned}{} & {} Q=\Big [ H\left( x_{0},t_{m} \right) -\phi _{l}\left( t_{m+1} \right) r_{0}^{-}, H\left( x_{1},t_{m} \right) ,H\left( x_{2},t_{m} \right) ,..., \\{} & {} H\left( x_{N},t_{m} \right) -\phi _{r}\left( t_{m+1} \right) r_{N}^{+}\Big ] ^{T}. \end{aligned}$$

The matrix associated with Eq. (11) is of size \( (N + 1)\times (N + 1) \) with its entries for \( n = 1, 2,..., N-1 \) are \( R_{1}\left( x_{n} \right)<0, R_{2}\left( x_{n} \right) >0, R_{3}\left( x_{n} \right) <0. \) Therefore, the matrix R in Eq. (11) is an M-matrix and therefore its inverse exist and positive. Hence, tridiagonal system in Eq. (11) easily solved by any existing methods.

Convergence analysis

Lemma 7

The extended cubic B-splines \( K_{-1}\left( x,\lambda \right) \), \(K_{0}\left( x,\lambda \right) \), . . . \(K_{N}\left( x,\lambda \right) \), \(K_{N+1}\left( x,\lambda \right) \) satisfy \( \sum _{n=-1}^{N+1} \vert { K_{n}\left( x,\lambda \right) }\vert \le 1.75, 0<x<1. \)


At \( x_{n} \),

$$\begin{aligned}&\sum _{n=-1}^{N+1} \vert { K_{n}\left( x,\lambda \right) }\vert =\vert { K_{n-1}\left( x_{n},\lambda \right) }\vert +\vert { K_{n}\left( x_{n},\lambda \right) }\vert +\vert {K_{n+1}\left( x_{n},\lambda \right) }\vert \\&\quad =\frac{4-\lambda }{24}+\frac{8+\lambda }{12}+\frac{4-\lambda }{24}=1. \end{aligned}$$

For \( x_{n-1}<x< x_{n+1} \),

$$\begin{aligned}&\vert { K_{n}\left( x,\lambda \right) }\vert<\frac{8+\lambda }{12}, \vert { K_{n-1}\left( x,\lambda \right) }\vert<\frac{4-\lambda }{24},\\&\vert { K_{n+1}\left( x,\lambda \right) }\vert<\frac{4-\lambda }{24} , \\&\vert { K_{n-2}\left( x,\lambda \right) }\vert <\frac{4-\lambda }{24}. \end{aligned}$$

Thus, for \( x_{n-1}<x< x_{n+1}\),

$$\begin{aligned}&\sum _{n=-1}^{N+1} \vert { K_{n}\left( x,\lambda \right) }\vert =\vert { K_{n-1}\left( x_{n},\lambda \right) }\vert +\vert { K_{n}\left( x_{n},\lambda \right) }\vert \\&\quad +\vert { K_{n+1}\left( x_{n},\lambda \right) }\vert +\vert { K_{n-2}\left( x_{n},\lambda \right) }\vert =\frac{20+\lambda }{12}. \end{aligned}$$

Since \( -8<\lambda <1 \), so \( \frac{20+\lambda }{12}\le 1.75\) and this complete the proof. \(\square \)

Theorem 1

Let \( u\left( x_{n},t_{m+1} \right) \) be the continuous solution of Eqs. (1) and (2) and \( S\left( x,\lambda \right) \) be the collocation approximation from the space of splines to the solution \(U^{m+1}\left( x \right) \) be the approximate solution of Eq. (3). Then, for sufficiently large N, the following error bound holds

$$\begin{aligned} \vert {{L}_{\varepsilon ,\mu }^{\Delta t,h}\left( {U}^{m+1}\left( x_{n} \right) -S\left( x_{n},\lambda \right) \right) }\vert \le CN^{-2}. \end{aligned}$$


Let \( Z_{N}\left( x_{n} \right) \) be a unique spline interpolate to the solution \(U^{m+1}\left( x_{n} \right) \) of the problem (3) given by

$$\begin{aligned} Z_{N}\left( x_{n} \right) =\sum _{n=-1}^{N+1}\bar{\zeta }_{n}K_{n}\left( x,\lambda \right) . \end{aligned}$$

