It is well known that the gradient-projection algorithm (GPA) is very useful in solving constrained convex minimization problems. In this paper, we combine a general iterative method with the gradient-projection algorithm to propose a hybrid gradient-projection algorithm and prove that the sequence generated by the hybrid gradient-projection algorithm converges in norm to a minimizer of constrained convex minimization problems which solves a variational inequality.

1. Introduction

Let H be a real Hilbert space and C a nonempty closed and convex subset of H. Consider the following constrained convex minimization problem:
(1.1)minimizex∈Cf(x),
where f:C→ℝ is a real-valued convex and continuously Fréchet differentiable function. The gradient ∇f satisfies the following Lipschitz condition:
(1.2)‖∇f(x)-∇f(y)‖≤L‖x-y‖,∀x,y∈C,
where L>0. Assume that the minimization problem (1.1) is consistent, and let S denote its solution set.

It is well known that the gradient-projection algorithm is very useful in dealing with constrained convex minimization problems and has extensively been studied ([1–5] and the references therein). It has recently been applied to solve split feasibility problems [6–10]. Levitin and Polyak [1] consider the following gradient-projection algorithm:
(1.3)xn+1:=ProjC(xn-λn∇f(xn)),n≥0.
Let {λn}n=0∞ satisfy
(1.4)0<liminfn→∞λn≤limsupn→∞λn<2L.
It is proved that the sequence {xn} generated by (1.3) converges weakly to a minimizer of (1.1).

Xu proved that under certain appropriate conditions on {αn} and {λn} the sequence {xn} defined by the following relaxed gradient-projection algorithm:
(1.5)xn+1=(1-αn)xn+αnProjC(xn-λn∇f(xn)),n≥0,
converges weakly to a minimizer of (1.1) [11].

Since the Lipschitz continuity of the gradient of f implies that it is indeed inverse strongly monotone (ism) [12, 13], its complement can be an averaged mapping. Recall that a mapping T is nonexpansive if and only if it is Lipschitz with Lipschitz constant not more than one, that a mapping is an averaged mapping if and only if it can be expressed as a proper convex combination of the identity mapping and a nonexpansive mapping, and that a mapping T is said to be ν-inverse strongly monotone if and only if 〈x-y,Tx-Ty〉≥ν∥Tx-Ty∥2forallx,y∈H, where the number ν>0. Recall also that the composite of finitely many averaged mappings is averaged. That is, if each of the mappings {Ti}i=1N is averaged, then so is the composite T1⋯TN [14]. In particular, an averaged mapping is a nonexpansive mapping [15]. As a result, the GPA can be rewritten as the composite of a projection and an averaged mapping which is again an averaged mapping.

Generally speaking, in infinite-dimensional Hilbert spaces, GPA has only weak convergence. Xu [11] provided a modification of GPA so that strong convergence is guaranteed. He considered the following hybrid gradient-projection algorithm:
(1.6)xn+1=θnh(xn)+(1-θn)ProjC(xn-λn∇f(xn)).

It is proved that if the sequences {θn} and {λn} satisfy appropriate conditions, the sequence {xn} generated by (1.6) converges in norm to a minimizer of (1.1) which solves the variational inequality
(1.7)x*∈S,〈(I-h)x*,x-x*〉≥0,x∈S.

On the other hand, Ming Tian [16] introduced the following general iterative algorithm for solving the variational inequality
(1.8)xn+1=αnγf(xn)+(I-μαnF)Txn,n≥0,
where F is a κ-Lipschitzian and η-strongly monotone operator with κ>0, η>0 and f is a contraction with coefficient 0<α<1. Then, he proved that if {αn} satisfying appropriate conditions, the {xn} generated by (1.8) converges strongly to the unique solution of variational inequality
(1.9)〈(μF-γf)x~,x~-z〉≤0,z∈Fix(T).

In this paper, motivated and inspired by the research work in this direction, we will combine the iterative method (1.8) with the gradient-projection algorithm (1.3) and consider the following hybrid gradient-projection algorithm:
(1.10)xn+1=θnγh(xn)+(I-μθnF)ProjC(xn-λn∇f(xn)),n≥0.

