We present a general result concerning the limit of the iterates of positive linear operators acting on continuous functions defined on a compact set. As applications, we deduce the asymptotic
behaviour of the iterates of almost all classic and new positive linear operators.
1. Introduction
The outstanding results of Kelisky and Rivlin [1] and Karlin and Ziegler [2] provided new insights into the study of the limit behavior of the iterates of linear operators defined on C[0,1]. They have attracted a lot of attention lately, and several alternative proofs and generalizations have been given in the last fifty years (see the references).
Nevertheless, the general problem concerning the overiterates of positive linear operators remained unsolved. This can be stated as follows.
Let X be a compact topological space, let C(X) be the linear space of all continuous real-valued functions defined on X and endowed with the norm ∥f∥∶=supx∈X|f(x)|, and let U:C(X)→C(X) denote a positive linear operator, that is, Uf≥0 for all f≥0. The problem is to provide sufficient conditions for the convergence of the sequence of iterates (Uk)k∈ℕ and find its limit, which is the goal of this paper.
Various techniques from different areas such as spectral theory, probability theory, fixed point theory, and the theory of semigroups of operators, have been employed in the attempts to find a solution (see, in chronological order, [1–22] and the references therein).
However, although many useful contributions have been made, the problem, in its generality, remained unsolved. The limit remained unknown for a long while even for the case restricted to classical particular positive linear operators.
For the first time, a solution to the general problem of the asymptotic behavior of the iterates of positive linear operators defined on C[0,1] was announced by the authors of this paper at the APPCOM08 conference held in Niš, Serbia, in August 2008. Related results appeared one year later in [19].
This paper describes the employment of a number of completely new methods in solving the general problem. To the best of our knowledge, this is the most general result known up to date. As an application of the main result, the asymptotic behavior of the iterates of many classical and new positive linear operators is deduced.
2. Notations and Preliminary Results
Throughout this paper, we will use the following notations:
X is a compact topological space;
C(X) is the normed linear space of all continuous real-valued functions defined on X; ∥f∥:=maxx∈X|f(x)|;
𝒱 is a linear subspace of C(X) including the space 𝒫0 of constants;
U:C(X)→C(X) is a positive linear operator preserving the elements of 𝒱;
L:C(X)→𝒱 is an interpolation operator;
YL is the interpolation set of L,
YL={y∈X∣Lf(y)=f(y),∀f∈C(X)}≠∅.
The existence of such an operator L is always assured. Indeed, for fixed x0∈X, the operator L:C(X)→𝒫0, Lf:=f(x0) is an interpolation operator with interpolation set YL={x0}.
We also emphasize that if U:C[0,1]→C[0,1] is a positive linear operator preserving the affine functions, then Uf interpolates f at the end points for all f∈C[0,1]. This well-known result is a particular case of a theorem of Bauer, see [23] and [24, Proposition 1.4]. It follows that YL⊇{0,1},
ei:[0,1]→ℝ are the monomial functions ei(x)=xi, i=0,1,…,
L1:C[0,1]→C[0,1] is the Lagrange interpolation operator L1f=f(0)e0+(f(1)-f(0))e1.
3. The Main Results
By using the notations presented in Section 2, the following theorem is the main result of the paper.
Theorem 3.1.
If there exists φ∈C(X) such that
Uφ≥φonX,Uφ≠φonX∖YL,
then
limkUkf=Lf,∀f∈C(X).
Moreover, if X is a compact metric space, then the convergence is uniform.
Proof.
Let f∈C(X). The case when Lf=f is trivial. Indeed, in this case, since Lf∈𝒱 and U preserves the elements of 𝒱, we have Uf=ULf=Lf, and hence Ukf=Lf, k=1,2,….
If Lf≠f, for sufficiently small ɛ>0, the inverse image of the open set (-ɛ,ɛ) under the continuous function Lf-f is an open set G, YL⊆G≠X. It follows that
|Lf-f|<ɛ,onG.
Since X is compact and G is open, it follows that X∖G⊆X∖YL is a nonempty compact subset of X, and we obtain
mɛ∶=infx∈X∖G(Uφ(x)-φ(x))>0.
