Skip to main content

Theory and Modern Applications

On the finite approximate controllability for Hilfer fractional evolution systems


In this paper, we consider the finite approximate controllability of some Hilfer fractional evolution systems. Using a variational approach and Schauder’s fixed point theorem, we give sufficient conditions for finite approximate controllability of semilinear controlled systems. An example is given to illustrate our theory.

1 Introduction

In this paper, we investigate the Hilfer fractional evolution system:

$$\begin{aligned} \left \{ \textstyle\begin{array}{l} D^{\nu,\mu}_{0^{+}}x(t)= Ax(t)+f(t,x(t))+Bu(t),\quad t\in J^{\prime}=(0,b],\\ I^{(1-\nu)(1-\mu)}_{0^{+}}x(0)=x_{0}, \end{array}\displaystyle \right . \end{aligned}$$

where \(D^{\nu,\mu}_{0^{+}}\) represents the Hilfer fractional derivative, \(0\leq\nu\leq1\), \(\frac{1}{2}<\mu<1\), \(x(\cdot)\) is assumed to be in a Hilbert space H, \(I^{(1-\nu)(1-\mu)}_{0^{+}}\) is the Riemann–Liouville fractional integral of order \((1-\nu)(1-\mu)\), \(A:D(A)\subseteq H\rightarrow H\) is the infinitesimal generator of a compact, uniformly bounded and \(C_{0}\)-semigroup \(\{T(t),t\geq0\}\) on a separable Hilbert space H. Now \(f:J^{\prime} \times H\rightarrow H\) is a given function that will be specified later. The control function u is taken in \(L^{2}(J^{\prime},U)\) and the admissible controls set U is a Hilbert space, B is a bounded linear operator from U into H, and finally, \(x_{0}\) is an element of H.

Fractional calculus and fractional dynamic equations [1, 2] arise naturally in phenomena in engineering, physics, science and controllability. For recent work on the existence of mild solutions, controllability and optimal control for some fractional evolution systems we refer the reader to [3, 4], and for approximate controllability of some linear and nonlinear systems see [57] and the references therein. Hilfer [8] consider a generalized Riemann–Liouville fractional derivative called the Hilfer fractional derivative and in [9] the approximate controllability of Hilfer fractional differential inclusions with nonlocal conditions was investigated. Existence, nonexistence, uniqueness involving Hilfer fractional derivatives was discussed in [1012] and in [13] the approximate controllability of fractional evolution equations involving Hilfer fractional derivatives was considered and in [14], we study the approximate controllability of Hilfer fractional evolution hemivariational inequalities by two resolvent operators and fixed point theorem. Compared with approximate controllability, finite approximate controllability is a stronger concept, and it is the consequence of approximate controllability in some linear heat equations. There are a number of papers on finite approximate controllability of differential systems. In [15], semilinear variational inequalities with distributed controls were studied, in [16] the author presented a finite-dimensional version of null controllability for the semilinear heat equation in bounded domains with Dirichlet boundary conditions, the author in [17] investigated finite approximate controllability for a nonlocal parabolic problem, and in [18, 19] the author considered approximate controllability and finite approximate controllability of some semilinear abstract equation, and finite approximate controllability for Sobolev-type nonlocal fractional semilinear evolution equations in Hilbert spaces.

There are only a few papers on finite approximate controllability of fractional evolution systems and motivated from the above (in particular [18, 19]), we will study the finite approximate controllability of some Hilfer fractional evolution systems. In Sect. 2, we present some preliminaries on fractional calculus and the definition of finite approximate controllability. In Sect. 3, sufficient conditions are given for the existence of mild solutions of system (1.1). In Sect. 4, by using the treatment in [14] and the variational method, the finite approximate controllability of system (1.1) is discussed. In Sect. 5, an example is given to illustrate the theory.

2 Preliminaries

Let \(J=[0,b]\) and E be a Banach space with norm \(\|\cdot\|_{E}\) (we usually write it as \(\|\cdot\|\)). Now \(E^{*}\) denotes its dual and \(\langle\cdot,\cdot\rangle_{E}\) denotes the duality pairing between \(E^{*}\) and E. We use \(L_{b}(E,E)\) to denote the space of bounded linear operators with the norm \(\|\cdot\|_{L_{b}(E,E)}\). Let \(C(J,E)\) be the Banach space of all continuous functions from J into E. Set \(\gamma=\nu+\mu-\nu\mu\), \(0<\gamma<1\), and then \(1-\gamma=(1-\nu )(1-\mu)\). Define

$$Y:=C_{1-\gamma}\bigl(J^{\prime},H\bigr)=\Bigl\{ x\in C \bigl(J^{\prime},H\bigr):\lim_{t\rightarrow 0^{+}}t^{1-\gamma}x(t) \textrm{ exist and finite}\Bigr\} $$

endowed with the norm \(\|x\|_{Y}=\textrm{sup}_{t\in J^{\prime}}\|t^{1-\gamma }x(t)\|_{H}\). Clearly, \((Y,\|\cdot\|_{Y})\) is a Banach space.

For brevity, let \(L^{p}_{H}=L^{p}(J, H)\), \(L^{p}_{R^{+}}=L^{p}(J, R^{+})\) and \(L^{p}_{U}=L^{p}(J, U)\) for \(1\leq p<\infty\).

We collect some definitions on fractional calculus of Riemann–Liouville type, Caputo type and Hilfer type; For more details, see [10, 12, 2022].

Definition 2.1

For a given integral function \(f:[a,\infty)\rightarrow E\), the integral

$$I^{\alpha}_{a^{+}}f(t)=\frac{1}{\varGamma(\alpha)} \int ^{t}_{a}(t-s)^{\alpha-1}f(s)\,ds,\quad t>a, \alpha>0, $$

is called the right-side Riemann–Liouville fractional integral of order α, where Γ is the gamma function.

Definition 2.2

The right-side Riemann–Liouville fractional derivative of order \(\alpha \in{(n-1,n)}\), \(n\in Z^{+}\) for a function \(f:[a,\infty)\rightarrow E\) is defined by

$${}^{\mathrm{RL}}D^{\alpha}_{a^{+}}f(t)=\frac{1}{\varGamma(n-\alpha)} \biggl( \frac {d}{dt} \biggr)^{(n)} \int^{t}_{a}(t-s)^{n-\alpha-1}f(s)\,dt,\quad t>a. $$

Definition 2.3

The right-side Hilfer fractional derivative of order ν, μ (\(0\leq \nu\leq1\), \({0<\mu<1}\)) for a function \(f:[a,\infty)\rightarrow E\) is defined by

$$D^{\nu,\mu}_{a^{+}}f(t)=I^{\nu(1-\mu)}_{a^{+}} \biggl( \frac{d}{dt}I^{(1-\nu )(1-\mu)}_{a^{+}}f(t) \biggr),\quad t>a. $$

Definition 2.4

The right-side Caputo’s fractional derivative of order \(\alpha\in (n-1,n)\), \(n\in Z^{+}\) for a function \(f:[a,\infty)\rightarrow E\) is defined by

$${}^{C}D^{\alpha}_{a^{+}}f(t)=\frac{1}{\varGamma(n-\alpha)} \int ^{t}_{a}(t-s)^{n-\alpha-1}f^{(n)}(s) \,dt,\quad t>a. $$

Remark 2.5

  1. (i)

    When \(\nu=0\), \(0<\mu<1\), and \(a=0\), the right-side Hilfer fractional derivative corresponds to the classical right-side Riemann–Liouville fractional derivative:

    $$D^{0,\mu}_{0^{+}}f(t)=\frac{d}{dt}I^{(1-\mu)}_{a^{+}}f(t)={}^{\mathrm{RL}}D^{\mu}_{0^{+}}f(t). $$
  2. (ii)

    When \(\nu=1\), \(0<\mu<1\), and \(a=0\), the right-side Hilfer fractional derivative corresponds to the classical right-side Caputo’s fractional derivative:

    $$D^{1,\mu}_{0^{+}}f(t)=I_{0^{+}}^{(1-\mu)} \frac{d}{dt}f(t)= {}^{C}D^{\mu}_{0^{+}}f(t). $$

Next we recall the definition of finite approximate controllability; see [15, 16]:

Definition 2.6

The system (1.1) is finite approximate controllable on \(J^{\prime}\), if \(x_{b}\in H\) and \(\epsilon>0\), there exists a control \(u_{\epsilon}\in L^{2}_{U}\), such that the solution \(x_{\epsilon}\) of system (1.1) satisfies the conditions:

$$\begin{aligned} \bigl\Vert x_{\epsilon}(b)-x_{b} \bigr\Vert < \epsilon \end{aligned}$$


$$\begin{aligned} \varPi_{\mathscr{E}}x_{\epsilon}(b)=\varPi_{\mathscr{E}}x_{b}, \end{aligned}$$

where \(\mathscr{E}\) is a finite-dimensional subspace of H and \(\varPi _{\mathscr{E}}\) is the orthogonal projection from H to \(\mathscr{E}\).

The following definition is based on [12, Definition 2.3] and [13, Definition 5].