The estimates given in [42] yields

$$\begin{aligned} \begin{aligned}\left\| U^{m+1}\left( x_{n} \right) -Z_{N}\left( x_{n} \right) \right\| _{\infty } \le C_{0}\left\| \frac{d^{4}U^{m+1}\left( x_{n} \right) }{dx^{4}}\right\| _{\infty }N^{-4} \\\left\| \frac{dU^{m+1}\left( x_{n} \right) }{dx}-\frac{dZ_{N}\left( x_{n} \right) }{dx}\right\| _{\infty } \le C_{1}\left\| \frac{d^{4}U^{m+1}\left( x_{n} \right) }{dx^{4}}\right\| _{\infty }N^{-3}\\\left\| \frac{d^{2}U^{m+1}\left( x_{n} \right) }{dx^{2}}-\frac{d^{2}Z_{N}\left( x_{n} \right) }{dx^{2}}\right\| _{\infty } \le C_{0}\left\| \frac{d^{4}U^{m+1}\left( x_{n} \right) }{dx^{4}}\right\| _{\infty }N^{-2}. \end{aligned} \end{aligned}$$

Using triangle inequality,

$$\begin{aligned} \begin{aligned}\left\| {U}^{m+1}\left( x_{n} \right) -S\left( x_{n},\lambda \right) \right\| _{\infty } \le \left\| U^{m+1}\left( x_{n} \right) -Z_{N}\left( x_{n} \right) \right\| _{\infty }\\+\left\| Z_{N}\left( x_{n} \right) -S\left( x_{n},\lambda \right) \right\| _{\infty }. \end{aligned} \end{aligned}$$

The collocating conditions are \( {L}_{\varepsilon ,\mu }^{\Delta t,h}{U}^{m+1}\left( x_{n} \right) ={L}_{\varepsilon ,\mu }^{\Delta t,h}S\left( x_{n},\lambda \right) =H\left( x_{n},t_{m} \right) \). Assume that \({L}_{\varepsilon ,\mu }^{\Delta t,h}Z_{N}\left( x_{n}\right) =\bar{H}\left( x_{n},t_{m} \right) \) which satisfies the boundary conditions \( Z_{N}\left( x_{1}\right) =Z_{N}\left( x_{N+1}\right) \). Then,

$$\begin{aligned} \begin{aligned}\vert { {L}_{\varepsilon ,\mu }^{\Delta t,h}{U}^{m+1}\left( x_{n} \right) - {L}_{\varepsilon ,\mu }^{\Delta t,h}Z_{N}\left( x_{n}\right) }\vert =\vert { {L}_{\varepsilon ,\mu }^{\Delta t,h}S\left( x_{n},\lambda \right) - {L}_{\varepsilon ,\mu }^{\Delta t,h}Z_{N}\left( x_{n}\right) }\vert \\\quad =\vert {-\varepsilon \left( \frac{d^{2}U^{m+1}\left( x_{n} \right) }{dx^{2}}-\sigma \left( \varepsilon ,\mu \right) \frac{d^{2}Z_{N}\left( x_{n} \right) }{dx^{2}} \right) }\vert \\\quad +\vert {-\mu a\left( x\right) \left( \frac{dU^{m+1}\left( x_{n} \right) }{dx}-\frac{dZ_{N}\left( x_{n} \right) }{dx} \right) +b^{m+1}\left( x\right) \left( U^{m+1}\left( x_{n} \right) - Z_{N}\left( x_{n} \right) \right) }\vert \\\quad \le \vert {\varepsilon }\vert \vert {\sigma \left( \varepsilon ,\mu \right) }\vert \left\| \frac{d^{2}U^{m+1}\left( x_{n} \right) }{dx^{2}} \right\| _{\infty } + \vert {\varepsilon }\vert \vert {\sigma \left( \varepsilon ,\mu \right) }\vert \left\| \frac{d^{2}U^{m+1}\left( x_{n} \right) }{dx^{2}}-\frac{d^{2}Z_{N}\left( x_{n} \right) }{dx^{2}}\right\| _{\infty }\\\quad + \vert {\mu }\vert \left\| a(x)\right\| _{\infty } \left\| \frac{dU^{m+1}\left( x_{n} \right) }{dx}-\frac{dZ_{N}\left( x_{n} \right) }{dx} \right\| _{\infty } +\left\| b^{m+1}\left( x\right) \right\| _{\infty }\left\| U^{m+1}\left( x_{n} \right) - Z_{N}\left( x_{n} \right) \right\| _{\infty }. \end{aligned} \end{aligned}$$

