更全的杂志信息网

算法随机性与逼近元

更新时间:2009-03-28

1 Introduction

In this paper we propose an effective version of Differentiation Theorem for Approximate Identities.Recent results have shown that algorithmic randomness is a very useful tool to study the effective aspects of classical analysis.Many theorems have been investigated,including the ergodic theorem([1,4]),Lebesgue differentiation theorem([7])and many other topics([3,2]).Generally,these results have this form:

A real xhas a certain property that holds almost everywhere

⇔xsatisfies a certain randomness notion.

Approximate identity is a very important topic in the study of functional and harmonic analysis.The Differentiation Theorem for Approximate Identities is a basic theorem concerning approximate identity.It says that for a k∈L1with a radially decreasing majorant such that kε=ε-1k(ε-1x),we have kε∗f→f,for all xexcept a null set as ε→0.

This theorem also indicate the Lebesgue differentiation theorem.It has been shown in[7]that the Lebesgue differentiation theorem can characterize Schnorr randomness.In this paper,our main result is that the Differentiation Theorem for Approximate Identities also characterize Schnorr randomness.And we will obtain an effective version of Lebesgue differentiation theorem which is slightly different from[7].

柿子炭疽病是柿树主要病害之一,富平尖柿属于极易感染炭疽病的品种。该病在高温高湿条件下传播蔓延快,防治难度大,只有准确区分病害症状特征,了解病害发生规律,并采取综合措施适时防治,才能取得良好效果。

In section 2,we present the necessary preliminaries.Section 3 and 4 contain the proof of the two directions of our main theorem respectively.

两种算法的运行时间以及ERLE参数如表1所示。输入信号为语音情况下的两种算法的失调量如图3所示,噪声输入情况下的失调量如图4所示。两组测试的结果一致,NLMS和FDNLMS算法的收敛时间、失调量、ERLE相差不大,但是,在处理速度方面,FDNLMS算法由于运算量大幅度降低,所用的处理时间只占NLMS算法的8.5%,说明算法的效率高,实用性强。

2 Preliminaries

The classical version of the Differentiation Theorem for Approximate Identities can be found in[5].To introduce this theorem,we begin with some definitions:

(一)转变观念,重新定位。新课改小学科学主要有两方面要求,首先以探究为其核心内容,其次是力求培养学生学习兴趣。此背景下小学不应再只依应试成绩为要求,则要重视学生的科学素养的培养,不仅是单纯地进行传授科学知识。另外也要改变对学生的评价方式,不能只以分数论好坏。同时教师进行拓展教学,主要指校方给学生提供可自主学习的内容,要成为学生学习科学的指导与引领者,花更多时间精力去帮助学生实践操作科学问题。

Definition 1(convolution) Let f,gbe in L1(R).The convolution f∗gis

 

Itisknownthatconvolutionsatisfiescommutativity,associativityanddistributivity.

Definition2 Anapproximateidentity(asε→0)isafamilyofL1(R)functionskε(0<ε≤1)with the following three properties:

在我国,青少年课业负担越发沉重起来的背景下,外部环境也逐渐发生一定变化,比方说网络和电子产品造成的冲击、社会浮躁的氛围等等,会对青少年的阅读兴趣造成一定影响。在阅读转化成一种负担的情况下,孩子的阅读兴趣自然没有办法得到保证,在此背景下产生了很多不良的阅读模式,诸如功利型阅读模式、浅层阅读模式以及快餐型阅读模式等。这些阅读模式实际应用的过程中,难以对阅读水平做出保证,并且阅读者不会积极地在阅读的过程中思考文章的深层意义。为了能够更好地继承并弘扬“耕读传家”这一我国优良传统,为了能够让更多的人参与到阅读当中,为了能够在整个社会当中创设出来更为优秀的阅读环境,全民阅读活动应运而生。

考虑到人们在工厂工作而Milk-run在生产供应路线上行驶的事实,必须解决某些安全问题。首先,火车必须可以自由进出Milk-run运输车道,并且路线上不应放置任何物料,因为这些因素可能对司机构成威胁并导致延误。为了使员工关注工厂内的交通情况,必须对车道进行清晰的标识,并且培训员工如何应对车辆的流通,Milk-run火车必须拥有绝对的优先通行权。

·There exists a constant c>0 such that for all ε.

