Abstract
The integro-differential Cauchy problem with exponential inhomogeneity and with a spectral value that turns zero at an isolated point of the segment of the independent variable is considered. The problem belongs to the class of singularly perturbed equations with an unstable spectrum and has not been considered before in the presence of an integral operator. A particular difficulty is its investigation in the neighborhood of the zero spectral value of inhomogeneity. Here, it is not possible to apply the well-known procedure of Lomov’s regularization method, so the authors have chosen the method of constructing the asymptotic solution of the initial problem based on the use of the regularized asymptotic solution of the fundamental solution of the corresponding homogeneous equation whose construction from the positions of the regularization method has not been considered so far. In the case of an unstable spectrum, it is necessary to take into account its point features. In this case, inhomogeneity plays an essential role. It significantly affects the type of singularities in the solution of the initial problem. The fundamental solution allows us to construct asymptotics regardless of the nature of the inhomogeneity (it can be both slowly changing and rapidly changing, for example, rapidly oscillating). The approach developed in the paper is universal with respect to arbitrary inhomogeneity. The first part of the study develops an algorithm for the regularization method to construct the asymptotic (of any order on the parameter) of the fundamental solution of the corresponding homogeneous integro-differential equation. The second part is devoted to constructing the asymptotics of the solution of the original problem. The main asymptotic term is constructed in detail, and the possibility of constructing its higher terms is pointed out. In the case of a stable spectrum, we can construct regularized asymptotics without using a fundamental solution.
Keywords:
singularly perturbed problem; integro-differential equation; exponential heterogeneity; spectral value instability; regularized asymptotics of the fundamental solution MSC:
34K26; 45J05
1. Introduction
Consider the integro-differential problem
where is some scalar function and is the so-called spectral value of the inhomogeneity (see, for example, [1]). As was indicated in [2,3], the structure of the asymptotics of the solution of the problem (1) essentially depends on the behavior of the functions and on the segment (here, the segment is taken for the sake of simplicity; any segment can be taken instead). This becomes clear if we pass from problem (1) to the system
The matrix of the differential part of this system can have both a stable spectrum and an unstable spectrum (see [1], pp. 39–40, 186–187). In each of these cases, the asymptotics has its structure. A particularly complicated structure arises in the case of spectrum instability, characterized by the presence of so-called point features. For example, if it has zero of the order r on the segment , then in the asymptotics of the solution of the corresponding differential problem
singularities appear, which are described by special functions of the type
absentees in the case A description of singularities in the presence of an integral term in (2) becomes very difficult in this case. There are results for singularly perturbed integro-differential systems in which the exponential factor of the kernel of the integral operator is unstable, while the spectrum of the corresponding differential part is stable (see, for example, [4,5,6,7]). Problems with the spectrum instability of the limit operator were considered mainly for ordinary differential systems [8,9,10]. In them, the role of instability is played by the zeros of the coefficient or of the matrices . Generalizing the algorithm of the regularization method [2,3], to singularly perturbed integral and integro-differential problems with rapidly oscillating coefficients [11,12,13,14,15,16], we studied the influence of rapidly oscillating inhomogeneities on the asymptotics of the solution [17,18,19,20,21,22]. An investigation of the instability of the general case of the spectrum in the integro-differential problem has not yet been carried out. In our case, the instability is the exponential factor of inhomogeneity. The problem has not been considered in this formulation before. The study of the general case instability of the spectrum in the integro-differential problem has not yet been carried out. We begin its study with problem (1), in which the spectral value of the inhomogeneity is unstable:
The idea of constructing an asymptotic solution of the problem (1) under the indicated assumptions is based on the use of regularized asymptotics of the fundamental solution of the corresponding homogeneous integro-differential system and the next statement.
Lemma 1.
Let be the fundamental solution of a homogeneous integro-differential Volterra equation (s is a parameter):
Then, the solution of the integro-differential problem
is given by the formula
Proof.