Using Lemma 1 and using Eq.(13)

$$\begin{aligned} \max _{x\in D}\vert { {L}_{\varepsilon ,\mu }^{\Delta t,h}{U}^{m+1}\left( x_{n} \right) - {L}_{\varepsilon ,\mu }^{\Delta t,h}Z_{N}\left( x_{n}\right) }\vert \le CN^{-2}, \end{aligned}$$

this is because \( \vert {\sigma \left( \varepsilon ,\mu \right) -1 }\vert \le CN^{-2}. \) Equation (11) and \( {L}_{\varepsilon ,\mu }^{\Delta t,h}{U}^{m+1}\left( x_{n} \right) - {L}_{\varepsilon ,\mu }^{\Delta t,h}Z_{N}\left( x_{n}\right) \) results

$$\begin{aligned} R\left( V-\bar{V}\right) = Q-\bar{Q}, \end{aligned}$$


$$\begin{aligned}&V-\bar{V}=\left( \varsigma _{0}-\bar{\varsigma }_{0},\varsigma _{1}-\bar{\varsigma }_{1},. . . ,\varsigma _{N}-\bar{\varsigma }_{N} \right) ,\\&Q-\bar{Q}=\Big (H\left( x_{0},t_{m} \right) - \bar{H}\left( x_{0},t_{m} \right) ,H\left( x_{1},t_{m} \right) - \bar{H}\left( x_{1},t_{m} \right) , . . . ,\\&H\left( x_{N},t_{m} \right) - \bar{H}\left( x_{N},t_{m} \right) \Big ). \end{aligned}$$

The matrices R is invertible, i.e\( \vert {R^{-1}}\vert \le C \), and the boundary conditions are bounded. Therefore, Eqs. (15) and (16) results \( \vert { V-\bar{V}}\vert \le CN^{-2}\). Thus, Eqs. (7) and (12) gives

$$\begin{aligned} \left\| S\left( x_{n},\lambda \right) -Z_{N}\left( x \right) \right\| _{\infty } =\vert {\varsigma _{n}-\bar{\varsigma }_{n}}\vert \sum _{n=0}^{N+2} \vert {K_{n}\left( x,\lambda \right) }\vert \le CN^{-2}. \end{aligned}$$

\(\square \)

Theorem 2

Let \(u\left( x_{n},t_{m+1}\right) \) be the solution of the continuous problem (1)-(2) and \(U_{n}^{m+1}\) be the numerical solution of (8). Then, there exists a constant C such that the following uniform error estimate holds:

$$\begin{aligned} \sup _{0<\varepsilon \le 1}\max _{0\le n\le N,0\le m\le M}\vert { u\left( x_{n},t_{m+1}\right) -U_{n}^{m+1}}\vert \le C\left( \Delta t + N^{-2}\right) . \end{aligned}$$


The proof is the consequence of Lemma 5 and Theorem 1. \(\square \)

Numerical examples and results

In this section, two numerical results are used to confirm the theoretical results using the proposed numerical scheme. The exact solution of the numerical example is not available. Therefore, double mesh principle is used to find the maximum absolute error \(E_{\varepsilon ,\mu }^{N,M}\) and the corresponding convergence order \(p^{N,M}_{\varepsilon ,\mu }\) as

$$\begin{aligned} E_{\varepsilon ,\mu }^{N,M}=\max _{0\le n\le N,0\le m\le M}\vert { U_{n}^{m+1}-U_{2n}^{2m+1}}\vert \text {and}~~~~p^{N,M}_{\varepsilon ,\mu }={\log _{2}}\left( \frac{E_{\varepsilon ,\mu }^{N,M}}{E_{\varepsilon ,\mu }^{2N,2M}}\right) . \end{aligned}$$