Definition 2.7

For each \(u\in L^{2}_{U}\), a function \(x\in Y\) is a mild solution of (1.1) if \(I^{(1-\nu)(1-\mu)}_{0^{+}} x(0)=x_{0}\) and

$$\begin{aligned} x(t)= \mathscr{L}_{\nu,\mu}(t)x_{0}+ \int_{0}^{t}\mathscr{T}_{\mu}(t-s) \bigl[f \bigl(s,x(s)\bigr)+Bu(s) \bigr]\,ds,\quad t\in J^{\prime}, \end{aligned}$$


$$\mathscr{P}_{\mu}(t):= \int_{0}^{\infty} \mu\theta M_{\mu}(\theta )T \bigl(t^{\mu}\theta\bigr)\,d\theta,\qquad \mathscr{T}_{\mu}(t):=t^{\mu-1} \mathscr {P}_{\mu}(t), \qquad\mathscr{L}_{\nu,\mu}(t):=I^{\nu(1-\mu)}_{0^{+}} \mathscr {T}_{\mu}(t), $$

and \(M_{\mu}(\theta)\) is the M-Wright function defined by

$$M_{\mu}(\theta)=\sum_{n=1}^{\infty} \frac{(-\theta )^{n-1}}{(n-1)!\varGamma(1-\mu n)},\quad 0< \mu< 1, \theta\in\mathcal{C}, $$

and it satisfies \(M_{\mu}(\theta)>0\), \(\int_{0}^{\infty}M_{\mu}(\theta )\,d\theta=1\) and \(\int_{0}^{\infty}\theta^{\delta}M_{\mu}(\theta)\,d\theta=\frac{\varGamma (1+\delta)}{\varGamma(1+\mu\delta)}\), \(\delta\in(-1,\infty)\).

We assume \(T(t)\) (\(t\geq0\)) is uniformly bounded, so there exists \(M>1\) with \(\sup_{t\in[0,\infty)}\|T(t)\|\leq M\).

Lemma 2.8

The operators \(\mathscr{T}_{\mu}(\cdot)\)and \(\mathscr{L}_{\nu,\mu }(\cdot)\)have the following properties:

  1. (i)

    ([12, Proposition 2.16]) For any fixed \(t> 0\), \(\mathscr {T}_{\mu}(t)\)and \(\mathscr{L}_{\nu,\mu}(t)\)are linear and bounded operators, i.e., for any \(x \in H\),

    $$\bigl\Vert \mathscr{T}_{\mu}(t)x \bigr\Vert _{H} \leq \frac{Mt^{\mu-1}}{\varGamma (\mu)} \Vert x \Vert _{H} \quad\textit{and}\quad \bigl\Vert \mathscr{L}_{\nu,\mu}(t)x \bigr\Vert _{H} \leq\frac {Mt^{\gamma-1}}{\varGamma(\gamma)} \Vert x \Vert _{H},\quad \gamma=\nu+\mu-\nu\mu. $$
  2. (ii)

    \(\{\mathscr{T}_{\mu}(t),t> 0\}\)and \(\{\mathscr{L}_{\nu,\mu }(t),t> 0\}\)are compact if \(T(t)\)is compact, \(t>0\).

Remark 2.9

From Lemma 2.8(ii), we see that \(\mathscr{T}_{\mu}(\cdot)\) and \(\mathscr{L}_{\nu,\mu}(\cdot)\) are continuous in the uniform operator topology for \(t>0\), i.e.,

$$\bigl\| \mathscr{T}_{\mu}(t_{2})-\mathscr{T}_{\mu}(t_{1})\bigr\| _{L_{b}(H,H)}\to0,\quad\quad \bigl\| \mathscr{L}_{\nu,\mu}(t_{2})-\mathscr{L}_{\nu,\mu}(t_{1})\bigr\| _{L_{b}(H,H)}\to 0$$

as \(t_{2}\to t_{1}\).

3 Existence of mild solutions

Consider the following assumptions:


\(f:J^{\prime}\times H\rightarrow H\) is a function such that:

  1. (i)

    the function \(t\mapsto f(t,x)\) is measurable for all \(x\in H\);

  2. (ii)

    the function \(x\mapsto f(t,x)\) is continuous for \(t\in J^{\prime}\);

  3. (iii)

    for each \(r>0\), there exists a positive integrable function \(\varPhi_{r}(t):J^{\prime}\rightarrow(0,+\infty)\) such that

    $$\sup_{ \Vert x \Vert _{Y}\leq r} \bigl\Vert f\bigl(t,x(t)\bigr) \bigr\Vert \leq \varPhi_{r}(t) \quad\hbox{for a.e. } t\in J^{\prime} \hbox{ (here $x \in Y$)} $$


    $$\lim_{r\rightarrow+\infty} \inf\frac{ \Vert \varPhi_{r} \Vert _{L^{2}_{R^{+}}}}{r}=\rho< +\infty; $$
  4. (iv)

    the inequality

    $$\frac{Mb^{\frac{1}{2}+\mu-\gamma}}{\sqrt{2\mu-1}\varGamma(\mu)}\rho< 1 $$



the linear fractional control system

$$\begin{aligned} \left \{ \textstyle\begin{array}{l} D^{\nu,\mu}_{0^{+}}x(t)= Ax(t)+Bu(t) , \\ I^{(1-\nu)(1-\mu)}_{0^{+}}x(0)=x_{0}, \end{array}\displaystyle \right . \end{aligned}$$

is approximately controllable on \(J^{\prime}\).

Next, take into account two relevant operators:

$$\varGamma_{0}^{b}= \int_{0}^{b}\mathscr{T}_{\mu}(b-s)BB^{*} \mathscr {T}_{\mu}^{*}(b-s)\,ds, $$


$$R_{\epsilon}^{b}=\bigl(\epsilon I+\varGamma_{0}^{b} \bigr)^{-1},\quad \epsilon>0, $$

where I denotes the identity operator, \(B^{*}\) denotes the adjoint of B and \(\mathscr{T}_{\mu}^{*}(\cdot)\) is the adjoint of \(\mathscr {T}_{\mu}(\cdot)\).

Let \(\epsilon>0\), \(y\in Y\) and \(x_{b} \in H\). We consider the functional \(\mathcal{J_{\epsilon}}(\cdot;y):H \rightarrow R\) defined by

$$\begin{aligned} \mathcal{J_{\epsilon}}(\varPsi;y)=\epsilon \bigl\Vert (I- \varPi_{\mathscr {E}})R_{\epsilon}^{b}\varPsi \bigr\Vert _{H}+\frac{1}{2} \int_{0}^{b} \bigl\Vert B^{*} \mathscr{T}_{\mu}^{*}(b-t)R_{\epsilon}^{b} \varPsi \bigr\Vert _{H}^{2}\,dt-\bigl\langle \mathscr{H}(y),R_{\epsilon}^{b}\varPsi\bigr\rangle , \end{aligned}$$


$$\mathscr{H}(y)=x_{b}-\mathscr{L}_{\nu,\mu}(b)x_{0} - \int_{0}^{b}\mathscr{T}_{\mu}(b-s)f \bigl(s,y(s)\bigr)\,ds. $$

We claim (and we will prove it after Lemma 3.3) that, for any \(y\in Y\), the functional \(\mathcal{J_{\epsilon}}(\cdot ;y)\) admits a unique minimum \(\widehat{\varPsi_{\epsilon}}\) which defines a map \(\mathscr {F}_{\epsilon}:Y \to H\) given by \(\mathscr{F}_{\epsilon}:y \to\widehat {\varPsi_{\epsilon}}\). Now let (here \(x\in Y\))

$$u_{\epsilon}(s,x)=B^{*}\mathscr{T}_{\mu}^{*}(b-s)R_{\epsilon }^{b} \mathscr{F}_{\epsilon}(x) $$


$$(\digamma_{\epsilon} x) (t)= \mathscr{L}_{\nu,\mu}(t)x_{0}+ \int_{0}^{t}\mathscr{T}_{\mu}(t-s) \bigl[f \bigl(s,x(s)\bigr)+Bu_{\epsilon}(s,x) \bigr]\,ds. $$

For \(r>0\), let \(B^{(1-\gamma)}_{r}(J^{\prime})=\{x\in Y:\|x\|_{Y}\leq r\}\) and \(B_{r}(J)=\{x\in C(J,H):\|x\|_{C}\leq r\}\).

Lemma 3.1

The set \(\mathcal{H}=\{\mathscr{H}(y):y\in B^{(1-\gamma)}_{r}(J^{\prime})\} \) is relatively compact in Y


The proof is similar to that in step 4 in the proof of Theorem 3.6. □

Lemma 3.2

\(\mathscr{H}:B^{(1-\gamma)}_{r}(J^{\prime})\rightarrow H\)is a continuous function.