· for all ε>0.

mials gksuch that‖f-gkL12-k,

Let B(x,θ)be the neighborhood(x-θ,x+θ)and Vcbe R-V.B(E,θ)denotes{y|∃x∈E(|y-x|<θ)}where E⊆R.We use χAas the characteristic function of A,where A⊆R.Next comes the basic theorem concerning approximate identities:

Theorem 1 ([5])Let kεbe an approximate identity on R.

(1)If f∈L1,then‖kε∗f-f‖L10 as ε→0.

(2)If fis continuous in a neighborhood of a compact set Eand bound on R,then kε∗fconverges uniformly to fon Eas ε→0.

In this paper we focus on a special kind of approximate identity.Let k(x)be a integrable function with integral one.Then we define kε:=ε-1k(ε-1x).It is straightforward to see kεis an approximate identity.

A function fis called radial,if f(x)=f(y)whenever|x|=|y|.If such fis decreasing on[0,∞)and f≥|g|in R,we call fa radial decreasing majorant of g.

Definition 3(Dominated Approximate Identity) Given a function k∈L1(R)such that,we call kεa dominated approximate identityif:

 

·khas a radial decreasing majorant Kwhich is continuous except at a finite number of points.

Definition 4 The function

 

is called the centered Hardy-Littlewood maximal functionof f.Actually,if we define,then we have

 

Here are two classical results which are useful in our proof.We refer to[5]for more details.

Lemma 1(2.1.12,[5]) If kεis a dominated approximate identity with the radially decreasing majorant K,then the estimate

 

is valid for all function f∈L1(R).

通过文献综述我们了解图式理论在中小学英语阅读教学的运用研究并非罕见。但是,以民族地区高中生为研究对象开展图式理论的实证研究却是凤毛麟角。语言的学习受环境和教学资源以及经济文化的发达程度的影响极大,因为语言的学习本身蕴含丰富的文化积累。然而民族地区的外语教育在这方面的研究存在很大缺口,因此,该研究具有及时性和研究价值。该研究的主要目的是调查图式理论在民族地区高中英语阅读教学中实际运用情况;了解语言图式、内容图式和结构图式与民族地区学生的英语阅读能力的相关性;探讨图式理论对民族地区英语阅读教学的启示。

Lemma 2(2.1.6,[5]) If f∈L1(R),then

 

Next comes the differentiation theorem for approximate identities,which can be viewed as a generalization of Lebesgue differentiation theorem:

Theorem2(DifferentiationTheoremforApproximateIdentities,2.1.17,[5]) Letkbea dominated approximate identity with a radially decreasing majorantK.Then kε∗f→fa.e.as ε→0 for all f∈L1(R).

Our job is to find the effective version of this theorem.The first thing we need to do is extending the computable notion from N to R.The next definitions can be found in[8].Let akbe a computable enumeration of all rationals on R.

Definition 5 A real number is computable if there are two computable functions h:NN and d:NN s.t.

 

This computable real xis coded by the indices of hand d.Let rebe the computable real on R indexed by e.

不同图案的图形密码安全性由高到低排列应依次是弦图式密码、八边形式密码、3*3式密码、正三角形式密码、等腰直角三角形式密码。

Definition 6 Eis a compact set on R.A function f:E→R is computable if:·fis sequentially computable:there is a computable function h:NN s.t.f(re)=rh(e)fis effectively uniformly continuous on E:there is a computable function d:NQ s.t.for all x,y∈E:

 

A function f:RR is computable if:

Lemma 5 ([7,6])If Aand Bare Σ1sets andµ(A)andµ(B)are computable,then A∩Band A∪Bare Σ1sets andµ(A∩B)andµ(A∪B)are computable.

·fis sequentially computable:there is a computable function h:NN s.t.f(re)=rh(e);

·fis effectively uniformly continuous on every compact subset:there is a computable function d:N2Q s.t.for all x,y∈[-N,N]:

 

The computable function fis coded by the indices of hand d.Notice that for every[-N,N],the functions sup[-N,N](f)andfdxare uniformly computable on N.([8])A function fis compactly supported,if f(x)=0 everywhere outside a compact set.