The uniform error before extrapolation \(E^{N,M}\) and the corresponding uniform order of convergence before extrapolation \(p^{N,M}\) by:

$$\begin{aligned} E^{N,M}=\max _{\varepsilon ,\mu }E^{N,M}_{\varepsilon ,\mu } ~~ \text {and}~~ p^{N,M}={log_{2}}\left( \frac{E^{N,M}}{E^{2N,2M}}\right) , \end{aligned}$$

where \(U_{m}^{n+1}\) is a numerical solution obtained using the space and time \( N \times M \) mesh spacing with a mesh size of h or \( \Delta t \).

Example 1

Consider problem

$$\begin{aligned}&\frac{\partial u}{\partial t}-\varepsilon \frac{\partial ^2 u}{\partial x^2}-\mu (1+x)\frac{\partial u}{\partial x}+u(x,t) =u(x,t-\tau )\\&\quad -16x^{2}\left( 1-x \right) ^{2},(x,t)\in (0,1)\times (0,2], \end{aligned}$$


$$\begin{aligned} \left\{ \begin{aligned}&u(0,t)=0, u(1,t)=0,t\in \left( 0,2\right] ,\\&u(x,t)=0,(x,t)\in \left[ 0,1\right] \times \left[ -\tau ,0\right] . \end{aligned} \right. \end{aligned}$$

Example 2

Consider problem

$$\begin{aligned}&\frac{\partial u}{\partial t}-\varepsilon \frac{\partial ^2 u}{\partial x^2}-\mu \left( 1+x\left( 1-x \right) +t^{2} \right) \frac{\partial u}{\partial x}+\left( 1+5xt\right) u(x,t)\\&\quad =u(x,t-\tau )+x(1-x)\left( e^{t} -1 \right) ,\\&\quad (x,t)\in (0,1)\times (0,2], \end{aligned}$$


$$\begin{aligned} \left\{ \begin{aligned}&u(0,t)=0, u(1,t)=0,t\in \left( 0,2\right] ,\\&u(x,t)=0,(x,t)\in \left[ 0,1\right] \times \left[ -\tau ,0\right] . \end{aligned} \right. \end{aligned}$$

Maximum pointwise errors \((E^{N,M}_{\varepsilon ,\mu })\) and rate of convergence \((p^{N,M}_{\varepsilon ,\mu })\) for Example 1 and Example 2 have been demonstrated by fixing \(\mu =10^{-4}\) and \( \lambda =-1e-03 \) in Tables 2, 3 respectively, for various values of \( \varepsilon \). The results given in Tables 2, 3 clearly indicate that the proposed numerical method is accurate of order \(O\left( \left( \Delta t\right) +N^{-2} \right) \). Also, tabulated results in Tables 4, 5 indicates that maximum point-wise errors going to stabilized as the two parameters \( \mu \) and \( \varepsilon \) approaches to zero. Comparisons of our numerical results with those of [35] are presented in Tables 6, 7. From these tables, we can confirm the more accurate of the proposed numerical method. The numerical solutions obtained by the numerical scheme presented in Example 1 are shown in Fig. 1a, b and numerical scheme presented in Example 2 are shown in Fig. 2a, b. From Figs. 1a,  2a, we confirm the occurrence of both left and right boundary layers near \( x = 0 \) and \( x = 1 \) for \( \mu = 10^{-6} \) and boundary layers near \( x = 0 \) for \( \mu = 10^{-1} \). The graphs between N and maximum pointwise errors of Examples 1 and 2 are plotted as the log-log scale respectively, in Fig. 3a, b. From these two graphs, one can observe that the numerical scheme converges uniformly as the perturbation parameters goes very small.