Consequently, the following decisive inequality
|Lf-f|<ɛ+‖f-Lf‖mɛ(Uφ-φ)
is satisfied. By applying the positive operator Uk to (3.5), we get
|Lf-Ukf|<ɛ+‖f-Lf‖mɛ|Uk+1φ-Ukφ|.
Since Uφ≥φ, we obtain
φ≤Ukφ≤Uk+1φ≤‖φ‖,k=1,2,….
The sequence (Ukφ)k≥1 is monotone and bounded. It follows that it is pointwise convergent. Since ɛ was chosen arbitrarily, by using (3.6) we deduce that Ukf→pointwiseLf.
In the particular case when X is a compact metric space, since
Ukφ→pointwiseLφ∈C(X),
by Dini's Theorem, we obtain that Ukφ→uniformlyLφ. From the inequalities
‖Lf-Ukf‖<ɛ+‖f-Lf‖mɛ‖Uk+1φ-Ukφ‖,
we deduce that Ukf→uniformlyLf.
In the following we give more information on the limit operator L.
Theorem 3.2.
The limit interpolation operator L is unique, positive, and satisfies the equalities
UL=LU=L.
Proof.
The unicity and positivity of the operator L follow from the existence of limit limkUk=L. Since U preserves the elements of 𝒱, we obtain that UL=L. Taking into account the relations
∅⊊YL⊆{y∈X∣LUf(y)=Uf(y),∀f∈C(X)},
we can repeat the proof of Theorem 3.1 by starting with Uf instead of f. We deduce that limkUkf=LUf, and hence LU=L.
An immediate corollary of Theorem 3.1 is the following.
Corollary 3.3.
If U:C(X)→C(X) is a positive linear operator possessing an interpolation point x0∈X and there exists φ∈C(X) such that Uφ≶φ on X∖{x0}, then
limkUkf=f(x0),∀f∈C(X).
Proof.
In Theorem 3.1 we take Lf=f(x0).
Remark 3.4.
The existence of the function φ is essential here, in the sense that, if it is not satisfied, then the statement of Theorem 3.1 might not be true. Indeed, the positive linear operator U:C[0,1]→C[0,1] defined by
Uf(x)=xf(0)+(1-x)f(1)
preserves the space of constants 𝒫0, and the operator L:C[0,1]→𝒫0, Lf(x)=f(0), is interpolator on YL={0}. However, there exists no continuous function φ∈C[0,1] such that Uφ(x)>φ(x), for all x∈[0,1]∖{0}(see(Uφ(0)-φ(0))(Uφ(1)-φ(1))=-(φ(1)-φ(0))2≤0) and the sequence (Une1)n≥1 has no limit.
4. Applications
In this section, as applications of Theorem 3.1 and Corollary 3.3, we rediscover known results and obtain new ones concerning the asymptotic behaviour of the iterates of positive linear operators.
4.1. Positive Operators on C[0,1] Preserving Linear Functions
In the case of the particularisations, X=[0,1], 𝒱 is the space of all linear functions in C[0,1], L is the Lagrange interpolation operator of degree one associated to f at the endpoints 0 and 1, and YL={0,1}, by Theorem 3.1, we have that the following corollary holds.
Corollary 4.1.
Let U:C[0,1]→C[0,1] be a positive linear operator preserving the linear functions. If there exists φ∈C[0,1] such that Uφ≶φ on (0,1), then the sequence of the iterates of U converges uniformly to the Lagrange operator L1.
4.2. The Meyer-König and Zeller Operators
In 1960 Meyer-König and Zeller, see [25], introduced a sequence of positive linear operators which were studied, modified, and generalized by several authors. The classical Meyer-König and Zeller operators MKZn:C[0,1]→C[0,1], n∈ℕ, in the modified version of Cheney and Sharma, see [26], are defined byMKZnf(x)={∑k=0∞(n+kk)(1-x)n+1xkf(kn+k),x∈[0,1),f(1),x=1.
Moreover, from [27, Equation (2.4)], see also [28], we have thatMKZne2-e2≥(n+1)-1e1(1-e1)2>0on(0,1),n≥1.