The proof is similar to step 2 in the proof of Theorem 3.6. □

Lemma 3.3

Let \(\epsilon>0\)and \(r>0\). Then with \(\epsilon_{1}= R_{\epsilon }^{b}\epsilon\), we have

$$\mathop{\underline{\lim} } _{ \Vert \varPsi \Vert _{H}\rightarrow+\infty} \inf_{y\in B^{(1-\gamma)}_{r}(J^{\prime})} \frac{\mathcal{J_{\epsilon}}(\varPsi ;y)}{ \Vert \varPsi \Vert _{H}}\geq\epsilon_{1}. $$


We follow the argument in [16, 18]. Suppose it is false. Then there exist sequences \(\{\varPsi_{n}\}\subset H\), \(\{ y_{n}\}\subset B^{(1-\gamma)}_{r}(J^{\prime})\) with \(\|\varPsi_{n}\|\rightarrow +\infty\) and

$$\begin{aligned} \mathop{\underline{\lim} } _{n\rightarrow+\infty} \frac{\mathcal {J_{\epsilon}}(\varPsi_{n};y_{n})}{ \Vert \varPsi_{n} \Vert _{H}}< \epsilon_{1}. \end{aligned}$$

Normalize with \(\widehat{\varPsi}_{n}=\frac{\varPsi_{n}}{\|\varPsi_{n}\|}\) (note \(\|\widehat{\varPsi}_{n}\|=1\)). The set \(\{\mathscr{H}(y): y\in B^{(1-\gamma)}_{r}(J^{\prime}) \}\) is relatively compact in H (see the argument later in Theorem 3.6), so without loss of generality assume \(\mathscr{H}(y_{n}) \xrightarrow{\text{strongly}} h\) in H for some \(h\in H\). Choose a subsequence which we will still denote by \(\widehat{\varPsi }_{n}\) with \(\widehat{\varPsi}_{n}\xrightarrow{\text{weakly}}\widehat{\varPsi}\) for \(\widehat{\varPsi} \in H\). From the compactness of \(\mathscr{T}_{\mu }(t)\) one has

$$B^{*}\mathscr{T}_{\mu}^{*}(b-t)R_{\epsilon}^{b} \widehat{\varPsi }_{n}\xrightarrow{\text{strongly}} B^{*} \mathscr{T}_{\mu}^{*}(b-t)R_{\epsilon}^{b} \widehat{\varPsi}. $$

From (3.1) we get

$$\begin{aligned} \begin{aligned}\frac{\mathcal{J_{\epsilon}}(\varPsi_{n};y_{n})}{ \Vert \varPsi_{n} \Vert _{H}^{2}}&=\frac {\epsilon}{ \Vert \varPsi_{n} \Vert _{H}} \bigl\Vert (I- \varPi_{\mathscr{E}})R_{\epsilon}^{b}\widehat{ \varPsi}_{n} \bigr\Vert _{H}+\frac{1}{2} \int_{0}^{b} \bigl\Vert B^{*}\mathscr {T}_{\mu}^{*}(b-t)R_{\epsilon}^{b}\widehat{ \varPsi}_{n} \bigr\Vert _{H}^{2}\,dt\\ &\quad- \frac {1}{ \Vert \varPsi_{n} \Vert _{H}}\bigl\langle \mathscr{H}(y_{n}),R_{\epsilon}^{b} \widehat {\varPsi}_{n}\bigr\rangle ,\end{aligned} \end{aligned}$$

and from Fatou’s lemma, as \(\|\varPsi_{n}\|\rightarrow+\infty\), we have

$$\int_{0}^{b} \bigl\Vert B^{*} \mathscr{T}_{\mu}^{*}(b-t)R_{\epsilon}^{b} \widehat {\varPsi} \bigr\Vert _{H}^{2}\,dt\leq\mathop{ \underline{\lim} } _{n\rightarrow +\infty}\inf \int_{0}^{b} \bigl\Vert B^{*} \mathscr{T}_{\mu}^{*}(b-t)R_{\epsilon}^{b} \widehat{\varPsi}_{n} \bigr\Vert _{H}^{2}\,dt=0, $$

which implies \(\widehat{\varPsi}_{n}\xrightarrow{\text{weakly}}0\) in H. Since \(\mathscr{E}\) is finite dimensional and we obtain \(\varPi_{\mathscr{E}} R_{\epsilon}^{b}\widehat{\varPsi}_{n} \xrightarrow{\text{strongly}} 0 \) in H and so

$$\begin{gathered} \bigl\Vert (I-\varPi_{\mathscr{E}})R_{\epsilon}^{b}\widehat{ \varPsi}_{n} \bigr\Vert _{H}=\sqrt { \bigl\Vert R_{\epsilon}^{b}\widehat{\varPsi}_{n} \bigr\Vert _{H}^{2} + \bigl\Vert \varPi_{\mathscr{E} }R_{\epsilon}^{b} \widehat{\varPsi}_{n} \bigr\Vert _{H}^{2} } \to \bigl\Vert R_{\epsilon}^{b}\widehat{\varPsi} \bigr\Vert _{H}, \\ \frac{\mathcal{J_{\epsilon}}(\varPsi_{n};y_{n})}{ \Vert \varPsi_{n} \Vert _{H}}=\epsilon \bigl\Vert (I-\varPi_{\mathscr{E}})R_{\epsilon}^{b} \widehat{\varPsi }_{n} \bigr\Vert _{H}+\frac{ \Vert \varPsi_{n} \Vert _{H}}{2} \int_{0}^{b} \bigl\Vert B^{*} \mathscr{T}_{\mu }^{*}(b-t)R_{\epsilon}^{b} \widehat{\varPsi}_{n} \bigr\Vert _{H}^{2}\,dt\\ \phantom{\frac{\mathcal{J_{\epsilon}}(\varPsi_{n};y_{n})}{ \Vert \varPsi_{n} \Vert _{H}}=}{}- \bigl\langle \mathscr {H}(y_{n}),R_{\epsilon}^{b} \widehat{\varPsi}_{n}\bigr\rangle ,\end{gathered} $$

and as a result

$$\mathop{\underline{\lim} } _{n\rightarrow+\infty} \frac{\mathcal {J_{\epsilon}}(\varPsi_{n};y_{n})}{ \Vert \varPsi_{n} \Vert _{H}}\geq\mathop { \underline{\lim} } _{n\rightarrow+\infty} \bigl(\epsilon \bigl\Vert (I-\varPi _{\mathscr{E}})R_{\epsilon}^{b}\widehat{\varPsi}_{n} \bigr\Vert _{H}-\bigl\langle \mathscr {H}(y_{n}),R_{\epsilon}^{b} \widehat{\varPsi}_{n}\bigr\rangle \bigr)=R_{\epsilon}^{b} \epsilon=\epsilon_{1}, $$

which contradicts (3.2). □

For \(y\in B^{(1-\gamma)}_{r}(J^{\prime})\) we have

$$\mathop{\underline{\lim} } _{ \Vert \psi \Vert _{H}\rightarrow+\infty} \frac {\mathcal{J_{\epsilon}}(\varPsi;y)}{ \Vert \varPsi \Vert _{H}}\geq \epsilon_{1}. $$

Also note for any \(y\in Y\) the map \(\varPsi\to \mathcal{J_{\epsilon }}(\varPsi;y)\) is continuous and strictly convex. Now for \(y\in B^{(1-\gamma)}_{r}(J^{\prime})\) let \(\{\varPsi_{\epsilon,n}\}\) be a minimizing sequence of \(\mathcal{J_{\epsilon}}(\cdot ;y)\) and we suppose without loss of generality (note from the above this sequence is bounded) \(\varPsi_{\epsilon,n}\) converges weakly to \(\widehat{\varPsi _{\epsilon}}\) in H. Now

$$\mathcal{J_{\epsilon}}(\widehat{\varPsi_{\epsilon}};y)\leq\mathop { \underline{\lim} } _{n\rightarrow+\infty} \mathcal{J_{\epsilon }}( \varPsi_{\epsilon,n};y)=\inf_{\varPsi\in H} \mathcal{J_{\epsilon }}( \varPsi:y). $$

Thus \(\widehat{\varPsi_{\epsilon}}\) is a minimum and from the convexity of \(\mathcal{J_{\epsilon}}(\cdot;y)\) the minimum is unique. Now we define a map \(\mathscr{F}_{\epsilon}:y\rightarrow\widehat {\varPsi_{\epsilon}}\) (which is the proof of the claim after (3.1)).

Lemma 3.4

For all \(y\in B^{(1-\gamma)}_{r}(J^{\prime})\), there exists \(R_{\epsilon}(r)>0\), such that \(\|\mathscr{F}_{\epsilon}(y)\|\leq R_{\epsilon}(r)\).


From Lemma 3.3, we see that there exists \(R_{\epsilon}(r)>0\) such that

$$\Vert \varPsi \Vert _{H}>R_{\epsilon}(r), \qquad\inf _{y\in B^{(1-\gamma )}_{r}(J^{\prime})}\frac{\mathcal{J_{\epsilon}}(\varPsi;y)}{ \Vert \varPsi \Vert _{H}}\geq\epsilon. $$

If \(\mathscr{F}_{\epsilon}\) is not bounded, we may as well suppose \(\| \mathscr{F}_{\epsilon}\|_{H}\geq R_{\epsilon}(r)\), such that

$$\begin{aligned} \inf_{y\in B^{(1-\gamma)}_{r}(J^{\prime})}\frac{\mathcal{J_{\epsilon }}(\mathscr{F}_{\epsilon};y)}{ \Vert \mathscr{F}_{\epsilon} \Vert _{H}}\geq \epsilon. \end{aligned}$$

But from the definition of the map \(\mathscr{F}_{\epsilon}\), we know

$$\mathcal{J_{\epsilon}}\bigl(\mathscr{F}_{\epsilon}(y);y\bigr)\leq \mathcal {J_{\epsilon}}\bigl(0^{+};y\bigr)=0, $$

which contradicts (3.3), thus, for all \(y\in B^{(1-\gamma )}_{r}(J^{\prime})\), we have \(\|\mathscr{F}_{\epsilon}(y)\|\leq R_{\epsilon}(r)\). □