Table 2 \(E_{\varepsilon ,\mu }^{N,M}\) and \(p^{N,M}_{\varepsilon ,\mu }\) with \(\mu =10^{-4}, \lambda =-1e-03,\) for Example 1
Table 3 \(E_{\varepsilon ,\mu }^{N,M}\) and \(p^{N,M}_{\varepsilon ,\mu }\) with \(\mu =10^{-4}, \lambda =-1e-03,\) for Example 2
Table 4 \(E^{N,M}_{\varepsilon ,\mu }\) and \(p^{N,M}_{\varepsilon ,\mu }\) with \( \lambda =-1e-03, \) for Example 1
Table 5 \(E^{N,M}_{\varepsilon ,\mu }\) and \(p^{N,M}_{\varepsilon ,\mu }\) with \( \lambda =-1e-03, \) for Example 2
Table 6 \(E^{N,M}_{\varepsilon ,\mu }\) and \(p^{N,M}_{\varepsilon ,\mu }\) with \(\mu =10^{-3}, \lambda =0, \) for Example 1
Table 7 \(E^{N,M}_{\varepsilon ,\mu }\) and \(p^{N,M}_{\varepsilon ,\mu }\) with \(\mu =10^{-9}, \lambda =0, \) for Example 2
Fig. 1
figure 1

Surface plot of the numerical solution for Example 1 with \(N=M=32\), (a) \(\varepsilon = 10^{-1}\), \(\mu = 10^{-6}\) (b) \(\varepsilon = 10^{-6}\), \(\mu = 10^{-1}.\)

Fig. 2
figure 2

Surface plot of the numerical solution for Example 2 with \(N=M=32\), (a) \(\varepsilon = 10^{-1}\), \(\mu = 10^{-6}\) (b) \(\varepsilon = 10^{-6}\), \(\mu = 10^{-1}.\)

Fig. 3
figure 3

Log-Log plot of the maximum error on left (a) for Example 1 with \(\mu = 10^{-4}\) and on right (b) for Example 2 with \(\mu = 10^{-4}\)


In this paper, the exponentially fitted strategy is applied to extended cubic B-spline scheme for solving a two-parameter singularly perturbed temporal delay parabolic problem. In our present study of continuous problem, the temporal direction is discretized by an implicit-Euler scheme with a uniform mesh, and the spatial direction is discretized by an exponentially fitted extended cubic B-spline finite difference method fitting only one parameter \( \varepsilon \). We have proved that the method provides first-order and second-order accurate uniformly convergent in time and space respectively. Two numerical tests are introduced to confirm the effectiveness of the proposed numerical scheme and approve the theoretical findings.


The proposed uniformly convergent numerical approach is based on a uniform mesh that does not resolve boundary layers because there are not a sufficient number of mesh points in boundary regions.

Availability of data and materials

No additional data is used for this research work.


  1. Liu X, Zhou M, Ansari AH, Chakrabarti K, Abbas M, Rathour L. Coupled fixed point theorems with rational type contractive condition via c-class functions and inverse c k-class functions. Symmetry. 2022;14(8):1663.

    Article  Google Scholar 

  2. Iqbal J, Mishra VN, Mir WA, Dar AH, Ishtyak M, Rathour L. Generalized resolvent operator involving (·,·)-co-monotone mapping for solving generalized variational inclusion problem. Georgian Math J. 2022;29(4):533–42.

    Article  Google Scholar 

  3. Gupta S, Rathour L. Approximating solutions of general class of variational inclusions involving generalized? i? j-(hp,?)-q-accretive mappings. FILOMAT. 2023;37(19):6255–75.

    Google Scholar 

  4. Kumar A, Verma A, Rathour L, Mishra LN, Mishra VN. Convergence analysis of modified szász operators associated with hermite polynomials. Rendiconti del Circolo Matematico di Palermo Series. 2023;2:1–15.

    Google Scholar 

  5. KRASNIQI XZ, Mishra LN. On the power integrability with weight of double trigonometric series. Adv Stud Contemp Math. 2021;31(2):221–42.

    Google Scholar 

  6. Pathak VK, Mishra LN. Application of fixed point theorem to solvability for non-linear fractional hadamard functional integral equations. Mathematics. 2022;10(14):2400.

    Article  Google Scholar 

  7. Mishra LN, Pathak VK, Baleanu D. Approximation of solutions for nonlinear functional integral equations. Aims Math. 2022;7(9):17486–506.