Definition 7 Eis a compact set of R.A function f:E→R is L1(E)-computable,or L1comp(E)for short if:there is a sequence of uniformly computable functions fkon E→R and a computable function d:NN s.t.

 

A function f:RR is L1(R)-computable,or L1comp(R)for short,if there is a sequence of uniformly computable compactly supported functions fkon RR and a computable function d:NN s.t.fkis supported on[-k,k]and:

 

f∈L1comp(R)is coded by the indices of fkand d.

It is easy to see that we can extend everyfunction to afunction uniformly.Notice that we actually define an L1-computable function fas an equivalence class on L1.We will select a representativefrom such equivalence class such thathas a certain value when xis Schnorr random.

Definition 8 A sequence of set UnR,n∈N is uniformly effectively open,or uniformly Σ1,if

 

where an,i,bn,iis a double sequence of uniformly computable reals.

Based on the conception of uniformly effectively open sets,we can define different versions of algorithmic randomness.Intuitively,a real is random if it avoids all‘effectively null’sets.The most commonly accepted randomness notions are Martin-Löf randomness and Schnorr randomness.

在Java程序设计里,线程的产生和运行都是通过Thread类来实现的。产生Thread类的对象是产生线程的唯一方法。Thread类定义了许多方法,帮助运行和处理线程,常用的方法如表1所示。

Definition 9 A Martin-Löf testis a uniformly Σ1sequence of UnR such thatµ(Un)2-nfor all n.A point x∈R is said to pass the test if x/nUn.xis Martin-Löf randomif it pass every Martin-Löf test.

A Schnorr testis a uniformly Σ1sequence of UnR such thatµ(Un)is uniformly computable on nandµ(Un)2-nfor all n.A point x∈R is said to pass the test if.xis Schnorr randomif it pass every Schnorr test.

Theorem3(EffectiveWeierstrassTheorem,[8])EisacompactsetonR.Afunctionfon Eis computable if and only if there is a computable sequence of rational polynomials pm(x)which converges effectively to fin uniform norm:there is a computable function d:NN such that|pm-f|≤2-nif m≥d(n).

A function fon R is computable if and only if for any N>0 there is a computable sequence of rational polynomials pm,N(x)which converges effectively to fon compact set[-N,N]i.e.,there is a computable function d:N×NN such that|pm,N-f|≤2-nif m≥d(n,N).

医用耗材SPD信息化建设打通了医院与上游供应商之间的信息互通,保证了耗材使用安全,有据可循,科学规范,改善了医疗服务质量,实现了医用耗材的精细化管理,降低了管理成本,加强和改善内部控制,为医院管理决策提供有力的依据。伴随信息技术的发展、行业政策的转变,医院的信息化建设将一直处于相对静止的动态变化过程中,并逐步向信息集成、资源共享的纵深领域发展。总之,信息化建设是促进医院现代化建设、提高医院管理效益和医疗服务质量、提升医院综合竞争力的不可缺少的必要因素,医用耗材SPD系统应用的深度和广度必将不断拓延,持续向低成本、高效率、高实用性的模式转变,为医院信息化建设提供有益借鉴。

for all x∈E.Notice that this δdepends on θ.

Proof See[8].□

Lemma 3 ([7,9])Suppose f∈L1comp[0,1]and fkis a sequence of uniformly computable rational polynomials such that‖f-fkL12-k.

·(Existence)The limit limk→∞fk(x)exists on all Schnorr random x.

In this section we will prove that the differentiation theorem for approximate identities holds for all Schnorr random x∈R.First,we show how it works when fis computable on a compact set.

·For any neighborhood Vof 0 we have as ε→0.

 

for all Schnorr random x

It is straightforward to see the result above also holds in L1([-N,N]).Note that if fis L1(R)-computable,then f↾[-N,N]is L1([-N,N])-computable,so actually this result also holds in L1(R).We useto denote limk→∞fk(x)when fkis the sequence as in the lemma above,which implies the value of(x)does not depend on the choice of fkwhen xis Schnorr random.