Lemma 3.5

Suppose for any \(y,y_{n}\in B^{(1-\gamma)}_{r}(J^{\prime})\), \(y_{n}\rightarrow y\)inY. Then

$$\mathscr{F}_{\epsilon}(y_{n})\xrightarrow{\textit{strongly}} \mathscr{F}_{\epsilon}(y). $$


Assume that \(y_{n}\in B^{(1-\gamma)}_{r}(J^{\prime})\) be a subsequence and \(y_{n}\rightarrow y \) as \(n\rightarrow+\infty\). By the boundedness of \(\mathscr{F}_{\epsilon}(y_{n})\) denoted by \(\widehat{\varPsi_{\epsilon ,n}}\), one can suppose \(\widehat{\varPsi_{\epsilon,n}}\) converges weakly to \(\widehat{\varPsi}_{\epsilon}\), then

$$\begin{aligned} \mathcal{J_{\epsilon}}\bigl(\widetilde{\varPsi_{\epsilon}(y)};y\bigr)& \leq \mathcal{J_{\epsilon}}(\widehat{\varPsi}_{\epsilon};y_{n}) \leq\mathop {\underline{\lim} } _{n\rightarrow+\infty} \mathcal{J_{\epsilon }}\bigl( \widehat{\varPsi_{\epsilon,n}}(y);y_{n}\bigr)\leq \mathop{\overline{ \lim} } _{n\rightarrow+\infty}\mathcal {J_{\epsilon}}\bigl(\widehat{ \varPsi_{\epsilon,n}}(y);y_{n}\bigr) \\ &\leq {\lim_{x \to+\infty}} \mathcal{J_{\epsilon}}\bigl( \widetilde{\varPsi_{\epsilon }(y)};y_{n}\bigr)=\mathcal{J_{\epsilon}} \bigl(\widetilde{\varPsi_{\epsilon}(y)};y\bigr).\end{aligned} $$

From the above section, we know that \(\widetilde{\varPsi_{\epsilon }(y)}\) is the unique minimum point, thus \(\widetilde{\varPsi_{\epsilon }(y)}=\widehat{\varPsi}_{\epsilon}(y)\). Combining the compactness of \(B^{*}\mathscr{T}_{\mu}^{*}(b-t)R_{\epsilon}^{b}\), the continuity of function \(H(y)\) and \(\widehat{\varPsi_{\epsilon,n}}\xrightarrow {w}\widehat{\varPsi}_{\epsilon}\), we have

$$\begin{gathered} {\lim_{n \to+\infty}}\mathcal{J_{\epsilon}}(\widehat{ \varPsi_{\epsilon ,n}};y_{n})=\mathcal{J_{\epsilon}}(\widehat{ \varPsi}_{\epsilon};y)=\mathcal{J_{\epsilon}}(\widetilde{ \varPsi}_{\epsilon};y), \\ {\lim_{n \to+\infty}} \int_{0}^{b} \bigl\Vert B^{*} \mathscr{T}_{\mu }^{*}(b-t)R_{\epsilon}^{b} \widehat{\varPsi_{\epsilon,n}} \bigr\Vert _{H}^{2}\,dt= \int _{0}^{b} \bigl\Vert B^{*} \mathscr{T}_{\mu}^{*}(b-t)R_{\epsilon}^{b} \widehat{\varPsi }_{\epsilon}\bigr\Vert _{H}^{2}\,dt, \\ {\lim_{n \to+\infty}}\bigl\langle H(y_{n}), R_{\epsilon}^{b}\widehat{\varPsi _{\epsilon,n}},\bigr\rangle = \bigl\langle H(y),R_{\epsilon}^{b}\widehat{\varPsi }_{\epsilon}\bigr\rangle ,\end{gathered} $$

thus, it follows that

$${\lim_{n \to+\infty}}\epsilon \bigl\Vert (I-\varPi_{\mathscr{E}})R_{\epsilon}^{b}\widehat{\varPsi_{\epsilon,n}} \bigr\Vert _{H}= \epsilon \bigl\Vert (I-\varPi_{\mathscr {E}})R_{\epsilon}^{b} \widehat{\varPsi}_{\epsilon}\bigr\Vert _{H}. $$

Using the compactness of \(\varPi_{\mathscr{E}}\), we infer that

$${\lim_{n \to+\infty}} \Vert \widehat{\varPsi_{\epsilon,n}} \Vert _{H}= \Vert \widehat {\varPsi}_{\epsilon} \Vert _{H}, $$

which implies that \(\mathscr{F}_{\epsilon}(y_{n})\xrightarrow {\text{strongly}}\mathscr{F}_{\epsilon}(y)\). □

For any \(x\in Y\subset L^{2}(J^{\prime},H)\), we consider the map \(\digamma :Y\rightarrow Y \)

$$ \begin{aligned}\digamma(x)={}& {\biggl\{ } g\in Y:\\& g(t)= \mathscr{L}_{\nu,\mu}(t)x_{0}+ \int_{0}^{t}\mathscr{T}_{\mu}(t-s)f \bigl(s,x(s)\bigr)\,ds + \int_{0}^{t}\mathscr{T}_{\mu}(t-s)Bu(s) \,ds,t\in J^{\prime} {\biggr\} }. \end{aligned} $$

Clearly, \(\lim_{t\to0^{+}}t^{1-\gamma} g(t)=\frac{x_{0}}{\varGamma(\gamma)}\).

We will work with the operator \(P:=\cdot^{1-\gamma}\digamma\cdot^{\gamma -1}\) from \(B_{r}(J)\) to \(B_{r}(J)\) (i.e. for \(y\in B_{r}(J)\), \(P y(t)=t^{1-\gamma}\digamma(t^{\gamma-1}y(t))\)). If we prove that P has a fixed point \(y^{*}\), then Ϝ has a fixed point \(x^{*}=\cdot^{\gamma-1}y^{*}\).

In our next result let \(r>0\) be such that

$$\frac{M}{\varGamma(\gamma)} \Vert x_{0} \Vert _{H}+ \frac{Mb^{\frac{1}{2}+\mu-\gamma }}{\sqrt{2\mu-1}\varGamma(\mu)}\bigl( \Vert \varPhi_{r} \Vert _{L^{2}_{R^{+}}} + \Vert B \Vert _{L_{b}(U,H)} \Vert u \Vert _{L^{2}_{U}}\bigr) \leq r . $$

Theorem 3.6

Assume that condition \(H(f)\)holds. Then (1.1) has a mild solution in \(B^{(1-\gamma)}_{r}(J^{\prime})\).


We prove Ϝ has a fixed point in \(B^{(1-\gamma)}_{r}(J^{\prime})\) (i.e. P has a fixed point in \(B_{r}(J)\)). We divide the proof into four steps.

Step 1: \(P:B_{r}(J) \to B_{r}(J)\).

Let \(z\in B_{r}(J)\) (and \(x=\cdot^{\gamma-1}z\) so \(x\in B^{(1-\gamma )}_{r}(J^{\prime})\)). Now

$$P z(t)= t^{1-\gamma}\mathscr{L}_{\nu,\mu}(t)x_{0}+t^{1-\gamma} \int _{0}^{t}\mathscr{T}_{\mu}(t-s) \bigl[f\bigl(s,s^{\gamma-1}z(s)\bigr)+Bu(s) \bigr]\,ds,\quad t\in J, $$


$$\begin{aligned} P z(t)= t^{1-\gamma}\mathscr{L}_{\nu,\mu}(t)x_{0}+t^{1-\gamma} \int _{0}^{t}\mathscr{T}_{\mu}(t-s) \bigl[f\bigl(s,x(s)\bigr)+Bu(s) \bigr]\,ds, \quad t\in J. \end{aligned}$$

From \(H(f)\)(iii), Lemma 2.8(i) and Hölder’s inequality we have for \(t\in J\)

$$\begin{aligned} \bigl\Vert P z(t) \bigr\Vert \leq& \frac{M}{\varGamma(\gamma)} \Vert x_{0} \Vert _{H}+\frac{Mt^{1-\gamma }}{\varGamma(\mu)} \int_{0}^{t}(t-s)^{\mu-1}\bigl[ \varPhi_{r}(s)+ \Vert B \Vert _{L_{b}(U,H)} \bigl\Vert u(s) \bigr\Vert _{U}\bigr]\,ds \\ \leq& \frac{M}{\varGamma(\gamma)} \Vert x_{0} \Vert _{H}+ \frac{Mb^{\frac{1}{2}+\mu -\gamma}}{\sqrt{2\mu-1}\varGamma(\mu)}\bigl( \Vert \varPhi_{r} \Vert _{L^{2}_{R^{+}}} + \Vert B \Vert _{L_{b}(U,H)} \Vert u \Vert _{L^{2}_{U}}\bigr) \leq r. \end{aligned}$$

Thus \(P:B_{r}(J) \to B_{r}(J)\).

Step 2: \(P:B_{r}(J) \to B_{r}(J)\) is continuous.

Let \(\{z_{n}\}\) be the sequence in \(B_{r}(J)\) with \(z_{n} \rightarrow z\) in \(B_{r}(J)\) as \(n\rightarrow+\infty\) (note \(\{x_{n}:=\cdot^{\gamma-1}z_{n}\} \) is a sequence in \(B^{(1-\gamma)}_{r}(J^{\prime})\) and \(x_{n}\rightarrow x\) as \(n\rightarrow+\infty\); here \(x=\cdot^{\gamma-1}z\)). Now

$$\begin{aligned} \bigl\Vert P z_{n}(t)-P z(t) \bigr\Vert \leq& t^{1-\gamma} \int_{0}^{t}\mathscr{T}_{\mu }(t-s) \bigl\Vert f\bigl(s,x_{n}(s)\bigr)-f\bigl(s,x(s)\bigr) \bigr\Vert \,ds, \end{aligned}$$

and a standard argument using the Lebesgue dominated convergence theorem guarantees that \(P:B_{r}(J) \to B_{r}(J)\) is continuous.