    Article  Google Scholar 

  8. Sundareswaran R, Vijayan S, Venkatesh S, Mishra LN, Broum S, et al. Failure analysis of pump piping system using dematel svn methodology. Neutrosophic Sets Syst. 2023;53(1):13.

    Google Scholar 

  9. Mehra S, Rathour L, Mishra LN, et al. Complete orthogonally modular metric space and fixed point results for orthogonal generalized f-contraction mappings. Adv Stud Euro-Tbilisi Mathl J. 2023;16(2):115–36.

    Google Scholar 

  10. Ma W-X. Nonlocal pt-symmetric integrable equations and related riemann-hilbert problems. Partial Differ Equ Appl Math. 2021;4: 100190.

    Article  Google Scholar 

  11. Ma W-X, Lee J-H. A transformed rational function method and exact solutions to the 3+ 1 dimensional jimbo-miwa equation. Chaos Solitons Fractals. 2009;42(3):1356–63.

    Article  Google Scholar 

  12. Ma W-X. A polynomial conjecture connected with rogue waves in the kdv equation. Partial Differ Equ Appl Math. 2021;3: 100023.

    Article  Google Scholar 

  13. Hailu WS, Duressa GF. Accelerated parameter-uniform numerical method for singularly perturbed parabolic convection-diffusion problems with a large negative shift and integral boundary condition. Results Appl Math. 2023;18: 100364.

    Article  Google Scholar 

  14. Gobena WT, Duressa GF. An optimal fitted numerical scheme for solving singularly perturbed parabolic problems with large negative shift and integral boundary condition. Results Control Optim. 2022;9: 100172.

    Article  Google Scholar 

  15. Ladyzhenskaia OA, Solonnikov VA, Ural’tseva NN. Linear and quasi-linear equations of parabolic type. Rhode: American mathematical soc; 1968.

    Book  Google Scholar 

  16. Negero NT, Duressa GF. Uniform convergent solution of singularly perturbed parabolic differential equations with general temporal-lag. Iran J Sci Technol Trans Sci. 2022;46(2):507–24.

    Article  Google Scholar 

  17. Woldaregay MM, Duressa GF. Accurate numerical scheme for singularly perturbed parabolic delay differential equation. BMC Res Notes. 2021;14(1):1–6.

    Article  Google Scholar 

  18. Bhathawala P, Verma A. A two-parameter singular perturbation solution of one dimension flow through unsaturated porous media. Appl Math. 1975;43(5):380–4.

    Google Scholar 

  19. Van Harten A, Schumacher J. On a class of partial functional differential equations arising in feed-back control theory. Amsterdam: Elsevier; 1978.

    Book  Google Scholar 

  20. Tikhonov AN, Samarskii AA. Equations of mathematical physics. Chelmsford: Courier Corporation; 2013.

    Google Scholar 

  21. Gupta A, Kaushik A, Sharma M. A higher-order hybrid spline difference method on adaptive mesh for solving singularly perturbed parabolic reaction-diffusion problems with robin-boundary conditions. Numer Methods Partial Differ Equ. 2023;39(2):1220–50.

    Article  Google Scholar 

  22. Duressa GF, Debela HG. Numerical solution of singularly perturbed differential difference equations with mixed parameters. J Math Model. 2021;9(4):691–705.

    Google Scholar 

  23. Gelu FW, Duressa GF. A parameter-uniform numerical method for singularly perturbed robin type parabolic convection-diffusion turning point problems. Appl Numer Math. 2023;190:50–64.

    Article  Google Scholar 

  24. Chandru M, Prabha T, Das P, Shanthi V. A numerical method for solving boundary and interior layers dominated parabolic problems with discontinuous convection coefficient and source terms. Differ Equ Dyn Syst. 2019;27(1):91–112.

    Article  Google Scholar 

  25. Kumar D. A parameter-uniform scheme for the parabolic singularly perturbed problem with a delay in time. Numer Methods Partial Differ Equ. 2021;37(1):626–42.

    Article  Google Scholar 

  26. Negero NT, Duressa GF. A method of line with improved accuracy for singularly perturbed parabolic convection-diffusion problems with large temporal lag. Results Appl Math. 2021;11: 100174.