We will prove that if the sequence {θn} of parameters and the sequence {λn} of parameters satisfy appropriate conditions, then the sequence {xn} generated by (1.10) converges in norm to a minimizer of (1.1) which solves the variational inequality (VI)(1.11)x*∈S,〈(μF-γh)x*,x-x*〉≥0,∀x∈S,
where S is the solution set of the minimization problem (1.1).

2. Preliminaries

This section collects some lemmas which will be used in the proofs for the main results in the next section. Some of them are known; others are not hard to derive.

Throughout this paper, we write xn⇀x to indicate that the sequence {xn} converges weakly to x, xn→x implies that {xn} converges strongly to x. ωw(xn):={x:∃xnj⇀x} is the weak ω-limit set of the sequence {xn}n=1∞.

Lemma 2.1 (see [<xref ref-type="bibr" rid="B19">17</xref>]).

Assume that {an}n=0∞ is a sequence of nonnegative real numbers such that
(2.1)an+1≤(1-γn)an+γnδn+βn,n≥0,
where {γn}n=0∞ and {βn}n=0∞ are sequences in [0,1] and {δn}n=0∞ is a sequence in ℝ such that

∑n=0∞γn=∞;

either limsupn→∞δn≤0 or ∑n=0∞γn|δn|<∞;

∑n=0∞βn<∞.

Then limn→∞an=0.
Lemma 2.2 (see [<xref ref-type="bibr" rid="B20">18</xref>]).

Let C be a closed and convex subset of a Hilbert space H, and let T:C→C be a nonexpansive mapping with FixT≠∅. If {xn}n=1∞ is a sequence in C weakly converging to x and if {(I-T)xn}n=1∞ converges strongly to y, then (I-T)x=y.

Lemma 2.3.

Let H be a Hilbert space, and let C be a nonempty closed and convex subset of H. h:C→C a contraction with coefficient 0<ρ<1, and F:C→C a κ-Lipschitzian continuous operator and η-strongly monotone operator with κ,η>0. Then, for 0<γ<μη/ρ,
(2.2)〈x-y,(μF-γh)x-(μF-γh)y〉≥(μη-γρ)‖x-y‖2,∀x,y∈C.
That is, μF-γh is strongly monotone with coefficient μη-γρ.

Lemma 2.4.

Let C be a closed subset of a real Hilbert space H, given x∈H and y∈C. Then, y=PCx if and only if there holds the inequality
(2.3)〈x-y,y-z〉≥0,∀z∈C.

3. Main Results

Let H be a real Hilbert space, and let C be a nonempty closed and convex subset of H such that C±C⊂C. Assume that the minimization problem (1.1) is consistent, and let S denote its solution set. Assume that the gradient ∇f satisfies the Lipschitz condition (1.2). Since S is a closed convex subset, the nearest point projection from H onto S is well defined. Recall also that a contraction on C is a self-mapping of C such that ∥h(x)-h(y)∥≤ρ∥x-y∥,forallx,y∈C, where ρ∈[0,1) is a constant. Let F be a κ-Lipschitzian and η-strongly monotone operator on C with κ,η>0. Denote by Π the collection of all contractions on C, namely,
(3.1)Π={h:hisacontractiononC}.
Now given h∈Π with 0<ρ<1, s∈(0,1). Let 0<μ<2η/κ2, 0<γ<μ(η-(μκ2)/2)/ρ=τ/ρ. Assume that λs with respect to s is continuous and, in addition, λs∈[a,b]⊂(0,2/L). Consider a mapping Xs on C defined by
(3.2)Xs(x)=sγh(x)+(I-sμF)ProjC(I-λs∇f)(x),x∈C.
It is easy to see that Xs is a contraction. Setting Vs:=ProjC(I-λs∇f). It is obvious that Vs is a nonexpansive mapping. We can rewrite Xs(x) as
(3.3)Xs(x)=sγh(x)+(I-sμF)Vs(x).
First observe that for s∈(0,1), we can get
(3.4)‖(I-sμF)Vs(x)-(I-sμF)Vs(y)‖2=‖Vs(x)-Vs(y)-sμ(FVs(x)-FVs(y))‖2=‖Vs(x)-Vs(y)‖2-2sμ〈Vs(x)-Vs(y),FVs(x)-FVs(y)〉+s2μ2‖FVs(x)-FVs(y)‖2≤‖x-y‖2-2sμη‖Vs(x)-Vs(y)‖2+s2μ2κ2‖Vs(x)-Vs(y)‖2≤(1-sμ(2η-sμκ2))‖x-y‖2≤(1-sμ(2η-sμκ2)2)2‖x-y‖2≤(1-sμ(η-μκ22))2‖x-y‖2=(1-sτ)2‖x-y‖2.
Indeed, we have
(3.5)‖Xs(x)-Xs(y)‖=‖sγh(x)+(I-sμF)Vs(x)-sγh(y)-(I-sμF)Vs(y)‖≤sγ‖h(x)-h(y)‖+‖(I-sμF)Vs(x)-(I-sμF)Vs(y)‖≤sγρ‖x-y‖+(1-sτ)‖x-y‖=(1-s(τ-γρ))‖x-y‖.
Hence, Xs has a unique fixed point, denoted xs, which uniquely solves the fixed-point equation
(3.6)xs=sγh(xs)+(I-sμF)Vs(xs).
The next proposition summarizes the properties of {xs}.