Step 3: \(\{P(z):z\in B_{r}(J)\}\) is equicontinuous.

Let \(z\in B_{r}(J)\) (so \(x=\cdot^{\gamma-1}z\)), \(0\leq\tau_{1}<\tau _{2}\leq b\) and \(\delta:=\tau_{2}-\tau_{1}>0\). Then

$$\begin{aligned} \bigl\Vert P z(\tau_{1})-P z(\tau_{2}) \bigr\Vert \leq Q_{1}+Q_{2}+Q_{3}+Q_{4}, \end{aligned}$$


$$\begin{aligned}& Q_{1}= \bigl\Vert \tau_{1}^{1-\gamma} \mathscr{L}_{\nu,\mu}(\tau_{1})x_{0}-\tau _{2}^{1-\gamma}\mathscr{L}_{\nu,\mu}(\tau_{2})x_{0} \bigr\Vert _{H}, \\& Q_{2}=(\tau_{2}-\tau_{1})^{1-\gamma} \int_{0}^{\tau_{1}} \bigl\Vert \mathscr{T}_{\mu }( \tau_{1}-s) \bigr\Vert \bigl( \bigl\Vert f\bigl(s,x(s)\bigr) \bigr\Vert _{H}+ \Vert B \Vert _{L_{b}(U,H)} \bigl\Vert u(s) \bigr\Vert _{U}\bigr)\,ds, \\& Q_{3}=\tau_{2}^{1-\gamma}\sup_{s\in[0,\tau_{1}]} \bigl\Vert \mathscr{T}_{\mu}(\tau _{1}-s)- \mathscr{T}_{\mu}(\tau_{2}-s) \bigr\Vert _{L_{b}(H,H)}\\& \phantom{Q_{3}=}{}\times \int_{0}^{\tau_{1}}\bigl( \bigl\Vert f\bigl(s,x(s)\bigr) \bigr\Vert _{H}+ \Vert B \Vert _{L_{b}(U,H)} \bigl\Vert u(s) \bigr\Vert _{U}\bigr)\,ds, \\& Q_{4}=\tau_{2}^{1-\gamma} \int_{\tau_{1}}^{\tau_{2}} \bigl\Vert \mathscr{T}_{\mu }( \tau_{2}-s) \bigr\Vert \bigl( \bigl\Vert f\bigl(s,x(s)\bigr) \bigr\Vert _{H}+ \Vert B \Vert _{L_{b}(U,H)} \bigl\Vert u(s) \bigr\Vert _{U}\bigr)\,ds. \end{aligned}$$


$$\begin{aligned} Q_{1} \leq& \bigl\Vert \tau_{1}^{1-\gamma} \mathscr{L}_{\nu,\mu}(\tau_{1})x_{0}-\tau _{2}^{1-\gamma}\mathscr{L}_{\nu,\mu}(\tau_{1})x_{0} \bigr\Vert _{H}+ \bigl\Vert \tau _{2}^{1-\gamma} \mathscr{L}_{\nu,\mu}(\tau_{1})x_{0}- \tau_{2}^{1-\gamma }\mathscr{L}_{\nu,\mu}(\tau_{2})x_{0} \bigr\Vert _{H} \\ \leq& (\tau_{2}-\tau_{1})^{1-\gamma}\frac{Mb^{\gamma-1}}{\varGamma (\gamma)} \Vert x_{0} \Vert _{H}+\tau_{2}^{1-\gamma} \bigl\Vert \mathscr{L}_{\nu,\mu}(\tau _{1})x_{0}- \mathscr{L}_{\nu,\mu}(\tau_{2})x_{0} \bigr\Vert _{L_{b}(H,H)} \\ := &Q_{11}+Q_{12}, \end{aligned}$$

\(Q_{11}\) tends to zero as \(\delta\rightarrow0\) and \(Q_{12}\) tends to zero as \(\delta\rightarrow0\) from Lemma 2.8(ii) and Remark 2.9. Thus, \(Q_{1}\) tends to zero as \(\delta\rightarrow0\).


$$\begin{aligned} Q_{2}&\leq(\tau_{2}-\tau_{1})^{1-\gamma} \int_{0}^{\tau_{1}} \bigl\Vert \mathscr {T}_{\mu}(\tau_{1}-s) \bigr\Vert \varPhi_{r}(s)+ \Vert B \Vert _{L_{b}(U,H)} \bigl\Vert u(s) \bigr\Vert _{U})\,ds \\ &\leq\delta^{1-\gamma} \int_{0}^{\tau_{1}}\frac{M(\tau_{1}-s)^{\mu -1}}{\varGamma(\mu)}\bigl( \varPhi_{r}(s)+ \Vert B \Vert _{L_{b}(U,H)} \bigl\Vert u(s) \bigr\Vert _{U}\bigr)\,ds \\ &\leq\delta^{1-\gamma} \biggl(\frac{M\tau_{1}^{\mu-\frac{1}{2}}}{\sqrt{2\mu -1}\varGamma(\mu)}\bigl( \Vert \varPhi_{r} \Vert _{L^{2}_{R^{+}}} + \Vert B \Vert _{L_{b}(U,H)} \Vert u \Vert _{L^{2}_{U}}\bigr) \biggr),\end{aligned} $$

which tends to zero as \(\delta\rightarrow0\).

In addition,

$$\begin{aligned} Q_{3} \leq&\tau_{2}^{1-\gamma}\sup_{s\in[0,\tau_{1}]} \bigl\Vert \mathscr{T}_{\mu }(\tau_{1}-s)- \mathscr{T}_{\mu}(\tau_{2}-s) \bigr\Vert _{L_{b}(H,H)} \int_{0}^{\tau _{1}}\bigl( \varPhi_{r}(s)+ \Vert B \Vert _{L_{b}(U,H)} \bigl\Vert u(s) \bigr\Vert _{U} \bigr)\,ds \\ \leq&\tau_{2}^{1-\gamma}\sup_{s\in[0,\tau_{1}]} \bigl\Vert \mathscr{T}_{\mu}(\tau _{1}-s)-\mathscr{T}_{\mu}( \tau_{2}-s) \bigr\Vert _{L_{b}(H,H)} \\ &{}\times\biggl(\frac{1}{\sqrt {\tau_{1}}} \Vert \varPhi_{r} \Vert _{L^{2}_{R^{+}}}+\frac{1}{\sqrt{\tau_{1}}} \Vert B \Vert _{L_{b}(U,H)} \Vert u \Vert _{L^{2}_{U}} \biggr), \end{aligned}$$

which tends to zero as \(\delta\rightarrow0\) via Lemma 2.8(ii) and Remark 2.9 (so \(\sup_{s\in[0,\tau_{1}]}\|\mathscr{T}_{\mu }(\tau_{1}-s)-\mathscr{T}_{\mu}(\tau_{2}-s)\|_{L_{b}(H,H)}\to0\) as \(\delta\rightarrow0\)).


$$\begin{aligned} Q_{4} \leq&\tau_{2}^{1-\gamma} \int_{\tau_{1}}^{\tau_{2}}\frac{M(\tau _{2}-s)^{\mu-1}}{\varGamma(\mu)}\bigl( \varPhi_{r}(s)+ \Vert B \Vert _{L_{b}(U,H)} \bigl\Vert u(s) \bigr\Vert _{U}\bigr)\,ds \\ \leq&\tau_{2}^{1-\gamma} \biggl(\frac{M\delta^{\mu-\frac{1}{2}}}{\sqrt {2\mu-1}\varGamma(\mu)}\bigl( \Vert \varPhi_{r} \Vert _{L^{2}_{R^{+}}} + \Vert B \Vert _{L_{b}(U,H)} \Vert u \Vert _{L^{2}_{U}}\bigr) \biggr), \end{aligned}$$

which tends to zero as \(\delta\rightarrow0\).

Thus \(\{P(z):z\in B_{r}(J)\}\) is an equicontinuous family of functions.

Step 4: \(\forall t\in J\), the set \(\varPi(t)=\{P z(t):z\in B_{r}(J)\}\) is relatively compact in H.