For φ=e2, we have, as a consequence of Corollary 3.3, that the following corollary holds.
Corollary 4.2.
The sequence of the iterates of the Meyer-König and Zeller operators (4.1) converges uniformly to the Lagrange interpolation operator L1.
4.3. The May Positive Linear Operators
The May operators, see [29], are defined byMnf(x)∶=∫01f(t)ρn(x,t)dt,n∈N,
where ρn denotes a kernel function. They satisfyMne2-e2=λn-1(e1-e2)>0on(0,1),n≥1,
for some λ>0 and preserve linear functions. For φ=e2, as a consequence of Corollary 3.3, the following corollary holds true.
Corollary 4.3.
The sequence (Mnk)k∈ℕ of the iterates of the May operators (4.3) converges uniformly to L1.
4.4. The Bernstein Operator on a Simplex
Consider the simplex Sp in ℝp, p≥1, given bySp={x∶=(x1,…,xp)∣xi≥0,|x|∶=x1+…+xp≤1}.
The p+1 vertices of the simplex Sp are the points vi∈ℝp, wherev0=(0,0,…,0),v1=(1,0,…,0),⋮vp=(0,0,…,1).
WithM={m∶=(m1,…,mp)∣mi∈{0,1,…,n},|m|≤n},
the Bernstein approximation operator Bn:C(Sp)→C(Sp) is defined by Bnf(x)=∑m∈Mn!m1!⋯mp!(n-|m|)!x1m1⋯xpmp(1-|x|)n-|m|f(mn).
The operator Bn preserves the subspace of linear functionsV={f∣f(x1,…,xp)=a0+a1x1+⋯+apxp,a0,a1,…,ap∈R}.
The Lagrange interpolation operator L:C(Sp)→𝒱 is defined byLf(x)=(1-∑k=1pxk)f(v0)+∑k=1pxkf(vk)
and interpolates all functions in C(Sp) on the setYL={v0,…,vp}.
For φ(x)=x12+…+xp2, we haveBnφ(x)-φ(x)=1n(x1(1-x1)+⋯+xp(1-xp))>0,∀x∈Sp-YL,
and, by using Theorem 3.1, we get the following.
Corollary 4.4.
The sequence of the iterates of the Bernstein operator associated with the simplex Sp (4.8) converges uniformly to the Lagrange interpolation operator (4.10).
4.5. Positive Operators on C[a,b] Preserving e0 and e2
In [15] Agratini introduced a sequence of positive linear operators Λn:C[a,b]→C[a,b] preserving e0 and e2. In the case of the particularisations, D=[a,b], YL={a,b}, 𝒱=span{e0,e2}, φ=e4, and L is the interpolation operator:Lf(x)=1b2-a2(f(a)b2-f(b)a2+(f(b)-f(a))x2),
as a corollary of Theorem 3.1, we obtain a result of Agratini [15, Theorem 3.1].
4.6. Bernstein-Type Operators Preserving e0 and ej
Let n,j∈ℕ, n>j>1. Aldaz et al. [30, Proposition 11] had recently considered the Bernstein-type operators ℬj,n:C[0,1]→C[0,1],Bj,nf(x)=∑k=0n(nk)f((k(k-1)⋯(k-j+1)n(n-1)⋯(n-j+1))1/j)xk(1-x)n-k.
The operators ℬj,n satisfyBj,ne0=e0,Bj,nej=ej.
Considering the particularisations, D=[0,1], YL={0,1}, 𝒱=span{e0,ej}, φ=e2j, and Ljis the interpolation operator: Ljf=f(0)e0+(f(1)-f(0))ej,
as a corollary of Theorem 3.1, we obtain the following corollary.
Corollary 4.5.
The sequence (ℬj,nk)k∈ℕ of the iterates of the Bernstein-type operators converges uniformly to the operator Lj in (4.16).
4.7. The Cesàro Operator on C[0,1]
In the case when X=[0,1], 𝒱 is the space 𝒫0 of constants, Lf=f(0), YL={0}; φ=e1, by Theorem 3.1, we generalize the following recent result of Galaz Fontes and Solís.