Proposition 3.1.

Let xs be defined by (3.6).

{xs} isboundedfor
s∈(0,(1/τ)).

lims→0∥xs-ProjC(I-λs∇f)(xs)∥=0.

xs
definesacontinuouscurvefrom
(0,1/τ)
into
H.

Proof.

(i) Take a x¯∈S, then we have
(3.7)‖xs-x¯‖=‖sγh(xs)+(I-sμF)ProjC(I-λs∇f)(xs)-x¯‖=‖(I-sμF)ProjC(I-λs∇f)(xs)-(I-sμF)ProjC(I-λs∇f)(x¯)+s(γh(xs)-μFProjC(I-λs∇f)(x¯))‖≤(1-sτ)‖xs-x¯‖+s‖γh(xs)-μF(x¯)‖≤(1-sτ)‖xs-x¯‖+sγρ‖xs-x¯‖+s‖γh(x¯)-μF(x¯)‖.
It follows that
(3.8)‖xs-x¯‖≤‖γh(x¯)-μF(x¯)‖τ-γρ.
Hence, {xs} is bounded.

(ii) By the definition of {xs}, we have
(3.9)‖xs-ProjC(I-λs∇f)(xs)‖=‖sγh(xs)+(I-sμF)ProjC(I-λs∇f)(xs)-ProjC(I-λs∇f)(xs)‖=s‖γh(xs)-μFProjC(I-λs∇f)(xs)‖⟶0,{xs} is bounded, so are {h(xs)} and {FProjC(I-λs∇f)(xs)}.

(iii) Take s, s0∈(0,1/τ), and we have
(3.10)‖xs-xs0‖=‖sγh(xs)+(I-sμF)ProjC(I-λs∇f)(xs)-s0γh(xs0)-(I-s0μF)ProjC(I-λs0∇f)(xs0)‖≤‖(s-s0)γh(xs)+s0γ(h(xs)-h(xs0))‖+‖(I-s0μF)ProjC(I-λs0∇f)(xs)-(I-s0μF)ProjC(I-λs0∇f)(xs0)‖+‖(I-sμF)ProjC(I-λs∇f)(xs)-(I-s0μF)ProjC(I-λs0∇f)(xs)‖≤‖(s-s0)γh(xs)+s0γ(h(xs)-h(xs0))‖+‖(I-s0μF)ProjC(I-λs0∇f)(xs)-(I-s0μF)ProjC(I-λs0∇f)(xs0)‖+‖(I-sμF)ProjC(I-λs∇f)(xs)-(I-sμF)ProjC(I-λs0∇f)(xs)‖+‖(I-sμF)ProjC(I-λs0∇f)(xs)-(I-s0μF)ProjC(I-λs0∇f)(xs)‖≤|s-s0|γ‖h(xs)‖+s0γρ‖xs-xs0‖+(1-s0τ)‖xs-xs0‖+|λs-λs0|‖∇f(xs)‖+‖sμFProjC(I-λs0∇f)(xs)-s0μFProjC(I-λs0∇f)(xs)‖=|s-s0|γ‖h(xs)‖+s0γρ‖xs-xs0‖+(1-s0τ)‖xs-xs0‖+|λs-λs0|‖∇f(xs)‖+|s-s0|‖μFProjC(I-λs0∇f)(xs)‖=(γ‖h(xs)‖+μ‖FProjC(I-λs0∇f)(xs)‖)|s-s0|+s0γρ‖xs-xs0‖+(1-s0τ)‖xs-xs0‖+|λs-λs0|‖∇f(xs)‖.
Therefore,
(3.11)‖xs-xs0‖≤γ‖h(xs)‖+μ‖FProjC(I-λs0∇f)(xs)‖s0(τ-γρ)|s-s0|+‖∇f(xs)‖s0(τ-γρ)|λs-λs0|.
Therefore, xs→xs0 as s→s0. This means xs is continuous.