For each \(\epsilon\in(0,t)\), \(t\in J\), \(z\in B_{r}\) (and \(x=\cdot^{\gamma -1}z\)) and any \(\delta>0\), we let

$$\begin{aligned} P z(t)=J_{1}(t)+J_{2}(t),\qquad P z^{\epsilon,\delta}(t)= \tilde{J_{1}}(t)+\tilde{J_{2}}(t), \end{aligned}$$


$$\begin{gathered} J_{1}(t):=t^{1-\gamma} \frac{x_{0}}{\varGamma(\nu(1-\mu))} \int^{t}_{0} \int _{0}^{\infty}(t-s)^{\nu(1-\mu)-1} s^{\mu-1} \mu\theta M_{\mu}(\theta )T\bigl(s^{\mu}\theta\bigr)\,d\theta \,ds, \\ J_{2}(t):=t^{1-\gamma} \int_{0}^{t} \int_{0}^{\infty}(t-s)^{\mu-1} \mu\theta M_{\mu}(\theta)T\bigl((t-s)^{\mu}\theta\bigr) \bigl[f\bigl(s, x(s)\bigr)+Bu(s) \bigr] \,d\theta \,ds, \\ \tilde{J_{1}}(t):=t^{1-\gamma} \frac{x_{0}}{\varGamma(\nu(1-\mu))} \int ^{t-\epsilon}_{0} \int_{\delta}^{\infty}(t-s)^{\nu(1-\mu)-1} s^{\mu-1} \mu\theta M_{\mu}(\theta)T\bigl(s^{\mu}\theta\bigr)\,d\theta \,ds, \\ \tilde{J_{2}}(t):=t^{1-\gamma} \int_{0}^{t-\epsilon} \int_{\delta}^{\infty }(t-s)^{\mu-1} \mu\theta M_{\mu}(\theta)T\bigl((t-s)^{\mu}\theta\bigr) \bigl[f\bigl(s, x(s)\bigr)+Bu(s) \bigr] \,d\theta \,ds.\end{gathered} $$

From Lemma 2.8(ii) we see that the set

$$\varPi_{\epsilon,\delta}(t)=\bigl\{ P z^{\epsilon,\delta}(t):z\in B_{r}(J) \bigr\} $$

is relatively compact in H for each \(\epsilon\in(0,t)\) and \(\delta>0\).

Moreover, we have

$$ \begin{aligned}[b] & \bigl\Vert J_{1}(t)-\tilde{J_{1}}(t) \bigr\Vert _{H} \\ &\quad=t^{1-\gamma}\frac{x_{0}}{\varGamma(\nu(1-\mu))} \biggl\Vert \int_{0}^{t} \int _{0}^{\infty} (t-s)^{\nu(1-\mu)-1} s^{\mu-1} \mu\theta M_{\mu}(\theta)T\bigl(s^{\mu}\theta \bigr)\,d\theta \,ds \\ & \qquad- \int_{0}^{t} \int_{\delta}^{\infty} (t-s)^{\nu(1-\mu)-1} s^{\mu-1} \mu\theta M_{\mu}(\theta)T\bigl(s^{\mu}\theta \bigr)\,d\theta \,ds \\ &\qquad+ \int_{t-\epsilon}^{t} \int_{\delta}^{\infty} (t-s)^{\nu(1-\mu)-1} s^{\mu-1} \mu\theta M_{\mu}(\theta)T\bigl(s^{\mu}\theta \bigr)\,d\theta \,ds \biggr\Vert _{H} \\ &\quad\leq\sup_{t\in[0,b]}t^{1-\gamma} \frac{x_{0}M\mu}{\varGamma(\nu(1-\mu ))} \biggl\{ \biggl\Vert \int_{0}^{t}(t-s)^{\nu(1-\mu)-1} s^{\mu-1} \,ds \int _{0}^{\delta}\theta M_{\mu}(\theta)\,d\theta \biggr\Vert _{H} \\ &\qquad+ \biggl\Vert \int_{t-\epsilon}^{t}(t-s)^{\nu(1-\mu)-1} s^{\mu-1} \,ds \biggl(\frac{1}{\varGamma(1+\mu)}- \int_{0}^{\delta}\theta M_{\mu}(\theta )\,d\theta \biggr) \biggr\Vert _{H} \biggr\} \\ &\quad\leq\frac{x_{0}M\varGamma(\mu+1)}{\varGamma(\nu(1-\mu)+\mu)} \int _{0}^{\delta}\theta M_{\mu}(\theta)\,d\theta \\ &\qquad+\frac{x_{0}M\mu b^{1-\gamma}}{\varGamma(\nu(1-\mu)+\mu)} {\biggl[}\frac {\epsilon^{\nu(1-\mu)-\frac{1}{2}}}{\sqrt{2\nu(1-\mu)-1}}-\frac{\epsilon ^{\mu-\frac{1}{2}}}{\sqrt{2\mu-1}} {\biggr]} { \biggl[}\frac{1}{\varGamma (1+\mu)}- \int_{0}^{\delta}\theta M_{\mu}(\theta)\,d\theta {\biggr]}\end{aligned} $$


$$ \begin{aligned}[b] & \bigl\Vert J_{2}(t)-\tilde{J_{2}}(t) \bigr\Vert _{H} \\ &\quad=t^{1-\gamma} \biggl\Vert \int_{0}^{t} \int_{0}^{\infty} (t-s)^{\mu-1} \mu\theta M_{\mu}(\theta)T\bigl((t-s)^{\mu}\theta\bigr)\bigl[f\bigl(s, x(s)\bigr)+Bu(s)\bigr]\,d\theta \,ds \\ & \qquad- \int_{0}^{t} \int_{\delta}^{\infty} (t-s)^{\mu-1} \mu\theta M_{\mu}(\theta)T\bigl((t-s)^{\mu}\theta\bigr)\bigl[f \bigl(s,x(s)\bigr)+Bu(s)\bigr]\,d\theta \,ds \\ &\qquad+ \int_{t-\epsilon}^{t} \int_{\delta}^{\infty} (t-s)^{\mu-1} \mu\theta M_{\mu}(\theta)T\bigl((t-s)^{\mu}\theta\bigr)\bigl[f\bigl(s, x(s)\bigr)+Bu(s)\bigr]\,d\theta \,ds \biggr\Vert _{H} \\ &\quad\leq \sup_{t\in[0,b]}t^{1-\gamma} M \biggl\{ \biggl\Vert \int _{0}^{t}(t-s)^{\mu-1}\mu\bigl[f\bigl(s, x(s)\bigr)+Bu(s)\bigr]\,ds \int_{0}^{\delta}\theta M_{\mu}(\theta)\,d\theta \biggr\Vert _{H} \\ &\qquad+ \biggl\Vert \int_{t-\epsilon}^{t}(t-s)^{\mu-1}\mu\bigl[f\bigl(s, x(s)\bigr)+Bu(s)\bigr]\,ds \biggl(\frac{1}{\varGamma(1+\mu)}- \int_{0}^{\delta}\theta M_{\mu}(\theta )\,d\theta \biggr) \biggr\Vert _{H} \biggr\} \\ &\quad\leq\frac{\mu Mb^{2-2\gamma}}{\sqrt{2\mu-1}} {\bigl(} \Vert \varPhi_{r} \Vert _{L^{2}_{R^{+}}}+ \Vert B \Vert \Vert u \Vert _{L^{2}_{U}} {\bigr)} { \biggl[}b^{\mu-\frac{1}{2}} \int_{0}^{\delta}\theta M_{\mu}(\theta )\,d\theta +\frac{\epsilon^{\mu-\frac{1}{2}}}{\varGamma(1+\mu)} {\biggr].}\end{aligned} $$

Since \(0\leq\int_{0}^{\delta}\theta M_{\mu}(\theta)\,d\theta\leq\int _{0}^{\infty}\theta M_{\mu}(\theta)\,d\theta=\frac{1}{\varGamma(1+\mu)}\), (3.5) and (3.6) tend to zero when \(\epsilon\rightarrow0\) and \(\delta\rightarrow0\). Therefore the set \(\{\varPi(t),t>0\}\) is relatively compact in H.

Schauder’s fixed point theorem guarantees that P has a fixed point \(z^{*}\in B_{r}(J)\). Let \(x^{*}=\cdot^{\gamma-1}z^{*}\in B^{(1-\gamma )}_{r}(J^{\prime})\) and then \(\digamma(x)\) has a fixed point \(x^{*}\in B^{(1-\gamma)}_{r}(J^{\prime})\) (i.e. a mild solution of system (1.1)). The proof of Theorem 3.6 is complete. □

Theorem 3.7

Assume that condition \(H(f)\)holds. Then there exists a fixed point of \(\digamma_{\epsilon}\)in Y.


The proof is similar to Theorem 3.6, so we omit it here. □

In fact, for any \(\epsilon>0\), there exist \(x_{\epsilon}\in Y\) and a map \(\boldsymbol{\varPsi}(x):Y\rightarrow H\) which related to a unique minimum \(\boldsymbol{\varPsi}_{\epsilon}\) of the functional \(\mathcal {J_{\epsilon}}\) such that

$$x_{\epsilon}(t)= \mathscr{L}_{\nu,\mu}(t)x_{0}+ \int_{0}^{t}\mathscr{T}_{\mu}(t-s) \bigl[f \bigl(s,x_{\epsilon}(s)\bigr)+Bu_{\epsilon}(s,x) \bigr]\,ds, $$


$$u_{\epsilon}(s,x)=B^{*}\mathscr{T}_{\mu}^{*}(b-s)R_{\epsilon }^{b} \boldsymbol{\varPsi}_{\epsilon}(x). $$

4 Finite approximate controllability for the semilinear case

In this section, we study the finite approximate controllability of system (1.1).

Theorem 4.1

Assume that assumptions \(H(f)\)and \(H(B)\)hold. Then system (1.1) is finite approximately controllable on \(J^{\prime}\).


