Results on exact controllability of second-order semilinear control system in Hilbert spaces

*Correspondence: n.sooppy@psau.edu.sa; sh.rezapour@azaruniv.ac.ir; sh.rezapour@mail.cmuh.org.tw; rezapourshahram@yahoo.ca 4Department of Mathematics, College of Arts and Sciences, Prince Sattam bin Abdulaziz University, Wadi Aldawaser 11991, Saudi Arabia 5Department of Mathematics, Azarbaijan Shahid Madani University, Tabriz, Iran Full list of author information is available at the end of the article Abstract


Introduction
Differential equations arise in many areas of science and technology, specifically whenever a deterministic relation involving some continuously varying quantities (modeled by functions) and their rates of change in space and/or time is known or postulated. This is illustrated in classical mechanics where the motion of a body is described by its position and velocity as time varies. For the studies related to the existence of solution for integerand fractional-order systems, one can refer to . The concept of controllability is one of the underlying ideas in mathematical control theory. Controllability analysis is used in several real-life problems which include, but are not limited to, rocket launching problems for satellite and aircraft control, missiles and anti-missiles problems in defense, regulating inflation rate in the economy, controlling sugar level in the blood, etc. A systematic study of controllability was initiated by Kalman [31] in 1963 when the theory of controllability for time-invariant and time-varying control systems in state-space form was developed.
Several engineering and scientific problems can be expressed by infinite-dimensional differential equations. Therefore, it becomes necessary to discuss the controllability results for infinite-dimensional systems. The controllability problems for finite-dimensional nonlinear systems have been immensely analyzed by several authors. Many authors have advanced the idea of controllability from finite-dimensional systems to infinitedimensional systems and determined appropriate requirements for the controllability of nonlinear systems. Different techniques have been practiced for the examination of controllability including fixed point theorems [32][33][34]. For more problems on controllability and recent progresses on fractional calculus and its applications, please refer to [35][36][37][38][39][40][41][42][43][44][45][46][47].
Second-order differential equations represent abstract mathematical interpretations of several partial differential equations which occur in many applications related to the oscillation of fastened bars, the transverse motion of an extensible beam, and several other realworld physical phenomena. Hence it becomes really important to determine the controllability outcomes for this kind of system. Controllability discussions for various secondorder nonlinear systems have been widely studied by several authors [33,[48][49][50][51][52][53][54][55][56][57][58]. Fixed point theorems have been utilized greatly in determining the existence and controllability results for different first-order and second-order systems which involve large estimations on system constants see [33,59,60]. Recently Bashirov [1] obtained the exact controllability results for first-order semilinear systems using a new technique that is based on the piecewise formulation of driving controls and without using fixed point theory. Earlier the same approach was applied in terms of approximate controllability [61]. Motivated by Bashirov [1], we extend the controllability results for second-order semilinear systems excluding the use of fixed point theory. It is centered on a piecewise formulation of steering controls and does not involve large estimations which are required to apply fixed point theorems. By constructing a piecewise sequence of controls, we determine that a secondorder semilinear system is exactly controllable to the domain of the system operator, that is, it drives the system from any starting position to the required final position from the domain of the system. To achieve the required results, we suppose the subsequent requirements: (a) The corresponding linear system is exactly controllable at every non-initial time period.
1 p(t) represents the state having values in Hilbert space X.
3 B is a bounded and linear operator from U into X. 4 The function r : [0, b] × X → X is a purely nonlinear function which produces nonlinearity in the system. 5 A : dom(A) ⊆ X → X is linear, closed where dom(A) is a dense subset of X. The linear system corresponding to (1.1) with state vector q(t) and control v is defined by The article is structured in the subsequent manner: 1 Sect. 2 presents a few basic results related to control theory and second-order systems. 2 Sect. 3 provides the assumptions which are required to obtain the controllability results. 3 Sect. 4 discusses the controllability results using the new technique. 4 Sect. 5 presents an example to verify the established outcomes.

Auxiliary results
Here, we will review fundamental theories and a few definitions which would be helpful for further discussions.
If {C(t), t ∈ R} is a strongly continuous cosine family in X, then {S(t), t ∈ R} is a oneparameter family of operators in X defined by The infinitesimal generator of a strongly continuous cosine family The domain of operator A is defined as x is a twice continuously differentiable function of t .
These cosine and sine families defined above and generator A fulfill the following properties.

Lemma 2.2 ([28]) Suppose that A is the infinitesimal generator of a cosine family of operators {C(t)
: t ∈ R}. Then the following hold: The uniform boundedness principle together with (1)  1 with domain x ∈ X for which this limit exists, is the infinitesimal generator of the cosine family {C(t), t ∈ R}.
, which is the set of admissible controls. We define the mild solution of the given semi linear system (1.1) and its corresponding linear system as follows.