Our main result in the following shows that {xs} converges in norm to a minimizer of (1.1) which solves some variational inequality.

Theorem 3.2.

Assume that {xs} is defined by (3.6), then xs converges in norm as s→0 to a minimizer of (1.1) which solves the variational inequality
(3.12)〈(μF-γh)x*,x~-x*〉≥0,∀x~∈S.
Equivalently, we have Projs(I-(μF-γh))x*=x*.

Proof.

It is easy to see that the uniqueness of a solution of the variational inequality (3.12). By Lemma 2.3, μF-γh is strongly monotone, so the variational inequality (3.12) has only one solution. Let x*∈S denote the unique solution of (3.12).

To prove that xs→x*(s→0), we write, for a given x~∈S,
(3.13)xs-x~=sγh(xs)+(I-sμF)ProjC(I-λs∇f)(xs)-x~=s(γh(xs)-μFx~)+(I-sμF)ProjC(I-λs∇f)(xs)-(I-sμF)ProjC(I-λs∇f)(x~).
It follows that
(3.14)‖xs-x~‖2=s〈γh(xs)-μFx~,xs-x~〉+〈(I-sμF)ProjC(I-λs∇f)(xs)-(I-sμF)ProjC(I-λs∇f)(x~),xs-x~〉≤(1-sτ)‖xs-x~‖2+s〈γh(xs)-μFx~,xs-x~〉.
Hence,
(3.15)‖xs-x~‖2≤1τ〈γh(xs)-μFx~,xs-x~〉≤1τ{γρ‖xs-x~‖2+〈γh(x~)-μFx~,xs-x~〉}.
To derive that
(3.16)‖xs-x~‖2≤1τ-γρ〈γh(x~)-μFx~,xs-x~〉.
Since {xs} is bounded as s→0, we see that if {sn} is a sequence in (0,1) such that sn→0 and xsn⇀x¯, then by (3.16), xsn→x¯. We may further assume that λsn→λ∈[0,2/L] due to condition (1.4). Notice that ProjC(I-λ∇f) is nonexpansive. It turns out that
(3.17)‖xsn-ProjC(I-λ∇f)xsn‖≤‖xsn-ProjC(I-λsn∇f)xsn‖+‖ProjC(I-λsn∇f)xsn-ProjC(I-λ∇f)xsn‖≤‖xsn-ProjC(I-λsn∇f)xsn‖+‖(λ-λsn)∇f(xsn)‖=‖xsn-ProjC(I-λsn∇f)xsn‖+|λ-λsn|‖∇f(xsn)‖.
From the boundedness of {xs} and lims→0∥ProjC(I-λs∇f)xs-xs∥=0, we conclude that
(3.18)limn→∞‖xsn-ProjC(I-λ∇f)xsn‖=0.
Since xsn⇀x¯, by Lemma 2.2, we obtain
(3.19)x¯=ProjC(I-λ∇f)x¯.
This shows that x¯∈S.