Corollary 4.6 (see [16, Theorem 3]).
Let ρ∈C[0,1] be positive on (0,1) such that ∫01ρ(t)dt=1, and let 𝒞:C[0,1]→C[0,1] be the Cesàro mean operator,
Cf(x)=∫01f(tx)ρ(t)dt,x∈[0,1].
Then,
Ckf→uniformlyf(0),∀f∈C[0,1].
4.8. The Bernstein-Stancu Operators
Let b>0. The Bernstein-Stancu operators Sn,0,b:C[0,1]→C[0,1] (see, e.g., [31]),Sn,0,bf(x)=∑k=0n(nn)(1-x)n-kxkf(kn+b),
satisfy the following:Sn,0,be0=e0,Sn,0,be1=nn+be1.
For X=[0,1], x0=0 and φ=e1, by Corollary 3.3, we obtain a result concerning the iterates of the Bernstein-Stancu operators.
Corollary 4.7.
The sequence of the iterates of Stancu's operators (4.19) converges uniformly to f(0).
4.9. The Cheney-Sharma Operator
Let tn≥0, n∈ℕ and let CSn be the nth Bernstein-Cheney-Sharma operator on C[0,1], defined byCSnf(x)=(1+ntn)-n∑k=0n(nk)x(x+ktn)k-1(1-x+(n-k)tn)n-kf(kn).
It is known that 0≤CSne1≤e1/(1+tn) (see, e.g., [32] and [3, (5.3.7)]. For tn>0 one has thatCSne0=e0,0≤CSne1≤11+tne1.
Taking X=[0,1], x0=0, and φ=e1, in Corollary 3.3, we obtain the following application.
Corollary 4.8.
The sequence of the iterates of the Cheney-Sharma operators (4.21) converges uniformly to f(0).
4.10. The Schurer Operator
For p,n∈ℕ, n≥p, the Bernstein-Schurer-type operator [3, (5.3.1)], ßn:C[0,1]→C[0,1] is defined byBnf(x)=∑k=0n-pf(kn)(n-pk)xk(1-x)n-p-k.
One can prove that this operator satisfiesBne0=e0,Bne1=n-pne1.
In the case, when X=[0,1], x0=0, and φ=e1, by Corollary 3.3, we obtain the following result.
Corollary 4.9.
The sequence of the iterates of the Schurer operators (4.23) converges uniformly to f(0).
4.11. Piecewise Bernstein Operators
Let f∈C[a,b]. Consider the Bernstein operator Bn,a,b:C[a,b]→C[a,b]:Bn,a,bf(x)=1(b-a)n∑k=0n(nk)(b-x)n-k(x-a)kf(a+k(b-a)n),n=0,1,….
In the case of the particularisations, X=[a,c], 𝒱 is the linear space of polygonal lines with vertices possessing abscissae at a,b, and c, Lf is the polygonal line with vertices at (a,f(a)), (b,f(b)), (c,f(c)), YL={a,b,c}, and φ(x)=x2, by Theorem 3.1, we have that the following corollary holds.
Corollary 4.10.
For a<b<c, consider the composite Bernstein operator B:C[a,c]→C[a,c],
Bf(x)={Bn,a,bf(x),x∈[a,b],Bn,b,cf(x),x∈[b,c].
Then,
Bkf→uniformly{L[a,b;f],on[a,b],L[b,c;f],on[b,c]∀f∈C[0,1].
Acknowledgments
The authors thank the referee for useful comments and suggestions. This research was supported by the Romanian CNCSIS Grant ID 162/2008 and by the Project PN2-Partnership no. 11018 MoDef.