From (3.1) we know that the functional \(\mathcal{J_{\epsilon }}(\varPsi;x_{\epsilon})\) is strictly convex, so we assume \(\widetilde {\varPsi_{\epsilon}}\) be the unique critical point which minimizes \(\mathcal{J_{\epsilon}}(\varPsi;x_{\epsilon})\), that is,

$$\mathcal{J_{\epsilon}}(\widetilde{\varPsi_{\epsilon}};x_{\epsilon })= \min_{\varPsi\in H}\mathcal{J_{\epsilon}}(\varPsi;x_{\epsilon}). $$

Because \(\mathcal{J_{\epsilon}}(\varPsi;x_{\epsilon})\) is Gateaux differentiable at \(\widetilde{\varPsi_{\epsilon}}\), then, for any \(\varPsi_{0}\in H\) and \(\theta>0\), we get

$$\begin{aligned}& \mathcal{J_{\epsilon}}(\widetilde{\varPsi_{\epsilon}}+\theta\varPsi _{0};x_{\epsilon})-\mathcal{J_{\epsilon}}(\widetilde{ \varPsi_{\epsilon }};x_{\epsilon}) \\& \quad=\epsilon \bigl\Vert (I-\varPi_{\mathscr{E}})R_{\epsilon}^{b}(\widetilde{\varPsi_{\epsilon}}+\theta\varPsi_{0}) \bigr\Vert _{H}+\frac {1}{2} \int_{0}^{b} \bigl\Vert B^{*} \mathscr{T}_{\mu}^{*}(b-t)R_{\epsilon}^{b}( \widetilde{\varPsi_{\epsilon}}+\theta\varPsi_{0}) \bigr\Vert _{H}^{2}\,dt\\& \qquad{}-\bigl\langle \mathscr{H}(x_{\epsilon}),R_{\epsilon}^{b}( \widetilde {\varPsi_{\epsilon}}+\theta\varPsi_{0}) \bigr\rangle \\& \qquad{}-\epsilon \bigl\Vert (I-\varPi _{\mathscr{E}})R_{\epsilon}^{b} \widetilde{\varPsi_{\epsilon}} \bigr\Vert _{H}-\frac {1}{2} \int_{0}^{b} \bigl\Vert B^{*} \mathscr{T}_{\mu}^{*}(b-t)R_{\epsilon}^{b} \widetilde{\varPsi_{\epsilon}} \bigr\Vert _{H}^{2}\,dt+ \bigl\langle \mathscr {H}(x_{\epsilon}),R_{\epsilon}^{b} \widetilde{\varPsi_{\epsilon}}\bigr\rangle \\& \quad= \epsilon\theta \bigl\Vert (I-\varPi_{\mathscr{E}})R_{\epsilon}^{b} \varPsi_{0} \bigr\Vert _{H}+\theta \int_{0}^{b}\bigl\langle B^{*} \mathscr{T}_{\mu}^{*}(b-t)R_{\epsilon}^{b} \widetilde{\varPsi_{\epsilon}},B^{*}\mathscr{T}_{\mu }^{*}(b-t)R_{\epsilon}^{b} \varPsi_{0}\bigr\rangle \,dt \\& \qquad{}+\frac{\theta^{2}}{2} \int_{0}^{b} \bigl\Vert B^{*} \mathscr{T}_{\mu }^{*}(b-t)R_{\epsilon}^{b} \varPsi_{0} \bigr\Vert _{H}^{2}\,dt-\theta\bigl\langle \mathscr {H}(x_{\epsilon}),\varPsi_{0}\bigr\rangle , \end{aligned}$$

such that

$$\begin{aligned} 0&={\lim_{\theta\to0^{+}}}\frac{\mathcal{J_{\epsilon}}(\widetilde {\varPsi_{\epsilon}}+\theta\varPsi_{0};x_{\epsilon})-\mathcal {J_{\epsilon}}(\widetilde{\varPsi_{\epsilon}};x_{\epsilon})}{\theta} \\ &={\lim_{\theta\to0^{+}}} \biggl(\epsilon \bigl\Vert (I- \varPi_{\mathscr {E}})R_{\epsilon}^{b} \varPsi_{0} \bigr\Vert _{H}+ \int_{0}^{b}\bigl\langle B^{*}\mathscr {T}_{\mu}^{*}(b-t)R_{\epsilon}^{b}\widetilde{ \varPsi_{\epsilon }},B^{*}\mathscr{T}_{\mu}^{*}(b-t)R_{\epsilon}^{b} \varPsi_{0}\bigr\rangle \,dt \\ &\quad+ \frac{\theta}{2} \int_{0}^{b} \bigl\Vert B^{*} \mathscr{T}_{\mu }^{*}(b-t)R_{\epsilon}^{b} \varPsi_{0} \bigr\Vert _{H}^{2}\,dt-\bigl\langle \mathscr {H}(x_{\epsilon}) ,R_{\epsilon}^{b} \varPsi_{0}\bigr\rangle \biggr) \\ &=\epsilon \bigl\Vert (I-\varPi_{\mathscr{E}})R_{\epsilon}^{b} \varPsi_{0} \bigr\Vert _{H}+ \int _{0}^{b}\bigl\langle B^{*} \mathscr{T}_{\mu}^{*}(b-t)R_{\epsilon}^{b} \widetilde {\varPsi_{\epsilon}},B^{*}\mathscr{T}_{\mu}^{*}(b-t)R_{\epsilon}^{b}\varPsi_{0}\bigr\rangle \,dt \\ &\quad-\bigl\langle \mathscr{H}(x_{\epsilon}) ,R_{\epsilon}^{b} \varPsi_{0}\bigr\rangle \,dt,\end{aligned} $$

as is well known

$$\begin{gathered} \int_{0}^{b}\bigl\langle B^{*} \mathscr{T}_{\mu}^{*}(b-t)R_{\epsilon}^{b} \widetilde{\varPsi_{\epsilon}},B^{*}\mathscr{T}_{\mu }^{*}(b-t)R_{\epsilon}^{b} \varPsi_{0}\bigr\rangle \,dt \\ \quad= \int_{0}^{b}\bigl\langle \mathscr{T}_{\mu}(b-t)BB^{*} \mathscr{T}_{\mu}^{*}(b-t)R_{\epsilon}^{b} \widetilde{\varPsi_{\epsilon}},R_{\epsilon}^{b} \varPsi_{0}\bigr\rangle \,dt \\ \quad= \int_{0}^{b}\bigl\langle \mathscr{T}_{\mu}(b-t)Bu_{\epsilon }(s,x),R_{\epsilon}^{b} \varPsi_{0}\bigr\rangle \,dt,\end{gathered} $$


$$\begin{gathered} \bigl\langle \mathscr{H}(x_{\epsilon}) ,R_{\epsilon}^{b} \varPsi_{0}\bigr\rangle \\ \quad=\epsilon \bigl\Vert (I-\varPi_{\mathscr{E}})R_{\epsilon}^{b} \varPsi_{0} \bigr\Vert _{H}+ \int _{0}^{b}\bigl\langle \mathscr{T}_{\mu}(b-t)BB^{*} \mathscr{T}_{\mu }^{*}(b-t)R_{\epsilon}^{b} \widetilde{\varPsi_{\epsilon}},R_{\epsilon}^{b} \varPsi_{0}\bigr\rangle \,dt \\ \quad=\epsilon \bigl\Vert (I-\varPi_{\mathscr{E}})R_{\epsilon}^{b} \varPsi_{0} \bigr\Vert _{H}+ \int_{0}^{b}\bigl\langle \mathscr{T}_{\mu }(b-t)Bu_{\epsilon}(s,x),R_{\epsilon}^{b} \varPsi_{0}\bigr\rangle \,dt.\end{gathered} $$

From the definition of \(\mathscr{H}(x_{\epsilon})\), \(x_{\epsilon}(b)\), one can get

$$\mathscr{H}(x_{\epsilon})=x_{b}-x_{\epsilon}(b)+ \int_{0}^{t}\mathscr{T}_{\mu }(t-s)Bu_{\epsilon}(s,x) \,ds, $$


$$\bigl\vert \bigl\langle x_{b}-x_{\epsilon}(b), \varPsi_{0} \bigr\rangle \bigr\vert = \epsilon \bigl\Vert (I-\varPi _{\mathscr{E}})\varPsi_{0} \bigr\Vert _{H}\leq\epsilon \Vert \varPsi_{0} \Vert _{H}, $$

which is equivalent to

$$\bigl\Vert x_{b}-x_{\epsilon}(b) \bigr\Vert _{H} \leq\epsilon. $$

On the other hand, if \(\theta<0\), we can get the same argument.