We next prove that x¯ is a solution of the variational inequality (3.12). Since
(3.20)xs=sγh(xs)+(I-sμF)ProjC(I-λs∇f)(xs),
we can derive that
(3.21)(μF-γh)(xs)=-1s(I-ProjC(I-λs∇f)(xs)+μ(F(xs)-FProjC(I-λs∇f)(xs)).
Therefore, for x~∈S,
(3.22)〈(μF-γh)(xs),xs-x~〉=-1s〈(I-ProjC(I-λs∇f))(xs)-(I-ProjC(I-λs∇f))(x~),xs-x~〉+μ〈F(xs)-FProjC(I-λs∇f)(xs),xs-x~〉≤μ〈F(xs)-FProjC(I-λs∇f)(xs),xs-x~〉.
Since ProjC(I-λs∇f) is nonexpansive, we obtain that I-ProjC(I-λs∇f) is monotone, that is,
(3.23)〈(I-ProjC(I-λs∇f))(xs)-(I-ProjC(I-λs∇f))(x~),xs-x~〉≥0.
Taking the limit through s=sn→0 ensures that x¯ is a solution to (3.12). That is to say
(3.24)〈(μF-γh)(x¯),x¯-x~〉≤0.
Hence x¯=x* by uniqueness. Therefore, xs→x* as s→0. The variational inequality (3.12) can be written as
(3.25)〈(I-μF+γh)x*-x*,x~-x*〉≤0,∀x~∈S.
So, by Lemma 2.4, it is equivalent to the fixed-point equation
(3.26)PS(I-μF+γh)x*=x*.

Taking F=A, μ=1 in Theorem 3.2, we get the following

Corollary 3.3.

We have that {xs} converges in norm as s→0 to a minimizer of (1.1) which solves the variational inequality
(3.27)〈(A-γh)x*,x~-x*〉≥0,∀x~∈S.
Equivalently, we have Projs(I-(A-γh))x*=x*.

Taking F=I, μ=1, γ=1 in Theorem 3.2, we get the following.

Corollary 3.4.

Let zs∈H be the unique fixed point of the contraction z↦sh(z)+(1-s)ProjC(I-λs∇f)(z). Then, {zs} converges in norm as s→0 to the unique solution of the variational inequality
(3.28)〈(I-h)x*,x~-x*〉≥0,∀x~∈S.

Finally, we consider the following hybrid gradient-projection algorithm,
(3.29){x0∈Carbitrarily,xn+1=θnγh(xn)+(I-μθnF)ProjC(xn-λn∇f(xn)),∀n≥0.
Assume that the sequence {λn}n=0∞ satisfies the condition (1.4) and, in addition, that the following conditions are satisfied for {λn}n=0∞ and {θn}n=0∞⊂[0,1]:

θn→0;

∑n=0∞θn=∞;

∑n=0∞|θn+1-θn|<∞;

∑n=0∞|λn+1-λn|<∞.

Theorem 3.5.

Assume that the minimization problem (1.1) is consistent and the gradient ∇f satisfies the Lipschitz condition (1.2). Let {xn} be generated by algorithm (3.29) with the sequences {θn} and {λn} satisfying the above conditions. Then, the sequence {xn} converges in norm to x* that is obtained in Theorem 3.2.

Proof.

(1) The sequence {xn}n=0∞ is bounded. Setting
(3.30)Vn:=ProjC(I-λn∇f).
Indeed, we have, for x¯∈S,
(3.31)‖xn+1-x¯‖=‖θnγh(xn)+(I-μθnF)Vnxn-x¯‖=‖θn(γh(xn)-μF(x¯))+(I-μθnF)Vnxn-(I-μθnF)Vnx¯‖≤(1-θnτ)‖xn-x¯‖+θnργ‖xn-x¯‖+θn‖γh(x¯)-μF(x¯)‖=(1-θn(τ-γρ))‖xn-x¯‖+θn‖γh(x¯)-μF(x¯)‖≤max{‖xn-x¯‖,1τ-γρ‖γh(x¯)-μF(x¯)‖},∀n≥0.
By induction,
(3.32)‖xn-x¯‖≤max{‖x0-x¯‖,‖γh(x¯)-μF(x¯)‖τ-γρ}.
In particular, {xn}n=0∞ is bounded.