    Article  Google Scholar 

  27. Negero N, Duressa G. An efficient numerical approach for singularly perturbed parabolic convection-diffusion problems with large time-lag. J Math Model. 2022;10(2):173.

    Google Scholar 

  28. Negero NT, Duressa GF. An exponentially fitted spline method for singularly perturbed parabolic convection-diffusion problems with large time delay. Tamkang Journal of Mathematics 2022.

  29. Negero NT, Duressa GF. Parameter-uniform robust scheme for singularly perturbed parabolic convection-diffusion problems with large time-lag. Comput Methods Differ Equ. 2022;10(4):954–68.

    Google Scholar 

  30. Negero NT. A robust fitted numerical scheme for singularly perturbed parabolic reaction-diffusion problems with a general time delay. Results Phys. 2023;51: 106724.

    Article  Google Scholar 

  31. Tesfaye SK, Woldaregay MM, Dinka TG, Duressa GF. Fitted computational method for solving singularly perturbed small time lag problem. BMC Res Notes. 2022;15(1):1–10.

    Article  Google Scholar 

  32. Li J, Navon IM. Uniformly convergent finite element methods for singularly perturbed elliptic boundary value problems i: reaction-diffusion type. Comput Math Appl. 1998;35(3):57–70.

    Article  Google Scholar 

  33. Li J. Convergence analysis of finite element methods for singularly perturbed problems. Comput Math Appl. 2000;40(6–7):735–45.

    Article  Google Scholar 

  34. Govindarao L, Mohapatra J, Sahu S. Uniformly convergent numerical method for singularly perturbed two parameter time delay parabolic problem. Int J Appl Comput Math. 2019;5(3):1–9.

    Article  Google Scholar 

  35. Kumar S, Kumar M, et al. A robust numerical method for a two-parameter singularly perturbed time delay parabolic problem. Comput Appl Math. 2020;39(3):1–25.

    Google Scholar 

  36. Negero NT. A uniformly convergent numerical scheme for two parameters singularly perturbed parabolic convection-diffusion problems with a large temporal lag. Results Appl Math. 2022;16: 100338.

    Article  Google Scholar 

  37. Negero NT. Fitted cubic spline in tension difference scheme for two-parameter singularly perturbed delay parabolic partial differential equations. Partial Differ Equ Appl Math. 2023.

    Article  Google Scholar 

  38. Negero NT. A fitted operator method of line scheme for solving two-parameter singularly perturbed parabolic convection-diffusion problems with time delay. J Math Model. 2023.

    Article  Google Scholar 

  39. Negero NT. A parameter-uniform efficient numerical scheme for singularly perturbed time-delay parabolic problems with two small parameters. Partial Differ Equ Appl Math. 2023;7: 100518.

    Article  Google Scholar 

  40. Negero N. A robust uniformly convergent scheme for two parameters singularly perturbed parabolic problems with time delay. Iran J Numer Anal Optim. 2023.

    Article  Google Scholar 

  41. Linß T, Roos H-G. Analysis of a finite-difference scheme for a singularly perturbed problem with two small parameters. J Math Anal Appl. 2004;289(2):355–66.

    Article  Google Scholar 

  42. Hall C. On error bounds for spline interpolation. J Approx Theory. 1968;1(2):209–18.

    Article  Google Scholar 

Download references


No funding organization for this research work.

Author information

Authors and Affiliations



The author read and approved the final manuscript.

Corresponding author

Correspondence to Naol Tufa Negero.

Ethics declarations

Ethics approval and consent to participate

Not applicable.

Consent for publication

Not applicable.

Competing interests

The author declares that he has no competing interests.

Additional information

Publisher's Note

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Rights and permissions

Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit The Creative Commons Public Domain Dedication waiver ( applies to the data made available in this article, unless otherwise stated in a credit line to the data.

Reprints and permissions

About this article

Check for updates. Verify currency and authenticity via CrossMark

Cite this article

Negero, N.T. Uniformly convergent extended cubic B-spline collocation method for two parameters singularly perturbed time-delayed convection-diffusion problems. BMC Res Notes 16, 282 (2023).

Download citation

  • Received:

  • Accepted:

  • Published:

  • DOI:


Mathematics Subject Classification