KeliskyR. P.RivlinT. J.Iterates of Bernstein polynomials1967215115200212457ZBL0177.31302KarlinS.ZieglerZ.Iteration of positive approximation operators19703310339027798210.1016/0021-9045(70)90055-9ZBL0199.44702AltomareF.CampitiM.199417Berlin, GermanyWalter de Gruyterxii+627de Gruyter Studies in MathematicsAppendix A by Michael Pannenberg and Appendix B by Ferdinand Beckhoff1292247GonskaH. H.ZhouX. L.Approximation theorems for the iterated Boolean sums of Bernstein operators1994531213110.1016/0377-0427(92)00133-T1305965ZBL0816.41020AdellJ. A.BadíaF. G.de la CalJ.On the iterates of some Bernstein-type operators1997209252954110.1006/jmaa.1997.53711474623ZBL0872.41009OruçH.TuncerN.On the convergence and iterates of q-Bernstein polynomials2002117230131310.1006/jath.2002.37031924655OstrovskaS.q-Bernstein polynomials and their iterates2003123223225510.1016/S0021-9045(03)00104-71990098ZBL1093.41013KingJ. P.Positive linear operators which preserve x2200399320320810.1023/A:10245711264551973095ZBL1027.41028RusI. A.Iterates of Bernstein operators, via contraction principle2004292125926110.1016/j.jmaa.2003.11.0562050229ZBL1056.41004ItaiU.On the eigenstructure of the Bernstein kernel2006254314382280385ZBL1160.42317GonskaH.KacsóD.PiţulP.The degree of convergence of over-iterated positive linear operators2006144034232220800ZBL1099.41011GonskaH.RaşaI.The limiting semigroup of the Bernstein iterates: degree of convergence20061111-211913010.1007/s10474-006-0038-42188976ZBL1121.41004GonskaH.PiţulP.RaşaI.Over-iterates of Bernstein-Stancu operators200744211712510.1007/s10092-007-0131-22330763ZBL1150.41013WenzH.-J.On the limits of (linear combinations of) iterates of linear operators199789221923710.1006/jath.1996.30391447840ZBL0871.41014AgratiniO.On the iterates of a class of summation-type linear positive operators20085561178118010.1016/j.camwa.2007.04.0442394357ZBL1151.41013Galaz FontesF.SolísF. J.Iterating the Cesàro operators200813662147215310.1090/S0002-9939-08-09197-12383520ZBL1146.47019AbelU.IvanM.Over-iterates of Bernstein's operators: a short and elementary proof2009116653553810.4169/193009709X4704342519492BadeaC.Bernstein polynomials and operator theory2009533-422923610.1007/s00025-008-0333-12524724ZBL1198.47028RasaI.Asymptotic behaviour of certain semigroups generated by differential operators20091127362597905ZBL1181.47047GavreaI.IvanM.On the iterates of positive linear operators preserving the affine functions2010372236636810.1016/j.jmaa.2010.07.0262678868ZBL1196.41014RasaI.C0 semigroups and iterates of positive linear operators: asymptotic behaviour20102supplement 82123142GavreaI.Ioan.Gavrea@math.utcluj.roIvanM.Mircea.Ivan@math.utcluj.roOn the iterates of positive linear operatorsJournal of Approximation Theory. In press10.1016/j.jat.2011.02.012BauerH.Šilovscher Rand und Dirichletsches Problem196111891360136983ZBL0098.06902PhelpsR. R.1966Princeton, NJ, USAD. Van Nostrandv+1300193470Meyer-KönigW.ZellerK.Bernsteinsche potenzreihen19601989940111965ZBL0091.14506CheneyE. W.SharmaA.Bernstein power series1964162412520179522ZBL0128.29001BeckerM.NesselR. J.A global approximation theorem for Meyer-König and Zeller operators19781603195206051074510.1007/BF01237033ZBL0376.41007SikkemaP. C.On the asymptotic approximation with operators of Meyer-König and Zeller1970324284400276660ZBL0205.08101MayC. P.Saturation and inverse theorems for combinations of a class of exponential-type operators1976286122412500435682ZBL0342.41018AldazJ. M.KounchevO.RenderH.Shape preserving properties of generalized Bernstein operators on extended Chebyshev spaces2009114112510.1007/s00211-009-0248-02557868ZBL1184.41011StancuD. D.Approximation of functions by a new class of linear polynomial operators196813117311940238001ZBL0167.05001CheneyE. W.SharmaA.On a generalization of Bernstein polynomials1964577840198074ZBL0146.08202