(2) We prove that ∥xn+1-xn∥→0 as n→∞. Let M be a constant such that
(3.33)M>max{supn≥0γ‖h(xn)‖,supκ,n≥0μ‖FVκxn‖,supn≥0‖∇f(xn)‖}.
We compute
(3.34)‖xn+1-xn‖=‖θnγh(xn)+(I-μθnF)Vnxn-θn-1γh(xn-1)-(I-μθn-1F)Vn-1xn-1‖=‖θnγ(h(xn)-h(xn-1))+γ(θn-θn-1)h(xn-1)+(I-μθnF)Vnxn-(I-μθnF)Vnxn-1+(I-μθnF)Vnxn-1-(I-μθn-1F)Vn-1xn-1‖=‖θnγ(h(xn)-h(xn-1))+γ(θn-θn-1)h(xn-1)+(I-μθnF)Vnxn-(I-μθnF)Vnxn-1+(I-μθnF)Vnxn-1-(I-μθnF)Vn-1xn-1+(I-μθnF)Vn-1xn-1-(I-μθn-1F)Vn-1xn-1‖≤θnγρ‖xn-xn-1‖+γ|θn-θn-1|‖h(xn-1)‖+(1-θnτ)‖xn-xn-1‖+‖Vnxn-1-Vn-1xn-1‖+μ|θn-θn-1|‖FVn-1xn-1‖≤θnγρ‖xn-xn-1‖+M|θn-θn-1|+(1-θnτ)‖xn-xn-1‖+‖Vnxn-1-Vn-1xn-1‖+M|θn-θn-1|=(1-θn(τ-γρ))‖xn-xn-1‖+2M|θn-θn-1|+‖Vnxn-1-Vn-1xn-1‖,(3.35)‖Vnxn-1-Vn-1xn-1‖=‖ProjC(I-λn∇f)xn-1-ProjC(I-λn-1∇f)xn-1‖≤‖(I-λn∇f)xn-1-(I-λn-1∇f)xn-1‖=|λn-λn-1|‖∇f(xn-1)‖≤M|λn-λn-1|.
Combining (3.34) and (3.35), we can obtain
(3.36)‖xn+1-xn‖≤(1-(τ-γρ)θn)‖xn-xn-1‖+2M(|θn-θn-1|+|λn-λn-1|).
Apply Lemma 2.1 to (3.36) to conclude that ∥xn+1-xn∥→0 as n→∞.

(3) We prove that ωw(xn)⊂S. Let x^∈ωw(xn), and assume that xnj⇀x^ for some subsequence {xnj}j=1∞ of {xn}n=0∞. We may further assume that λnj→λ∈[0,2/L] due to condition (1.4). Set V:=ProjC(I-λ∇f). Notice that V is nonexpansive and FixV=S. It turns out that
(3.37)‖xnj-Vxnj‖≤‖xnj-Vnjxnj‖+‖Vnjxnj-Vxnj‖≤‖xnj-xnj+1‖+‖xnj+1-Vnjxnj‖+‖Vnjxnj-Vxnj‖≤‖xnj-xnj+1‖+θnj‖γh(xnj)-μFVnjxnj‖+‖ProjC(I-λnj∇f)xnj-ProjC(I-λ∇f)xnj‖≤‖xnj-xnj+1‖+θnj‖γh(xnj)-μFVnjxnj‖+|λ-λnj|‖∇f(xnj)‖≤‖xnj-xnj+1‖+2M(θnj+|λ-λnj|)⟶0asj⟶∞.
So Lemma 2.2 guarantees that ωw(xn)⊂FixV=S.

(4) We prove that xn→x* as n→∞, where x* is the unique solution of the VI (3.12). First observe that there is some x^∈ωw(xn)⊂S Such that
(3.38)limsupn→∞〈(μF-γh)x*,xn-x*〉=〈(μF-γh)x*,x^-x*〉≥0.