Definition 2.5
The mild solution of system (1.1) is defined by a function p(·) ∈ X which satisfies the following integral equation: and the mild solution of the corresponding linear system (1.2) is described by the following integral equation: (p F , q F ) ∈ X and > 0, there exists a control function v ∈ U ad such that the solution of (1.1) satisfies where p(b) is the state value of system (1.1) at time t = b. If p(b) = p F , then the system is said to be exactly controllable. The system is said to be exactly controllable to dom(A) on [0, b] if, for the given starting position (p 0 , q 0 ) ∈ X and the required final position (p F , q F ) ∈ dom(A), there exists a control function v ∈ U ad such that the solution of the system satisfies Remark 2.7 Note that the exact controllability to dom(A) lies in between the exact and approximate controllability. Therefore, it is a weaker concept than the exact controllability. In real life applications, sometimes we are more concerned with attaining the points from dom(A). If it is possible to reach the points from X\ dom(A) as well, then it can be considered as an additional capability of the system.

Assumptions
Let us introduce the controllability Grammian operator W associated with linear system (1.2) by where S * (s) denotes the adjoint of S(s).

Theorem 3.1 The corresponding linear system (1.2) is exactly controllable on the interval
The control which drives the system from the starting position (q(h), q (h)) ∈ X to the final position (p F , q F ) ∈ X is given by Proof The result can be seen in [59]. Moreover, it can be easily verified by substituting the above defined control in the mild solution of the corresponding linear system that it

Remark 3.2 The coercivity of W (t) indicates that (W (t)) -1 is a bounded linear operator.
We say that W (t) is coercive if there exists γ > 0 such that W (t)x, x ≥ γ x 2 for all x ∈ X.
Here W (0) = 0 and, therefore, it fails to be coercive. But it may be coercive for 0 < t ≤ b. Therefore, the above result holds on [h, b].
To determine the main result, we make the following assumptions on the controllability Grammian operator W and the nonlinear function r(t, p): That is, (W (t)) -1 → ∞ as t → 0 + with the slower rate in comparison to the reciprocal of the square function as for small values of t. (III) The nonlinear function r is Lebesgue measurable in t.

Results on controllability
In this section, we primarily focus on the study of exact controllability of the assumed system.
which steers the initial state p 0 to C(h 1 )p F + S(h 1 )q F . That is, Define v on [b 0 , b 1 ] by letting v( ) = v 1 ( ). Then, from (2.1), we obtain For brevity, let p(b 1 ) = p 1 . Next, consider (1.2) on [b 1 , b 2 ]. By Theorem 3.1, the control steers p 1 to C(h 2 )p F + S(h 2 )q F . Writing p (b 1 ) = q 1 , then the control u 2 ( ) can be written as That is, Define v on (b 1 , b 2 ] by letting v( ) = v 2 ( ). Then, from (2.1), we obtain For the sake of convenience, let p(b 2 ) = p 2 . Progressing in this fashion, we acquire a sequence of driving controls where q n-1 = p (b n-1 ). After combining the above sequence of controls, we get the control function as follows: Now by using the assumption (V), we get where M = sup [0,b] C( ) and K = sup [0,b]×X r( , p) . Since C( ) is strongly continuous, S(0) = 0 and lim n→∞ h n = 0. Therefore, lim n→∞ p n = p F . Also, Thus, we have Since C( ) is strongly continuous, S(0) = 0 and lim n→∞ h n = 0.
Therefore, lim n→∞ q n = q F .
Next we prove that v ∈ U ad . Since every v n of v is continuous on the interval (b n-1 , b n ] for n = 0, 1, 2, . . . , hence v is measurable. Also, Since C(h 0 )p F = p F and KMh 0 = 0, we obtain Since lim →0 for some P, P > 0. Then, by using assumption (II), where q F ≤ C and Ap F ≤ L for some C, L > 0. Since 0 < α < 1, therefore the series This shows that v ∈ U ad . Therefore, p is continuous and
Define the controllability operator in the following way: Let us choose the system operator A in such a way that W ( ) is coercive for all 0 < ≤ b and there exists some N ≥ 0 such that 1+α W ( ) -1 ≤ N ∀0 < ≤ b and 0 ≤ α < 1.
Also, the nonlinear function σ can be considered satisfying conditions (III)-(V). The considered PDE (5.1) can be converted to (1.1). Therefore, system (5.1) is exactly controllable to dom(A).

Conclusion
In the present manuscript, the exact controllability to dom(A) for a second-order semilinear system has been discussed using a new technique which avoids fixed point theorems and does not involve large estimations on the system constants. The control function has been formulated by the piecewise construction of steering controls. These results can be further extended for systems with delay or deviated arguments with impulses and fractional-order systems.