Although the justification for this fact is given in [1] (pp. 268–269), we will give it again since in [1] it is given very concisely. Substitute (5) into (4); this will result in:
Here, we have changed the order of integration in the iterated integral:
and used equality (3) twice in the underlined expressions, replacing them with zeros. The lemma is proved. □
As mentioned earlier, to construct an asymptotic solution of the problem (1), we will use the first-order asymptotic of the fundamental solution . In this case, the main term of the asymptotic solution of the original problem (1) will be obtained. To construct the higher terms of the asymptotics, we also need the higher terms of the asymptotics of the fundamental solution. Therefore, in the following sections of our work, we construct an algorithm for constructing a complete asymptotic solution of the function
2. Regularization of the Problem
Problem (3) will be considered under the following conditions:
- (i)
- the function the kernel
- (ii)
Following the regularization method of S.A. Lomov [2,3], we introduce an additional variable
and instead of problem (3), consider the “extended” problem
for the function such that the function is an exact solution of the problem (3). However, the problem (7) cannot be considered fully regularized since it has not been regularized integral operator
As was shown in [2], for its regularization it is necessary to introduce an invariant space with respect to the integral operator J (see [2], p. 62). It is done like this. Let us first introduce the space U of functions of the form
As we take the class It is necessary to show that the image
is represented by a series of asymptotically converging to at (uniformly in Substituting the element of the space (9) into (10), we have
The second integral standing here can be expanded into an asymptotic series. Applying the operation integration by parts, we have
Continuing this process further, we obtain the series
where the operators
are introduced. It is easy to show (see, for example, [4], pp. 291–293 ), that the series on the right-hand side of the equality (12) converges asymptotically to the integral when (uniformly in ). Thereby, it is shown that the class is asymptotically invariant with respect to the operator
The class , which is invariant with respect to the operator J, allows us to regularize this operator, using its image on an element of the class It is done like this. We introduce operators
where operator has the form (13). These operators are called order operators because they are coefficients at the corresponding degree in Using operators the operator can be written shorter:
Now, is an arbitrary continuous function in
with asymptotic expansion
converging at (uniformly in Then, the image expands into an asymptotic series
Equality (16) is the basis for introducing the following concept.
Definition 1.
A formal extension of an operator J is the operator acting for each function of the form (9) according to the law
From (16), the asymptotic equality follows, which shows that the operator , which we have introduced, is indeed an extension of the operator Although the operator is formally defined, its usefulness is obvious since in practice usually the N-th approximation of the asymptotic solution (15) is constructed, in which only N-th partial sums of the series (15) are used, which have not a formal, but a true meaning.
Now, we can write the problem that is completely regularized with respect to the original problem (3):
where has the form (17).
3. Solvability of Iterative Problems
Defining the solution of the problem (18) in the form of series (15), we obtain the following iterative problems:
⋯
Let us turn to the study of their solvability in the space
Each of the problems (19a–e) can be represented as
where is the known function of space b is equal to zero or one, and is the operator, acting on each function (9) of the space U according to the law:
In the space U, we introduce the scalar product (for each )
We prove the following assertion (see, [5,8,9]).
Theorem 1.
Let conditions (i) and (ii) be satisfied and the right side of the Equation (20) belongs to the space Then, for the solvability of the Equation (20) in the space U, it is necessary and sufficient that the identity
holds.
Proof.
Let us substitute the element (9) of the space U into Equation (20); we obtain the equality
Equating here separately the free terms and the coefficients at the exponent, we obtain the equations
The second Equation (22) is a Volterra-type integral equation; it is uniquely solvable in the space . For the solvability of the first Equation (22) in this space, it is necessary and sufficient, that i.e., that condition (21) be satisfied. The theorem has proved. □
Remark 1.
When conditions (i) and (ii), as well as condition (21), hold, Equation (20) has the following solution in the space U:
where is an arbitrary function, is the solution of the integral Equation (22).
We will not formulate a theorem on the unique solvability of the general iterative problem (20). Note that the application of Theorem 1 to two successive iterative problems and leads to a unique calculation of the solution of the iterative problem in the space Let us show this is in the example of the problems (19a–e) (see, for example [16,20,22]).