We now compute
(3.39)‖xn+1-x*‖2=‖θnγh(xn)+(I-μθnF)ProjC(I-λn∇f)(xn)-x*‖2=‖θn(γh(xn)-μFx*)+(I-μθnF)Vn(xn)-(I-μθnF)Vnx*‖2=‖θnγ(h(xn)-h(x*))+(I-μθnF)Vn(xn)-(I-μθnF)Vnx*+θn(γh(x*)-μFx*)‖2≤‖θnγ(h(xn)-h(x*))+(I-μθnF)Vn(xn)-(I-μθnF)Vnx*‖2+2θn〈(γh-μF)x*,xn+1-x*〉=‖θnγ(h(xn)-h(x*))‖2+‖(I-μθnF)Vn(xn)-(I-μθnF)Vnx*‖2+2θnγ〈h(xn)-h(x*),(I-μθnF)Vn(xn)-(I-μθnF)Vnx*〉+2θn〈(γh-μF)x*,xn+1-x*〉≤θn2γ2ρ2‖xn-x*‖2+(1-θnτ)2‖xn-x*‖2+2θnγρ(1-θnτ)‖xn-x*‖2+2θn〈(γh-μF)x*,xn+1-x*〉=(θn2γ2ρ2+(1-θnτ)2+2θnγρ(1-θnτ))‖xn-x*‖2+2θn〈(γh-μF)x*,xn+1-x*〉≤(θnγ2ρ2+1-2θnτ+θnτ2+2θnγρ)‖xn-x*‖2+2θn〈(γh-μF)x*,xn+1-x*〉=(1-θn(2τ-γ2ρ2-τ2-2γρ))‖xn-x*‖2+2θn〈(γh-μF)x*,xn+1-x*〉.
Applying Lemma 2.1 to the inequality (3.39), together with (3.38), we get ∥xn-x*∥→0 as n→∞.

Corollary 3.6 (see [<xref ref-type="bibr" rid="B13">11</xref>]).

Let {xn} be generated by the following algorithm:
(3.40)xn+1=θnh(xn)+(1-θn)ProjC(xn-λn∇f(xn)),∀n≥0.
Assume that the sequence {λn}n=0∞ satisfies the conditions (1.4) and (iv) and that {θn}⊂[0,1] satisfies the conditions (i)–(iii). Then {xn} converges in norm to x* obtained in Corollary 3.4.

Corollary 3.7.

Let {xn} be generated by the following algorithm:
(3.41)xn+1=θnγh(xn)+(I-θnA)ProjC(xn-λn∇f(xn)),∀n≥0.
Assume that the sequences {θn} and {λn} satisfy the conditions contained in Theorem 3.5, then {xn} converges in norm to x* obtained in Corollary 3.3.

Acknowledgments

Ming Tian is Supported in part by The Fundamental Research Funds for the Central Universities (the Special Fund of Science in Civil Aviation University of China: No. ZXH2012 K001) and by the Science Research Foundation of Civil Aviation University of China (No. 2012KYM03).

LevitinE. S.PoljakB. T.Minimization methods in the presence of constraintsCalamaiP. H.MoréJ. J.Projected gradient methods for linearly constrained problemsPolyakB. T.SuM.XuH. K.Remarks on the gradient-projection algorithmYaoY.XuH.-K.Iterative methods for finding minimum-norm fixed points of nonexpansive mappings with applicationsCensorY.ElfvingT.A multiprojection algorithm using Bregman projections in a product spaceByrneC.A unified treatment of some iterative algorithms in signal processing and image reconstructionJungJ. S.Strong convergence of composite iterative methods for equilibrium problems and fixed point problemsLopezG.MartinV.Xu :H. K.CensorY.JiangM.WangG.Iterative algorithms for the multiple-sets split feasibility problemKumamP.A hybrid approximation method for equilibrium and fixed point problems for a monotone mapping and a nonexpansive mappingXuH.-K.Averaged mappings and the gradient-projection algorithmKumamP.A new hybrid iterative method for solution of equilibrium problems and fixed point problems for an inverse strongly monotone operator and a nonexpansive mappingBrezisH.CombettesP. L.Solving monotone inclusions via compositions of nonexpansive averaged operatorsYaoY.LiouY.-C.ChenR.A general iterative method for an infinite family of nonexpansive mappingsTianM.A general iterative algorithm for nonexpansive mappings in Hilbert spacesXuH.-K.Iterative algorithms for nonlinear operatorsGoebelK.KirkW. A.