Thus given \(\varPsi_{0} \in H\), we conclude that system (1.1) is approximately controllable on \(J^{\prime}\), and if \(\varPsi_{0} \in\mathscr {E}\), system (1.1) is finite approximately controllable on \(J^{\prime}\), that is, \(\varPi_{\mathscr{E}}x_{\varepsilon}(b)=\varPi_{\mathscr {E}}x_{b}\). □

5 An example

As an application of our result, consider the Hilfer fractional partial equation:

$$\begin{aligned} \left \{ \textstyle\begin{array}{llll} D^{\nu,\mu}_{0^{+}}x(t,y)=x_{yy}(t,y)+Bu(t,y)+\int_{0}^{t}e^{-s}\frac { \vert x(s,y) \vert }{1+ \vert x(s,y) \vert }\,ds,\quad 0< t\leq1, 0\leq y\leq\pi,\\ x(t,0)=x(t,\pi)=0,\quad 0< t\leq1,\\ I^{(1-\nu)(1-\mu)}_{0^{+}}x(0,y)=x_{0}(y),\quad 0\leq y\leq\pi, \end{array}\displaystyle \right . \end{aligned}$$

where \(\nu=1/2\), \(\mu=3/4 \), and \(x(t,y)\) represents the temperature function at the point \(y\in[0,\pi]\) and time \(t\in(0,1]\). Now, set \(H=L^{2}[0,\pi]\) and \(e_{n}(y)=\sqrt{2/\pi}\sin(ny)\), \(n=1,2,\dots \). Then \(\{e_{n}(y)\}\) is an orthonormal basis on H. Define \(A:D(A)\subset H\rightarrow H\) by \(Ax=x_{yy}\) with domain

$$\bigl\{ x\in H: x, x'\textrm{ are absolutely continuous}, x''\in H, x(0)=x(\pi )=0\bigr\} . $$


$$Ax=\sum_{n=1}^{\infty}\bigl(-n^{2} \bigr)\langle x,e_{n}\rangle e_{n},\quad x\in D(A), $$

one can see that A generates a compact semigroup \(T(t)\) (\(t>0\)) on H and

$$T(t)x=\sum_{n=1}^{\infty}e^{-n^{2}t} \langle x,e_{n}\rangle e_{n},\quad x\in H. $$

Hence \(T(t)\) is compact and \(\|T(t)\|\leq1\).

The infinite-dimensional Hilbert space U is

$$U:=\Biggl\{ u:u=\sum_{n=2}^{\infty}u_{n}e_{n}, \sum_{n=2}^{\infty}u_{n}^{2}< \infty\Biggr\} , $$

with the norm \(\|u\|_{U}=(\sum_{n=2}^{\infty}u_{n}^{2})^{1/2}\). We define a mapping \(B\in\mathcal{L}(U,H)\) by

$$Bu=4u_{2}e_{1}+3u_{2}e_{2}+\sum _{n=3}^{\infty}u_{n}e_{n} \quad\textrm{for } u=\sum_{n=2}^{\infty}u_{n}e_{n}\in U, $$

and for \(v=\sum_{n=1}^{\infty}v_{n}e_{n}\in H\), the inner product \(\langle Bu,v\rangle=\langle u,B^{*}v\rangle\), and thus

$$B^{*}v=(4v_{1}+3v_{2})e_{2}+\sum _{n=3}^{\infty}v_{n}e_{n} $$


$$B^{*}T^{*}(t)x=\bigl(4x_{1}e^{-t}+3x_{2}e^{-4t} \bigr)e_{2}+\sum_{n=3}^{\infty}e^{-n^{2}t}x_{n}e_{n}. $$

Assume \(\|B^{*}T^{*}(t)x\|=0\) for some \(t\in J^{\prime}\), and it follows that

$$\bigl\Vert 4x_{1}e^{-t}+3x_{2}e^{-4t} \bigr\Vert ^{2}+\sum_{n=3}^{\infty}\bigl\Vert e^{-n^{2}t}x_{n} \bigr\Vert ^{2}=0, $$

which implies that \(x=0\), and thus the linear part of system (5.1) is approximately controllable on \(J^{\prime}\) (see Theorem 4.1.7 of [23]). Now

$$f\bigl(t,x(t,y)\bigr)= \int_{0}^{t}e^{-s}\frac{ \vert x(s,y) \vert }{1+ \vert x(s,y) \vert }\,ds \leq \int _{0}^{t}e^{-s}\,ds=1-e^{-t}, $$

so the conditions of \(H(f)\) hold. Thus system (5.1) is finite approximately controllable on \(J^{\prime}\).


  1. Zhou, Y., Wang, J., Zhang, L.: Basic Theory of Fractional Differential Equations, 2nd edn. World Scientific, Singapore (2016)

    Book  Google Scholar 

  2. Diethelm, K.: The Analysis of Fractional Differential Equations. Springer, Berlin (2010)

    Book  Google Scholar 

  3. Hernández, E., O’Regan, D., Balachandran, K.: On recent developments in the theory of abstract differential equations with fractional derivatives. Nonlinear Anal. TMA 73, 3462–3471 (2010)

    Article  MathSciNet  Google Scholar 

  4. Wang, J., Fečkan, M., Zhou, Y.: On the new concept of solutions and existence results for impulsive fractional evolution equations. Dyn. Partial Differ. Equ. 8, 345–361 (2011)

    Article  MathSciNet  Google Scholar 

  5. Mahmudov, N.I.: Approximate controllability of semilinear deterministic and stochastic evolution equations in abstract spaces. SIAM J. Control Optim. 42, 1604–1622 (2003)

    Article  MathSciNet  Google Scholar 

  6. Mahmudov, N.I.: Approximate controllability of evolution systems with nonlocal conditions. Nonlinear Anal. 68, 536–546 (2008)

    Article  MathSciNet  Google Scholar 

  7. Mahmudov, N.I., Zorlu, S.: On the approximate controllability of fractional evolution equations with compact analytic semigroup. J. Comput. Appl. Math. 259, 194–204 (2014)

    Article  MathSciNet  Google Scholar 

  8. Hilfer, R.: Applications of Fractional Calculus in Physics. World Scientific, Singapore (2000)

    Book  Google Scholar 

  9. Yang, M., Wang, Q.: Approximate controllability of Hilfer fractional differential inclusions with nonlocal conditions. Math. Methods Appl. Sci. 40, 1126–1138 (2007)

    Article  MathSciNet  Google Scholar 

  10. Furati, K.M., Kassim, M.D., Tatar, N.: Existence and uniqueness for a problem involving Hilfer fractional derivative. Comput. Math. Appl. 64, 1616–1626 (2012)

    Article  MathSciNet  Google Scholar 

  11. Furati, K.M., Kassim, M.D., Tatar, N.: Non-existence of global solutions for a differential equation involving Hilfer fractional derivative. Electron. J. Differ. Equ. 235, 1 (2013)

    MathSciNet  MATH  Google Scholar 

  12. Gu, H., Trujillo, J.J.: Existence of mild solution for evolution equation with Hilfer fractional derivative. Appl. Math. Comput. 257, 344–354 (2015)

    MathSciNet  MATH  Google Scholar 

  13. Mahmudov, N.I., McKibben, M.A.: On the approximate controllability of fractional evolution equations with generalized Riemann–Liouville fractional derivative. J. Funct. Spaces 2015, Article ID 263823 (2015)

    MathSciNet  MATH  Google Scholar 

  14. Wang, J., Liu, X., O’Regan, D.: On the approximate controllability for Hilfer fractional evolution hemivariational inequalities. Numer. Funct. Anal. Optim. 40, 743–762 (2019)

    Article  MathSciNet  Google Scholar 

  15. Li, X., Yong, J.: Optimal Control Theory for Infinite Dimensional Systems. Birkhäuser, Boston (1994)

    MATH  Google Scholar 

  16. Zuazua, E.: Finite dimensional null controllability for the semilinear heat equation. J. Math. Pures Appl. 76, 237–264 (1997)

    Article  MathSciNet  Google Scholar 

  17. de Menezes, S.B.: Finite-dimensional approximate controllability for a nonlocal parabolic problem. Appl. Math. Sci. 2, 1307–1326 (2008)

    MathSciNet  MATH  Google Scholar 

  18. Mahmudov, N.I.: Finite-approximate controllability of evolution equations. Appl. Comput. Math. 16, 159–167 (2017)

    MathSciNet  MATH  Google Scholar 

  19. Mahmudov, N.I.: Variational approach to finite-approximate controllability of Sobolev-type fractional systems. J. Optim. Theory Appl. (in press)

  20. Miller, K.S., Ross, B.: An Introduction to the Fractional Calculus and Differential Equations. Wiley, New York (1993)

    MATH  Google Scholar 

  21. Podlubny, I.: Fractional Differential Equations. Academic Press, San Diego (1999)

    MATH  Google Scholar 

  22. Kilbas, A.A., Srivastava, H.M., Trujillo, J.J.: Theory and Applications of Fractional Differential Equations (2006)

    MATH  Google Scholar 

  23. Curtain, R.F., Zwart, H.: An Introduction to Infinite Dimensional Linear Systems Theory. Springer, New York (1995)

    Book  Google Scholar 

Download references


Not applicable.

Availability of data and materials

All data generated or analyzed during this study are included in this published article.


The work is supported by NSF of China (No. 11661084), Guizhou Province Science and technology fund [2016]1160, [2017]1201, Guizhou Province Innovative talents fund [2016]046, Zunyi Science and technology talents fund [2016]15, Sci-Tec Innovative Talents of Guizhou Province (No. [2015]502), Qian Ke He Ping Tai Ren Cai [2018]5784-08.

Author information

Authors and Affiliations



XL researched and analyzed the model, and was a major contributor in writing the manuscript. GX analyzed the data. YL reviewed the format of this manuscript. All authors read and approved the final manuscript.

Corresponding author

Correspondence to Xianghu Liu.

Ethics declarations

Competing interests

The authors declare that they have 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

Reprints and Permissions

About this article

Check for updates. Verify currency and authenticity via CrossMark

Cite this article

Liu, X., Li, Y. & Xu, G. On the finite approximate controllability for Hilfer fractional evolution systems. Adv Differ Equ 2020, 22 (2020).

Download citation

  • Received:

  • Accepted:

  • Published:

  • DOI:


  • 93C10
  • 93C40


  • Finite approximate controllability
  • Hilfer fractional derivative
  • Mild solutions
  • Variational approach
  • Schauder’s fixed point theorem