The solution of equation has the form of the sum (23). Since there is no inhomogeneity in , the integral Equation (22) will be homogeneous and, therefore, and the solution (23) itself takes form Subordinating it to the initial condition we find that Thus, the solution of the problem will be written in the form where the function is found only at the point For its final calculation, let us move on to the Equation :
By Theorem 1, this equation is solvable in the space U if and only if condition
holds. Attaching the initial condition to it, we uniquely find the function
which means that we uniquely construct the solution of the problem in the space In this case, the problem will be already heterogeneous:
Since the right-hand side of this equation has no coefficient at the exponent , condition (21) is satisfied, and we can calculate the solution to this equation (see Formula (23)):
where is an arbitrary function, and is the solution of the Volterra-type integral equation:
It exists, is unique, and belongs to the class . Subordinating (25) to the initial condition we obtain the equation
For the final calculation of the function (25), it is necessary to proceed to the following problem :
Given the type of operators and :
we select the coefficient at the exponent on the right side of the Equation . It will be as follows:
where is a known function. Adding the initial condition (27) to this equation, we obtain the function , and, therefore, we construct uniquely the solution (25) of the problem . Similar solutions are calculated in the space U of the following iterative problems for
4. Asymptotic Convergence of Formal Solutions to the Exact Solution
Applying Theorem 1 to iterative problems , we uniquely calculate their solutions in the space Let us denote by the N-th partial sum of the series (15) and through is the restriction of this sum at It is easy to prove the following assertion (see [2,4,14,16,20,21]).
Lemma 2.
Let conditions (i) and (ii) be satisfied. Then, the function is a formal asymptotic solution of the problem (3) of order , i.e., satisfies the problem
where is a constant independent on ε for sufficiently small ).
To estimate the difference between the exact and approximate solutions of the problem (3), we must consider the integro-differential problem
where is a known function. This problem has a unique solution for each , and it is necessary to estimate the norm of the solution in terms of the right-hand side Let us introduce another unknown function
Differentiating it with respect to we obtain the integro-differential problem:
As a result, for the vector function we obtain the integro-differential problem
Denote by the Cauchy matrix of the differential problem
The matrix is a matrix of simple structure with spectrum Indeed, we have an obvious equality
from which follows the simplicity of structure of the matrix Since the spectrum of the matrix lies in the half-plane the matrix is uniformly bounded (see, for example, [4,13,18,19], pp. 119–120), i.e.,
Let us use this matrix to invert the problem (31); we obtain the integral system
Passing here to norms, we obtain an integral inequality with respect to the norm Taking into account uniform boundedness of the matrix as well as the continuity of the functions and (and hence their boundedness) and applying the Gronwall–Bellman inequality [23], we obtain for sufficiently small inequality
from which follows the estimate
The following assertion is proved.
Lemma 3.
Let conditions (i) and (ii) be satisfied. Then, the solution of the problem (29) exists; is unique for sufficiently small ; and satisfies estimate (32), where the constant does not depend on ε for sufficiently small
Let us apply this lemma to prove the following assertion.
Theorem 2.
Let conditions (i) and (ii) be satisfied. Then, the problem (3) is uniquely solvable in the class and its solution satisfies the estimate
where the constant does not depend on ε for sufficiently small .
Proof.
The problem (3) is uniquely solvable since it reduces to the problem (29) by replacing By Lemma 3, for the difference we obtain the problem
It has the form of the problem (29) with inhomogeneity By Lemma 3, we have the estimate
and hence for there will be an estimate
Consequently, the following relation follows from the resulting inequality:
or
where and the constant does not depend on where is sufficiently small. The theorem has proved. □
Since in the next section we will use the first order asymptotics fundamental solution, we write it explicitly:
where is the function (), is the function (26), and the function satisfies the Equation with the initial condition (27).
5. Construction of an Asymptotic Solution to the Original Problem
We will study the original problem (1) under the following assumptions:
- (iii)
- (iv)
By Lemma 1, the problem (1) has the following solution:
where is the solution of the problem (3) at Let us use the first-order asymptotics (33) of the solution of the problem (3):
Substituting (35) into (34), we arrive at the equality
or
The integrals enclosed in a box can be integrated by parts; such an operation cannot be applied to the integral since the spectral value turns to zero at the point . Let us first deal with the integrals in the box. We have:
For the second integral in the box, we introduce the notation and again apply the integration operation in parts:
Since we are going to construct the main term of the asymptotic of the solution of the original problem (1), then this term can be neglected. The most difficult is the construction of the asymptotics of the integral since the classical operation of integration by parts does not work here. For this, as is known, it is necessary to apply the modified operation of integration by parts (see, for example, [4], Ch. 6). Let us denote for convenience Let be the basic system of Lagrange–Sylvester polynomials for the spectral value ([24], pp. 87–92). Then,
where is some function. Let us do it the following operations:
It is already possible to apply the classical operation of integration by parts to the integral :
Leaving in (36) -independent terms, as well as the term
we obtain the following equality:
Here, the second term contains the so-called surge functions which in the neighborhood of the point have a maximum proportional to the value
(see [25], pp. 207–212). They do not belong to the class and therefore should be included in the main term of the asymptotics of the solution of problem (1):
where is the function
Remark 2.
We have constructed the main term (38) of the asymptotic solution of the problem (1) using the first-order asymptotics (35) for the fundamental solution It is clear that in order to obtain the N-th order asymptotic for the solution one should use the -th order asymptotic for the function
6. Conclusions
In this paper, we consider the case of one exponential inhomogeneity. Similarly, the case of several inhomogeneities with unstable spectral values can be considered. However, in this case, different exponential inhomogeneities may have different points of instability belonging to the considered time interval. Therefore, it will not be easy to single out essentially special singularities in the solution of the original problem and to carry out further regularization of the problem. Using our method of constructing asymptotics with the help of a fundamental solution, this difficulty can also be overcome. The specific implementation of this case is the subject of a separate article. It will be more difficult to study the case of continual irreversibility of the spectral value, i.e., the case when the spectral value vanishes on some subset S of positive measure. We intend to study it in our future works.
Author Contributions
Methodology, B.K., V.S. and D.Z. All authors have read and agreed to the published version of the manuscript.
Funding
The work was partially supported by a grant No. 23-21-00496 Russian Science Foundation.
Data Availability Statement
Data is contained within the article.
Conflicts of Interest
The authors declare no conflict of interest. The funders had no role in the design of the study; in the collection, analyses, or interpretation of data; in the writing of the manuscript; or in the decision to publish the results.
References
- Imanaliev, M.I. Oscillations and Stability of Solutions of Singularly Perturbed Integro-Differential Systems; Publishing House “Ilim”: Frunze, Kyrgystan, 1974. [Google Scholar]
- Lomov, S.A. Introduction to the General Theory of Singular Perturbations; Science: Moscow, Russia, 1981. [Google Scholar]
- Lomov, S.A.; Lomov, I.S. Fundamentals of the Mathematical Theory of the Boundary Layer; Moscow University Press: Moscow, Russia, 2011. [Google Scholar]
- Safonov, V.F.; Bobodzhanov, A.A. Higher Mathematics Course. Singularly Perturbed Problems and the Regularization Method; MPEI Publishing House: Moscow, Russia, 2012. [Google Scholar]
- Bobodzhanov, A.A.; Safonov, V.F. Volterra integral equations with rapidly changing kernels and their asymptotic integration. Sb. Math. 2001, 192, 1139–1164. [Google Scholar] [CrossRef] [Scilit]
- Bobodzhanov, A.A.; Safonov, V.F. Singularly perturbed integro-differential systems with contrast structures. Sb. Math. 2005, 196, 173–200. [Google Scholar] [CrossRef] [Scilit]
- Kalimbetov, B.T.; Safonov, V.F. Regularization method for systems with unstable spectral value of the kernel of an integral operator. Differ. Equ. 1995, 31, 647–656. [Google Scholar]
- Lomov, S.A.; Safonov, V.F. Regularization and asymptotic solutions of singularly perturbed problems with point singularities of the spectrum of the limit operator. Ukr. Math. J. 1984, 36, 172–180. [Google Scholar] [CrossRef] [Scilit]
- Eliseev, A.G.; Lomov, S.A. Theory of singular perturbations in the case of spectral singularities of the limit operator. Math. Sb. 1986, 131, 544–557. [Google Scholar] [CrossRef] [Scilit]
- Kalimbetov, B.T.; Safonov, V.F. Contrasting structures for linear problems with violation of spectral stability on sets of positive measure. Uzbek. Math. J. 2002, 1, 36–43. [Google Scholar]
- Kalimbetov, B.T.; Safonov, V.F.; Tuychiev, O.D. Singular perturbed integral equations with rapidly oscillation coefficients. SEMR 2020, 17, 2068–2083. [Google Scholar] [CrossRef] [Scilit]
- Bobodzhanov, A.A.; Kalimbetov, B.T.; Safonov, V.F. Nonlinear singularly perturbed integro-differential equations and regularization method. WSEAS Trans. Math. 2020, 19, 301–311. [Google Scholar] [CrossRef] [Scilit]
- Kalimbetov, B.T.; Safonov, V.F. Regularization method for singularly perturbed integro-differential equations with rapidly oscillating coefficients and with rapidly changing kernels. Axioms 2020, 9, 131. [Google Scholar] [CrossRef] [Scilit]
- Kalimbetov, B.T.; Safonov, V.F. Singularly perturbed integro-differential equations with rapidly oscillating coefficients and with rapidly changing kernel in the case of a multiple spectrum. WSEAS Trans. Math. 2021, 20, 84–96. [Google Scholar] [CrossRef] [Scilit]
- Bobodzhanov, A.A.; Kalimbetov, B.T.; Safonov, V.F. Asymptotic solutions of singularly perturbed integro-differential systems with rapidly oscillating coefficients in the case of a simple spectrum. AIMS Math. 2021, 6, 8835–8853. [Google Scholar] [CrossRef] [Scilit]
- Bobodzhanov, A.A.; Kalimbetov, B.T.; Safonov, V.F. Integro-differential problem about parametric amplification and its asymptotical integration. Int. J. Appl. Math. 2020, 33, 331–353. [Google Scholar] [CrossRef] [Scilit]
- Kalimbetov, B.T.; Safonov, V.F.; Madikhan, E. Singularly perturbed integral equations with a rapidly oscillating inhomogeneity. IJAM 2021, 34, 653–668. [Google Scholar] [CrossRef] [Scilit]
- Bobodzhanov, A.A.; Kalimbetov, B.T.; Safonov, V.F. Generalization of the regularization method to singularly perturbed integro-differential systems of equations with rapidly oscillating inhomogeneity. Axioms 2021, 10, 40. [Google Scholar] [CrossRef] [Scilit]
- Bibulova, D.A.; Kalimbetov, B.T.; Safonov, V.F. Regularized asymptotic solutions of a singularly perturbed Fredholm equation with a rapidly varying kernel and a rapidly oscillating inhomogeneity. Axioms 2022, 11, 141. [Google Scholar] [CrossRef] [Scilit]
- Kalimbetov, B.T.; Tuychiev, O.D. Asymptotic solution of the Cauchy problem for the singularly perturbed partial integro-differential equation with rapidly oscillating coefficients and with rapidly oscillating heterogeneity. OPEN Math. 2021, 19, 244–258. [Google Scholar] [CrossRef] [Scilit]
- Bobodzhanov, A.A.; Kalimbetov, B.T.; Safonov, V.F. Algorithm of the regularization method for a nonlinear singularly perturbed integro-differential equation with rapidly oscillating inhomogeneities. Differ. Equ. 2022, 58, 392–404. [Google Scholar] [CrossRef] [Scilit]
- Bobodzhanov, A.A.; Kalimbetov, B.T.; Safonov, V.F. Algorithm of the regularization method for a singularly perturbed integro-differential equation with a rapidly decreasing kernel and rapidly oscillating inhomogeneity. J. Sib. Fed. Univ. Math. Phys. 2022, 15, 216–225. [Google Scholar] [CrossRef] [Scilit]
- Hartman, P. Ordinary Differential Equations; Wiley: New York, NY, USA, 1964. [Google Scholar]
- Gantmacher, F.R. The Theory of Matrices; GITTL: Moscow, Russia, 1953. [Google Scholar]
- Safonov, V.F.; Bobodzhanov, A.A. Singularly Perturbed Fredholm-Type Integro-Differential Equations and Systems with Internal Transition Layers; Sputnik+ Publishing House: Moscow, Russia, 2018. [Google Scholar]
Disclaimer/Publisher’s Note: The statements, opinions and data contained in all publications are solely those of the individual author(s) and contributor(s) and not of MDPI and/or the editor(s). MDPI and/or the editor(s) disclaim responsibility for any injury to people or property resulting from any ideas, methods, instructions or products referred to in the content. |
© 2023 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https://creativecommons.org/licenses/by/4.0/).