A dominated approximate identity kεis computable if kis L1-computable.Our main result is

土壤污染防治法规定,各类涉及土地利用的规划和可能造成土壤污染的建设项目,应当依法进行环境影响评价。各级人民政府生态环境、自然资源主管部门依法加强对矿产资源开发区域土壤污染防治的监督管理,严控可能造成土壤污染的重点污染物排放。

Theorem 4(effective version of Differentiation Theorem for Approximate Identities)Let kεbe a computable dominated approximate identity.Thenfor allif and only if xis Schnorr random.

We will use the following lemmas.

Lemma 4 ([7])Letsuch that Anis uniformly c.e.andµ(An)is uniformly computable andµ(An)2-n.Then Ais c.e.andµ(A)is computable.Moreover,this holds uniformly.

2.2.2 我国群众体育研究的发文机构分析 从发文机构(表2)的统计来看,以第1作者身份发文量最多的单位是南京师范大学,北京体育大学和西安体育学院次之,发文量在5篇及以上的机构共11家,其中体育专业院校2家,师范类院校4家。从这些分析可以看出,我国体育专业院校对于群众体育的研究力度还远远不够,体育专业院校作为我国体育学研究的主要贡献单位,应当为我国群众体育研究贡献自己的力量,从而推动我国群众体育的科学发展。

针对第二问题,首先,鉴于当前住房贷款规模空前扩大是导致居民负债率大幅攀升的主要原因,在房地产管控政策方面,应加快建立促进房地产市场平稳健康发展的长效机制,大力整治围绕房地产开发贷款的各类违规行为,在总量限制国内贷款流入房地产投资的同时,坚持因城施策,发展住房租赁市场,促进地区房地产市场的供求平衡。其次,进一步推进财税体制改革,使地方政府摆脱对土地财政的依赖。通过改革和完善财税体制,精简地方政府机构,以保证地方财政收入与支出的相匹配。这其中房地产税的落实将势在必行。

Let anR for n∈N and limnan=a.Then we say that it has a computable speed of approximation if there is a computable function hs.t.for all k≥h(n)we have|ak-a|<2-n.

3 Schnorr random points satisfy differentiation theorem for approximate identities

·(Uniqueness)Given another sequence of uniformly computable rational polyno

Lemma 6 ,thendyis a computable function of xon R.In particular, is computable.

According to effective Weierstrass Theorem,the‘a sequence of uniformly computable functions’in the definition of L1-computable function can be replaced by ‘a sequence of uniformly computable rational polynomials’.

Lemma7 Let kεbe a computable dominated approximate identity.Eis a compact set andFisacompactneighborhoodofE,B(E,θ)⊆F,fiscomputableonFandbounded on R,i.e.there is a M>0 s.t.|f|<M.Then kε∗f→funiformly as ε→0 on E.Moreover,the speed of this approximation is uniform computable on fF,Mand θ.

Proof Since kεis a dominated approximate identity,let c=‖kεL1=‖k‖L1.Let η>0 be an arbitrary real.We know that∫kε(y)dy=1,so we have

 

where Vis the neighborhood B(0).We need to find a sufficient small δ<1.Since fis computable on F,there is a δsuch thaif x∈Eand|y|<δ.So

 

Moreover,the pm(x)and pm,N(x)above can be obtained effectively from the index of f.

We know fhas a bound|f|<M.Then

 

Since kεis an approximate identity,the

It remains to calculate the speed of this convergence.Obviously it is determined by the speed of this convergence∫Vc|kε|dy→0 as ε→0.For convenience we let δbe a rational.Replace t=ε-1ywe have

NLMS系统的结构如图1所示,x(n)表示远端参考信号,v(n)表示近端信号,w(n)表示实际的回声路径,d(n)表示麦克风所采集到的信号,e(n)表示误差信号,y(n)表示表示输入,x(n)通过回声信道所得的回声信号,表示自适应回声抵消滤波器的权矢量即w(n)的估计值,表示自适应滤波器模拟回声信号的估计值。单远端模式下,v(n)为0。NLMS算法利用e(n)来对进行自适应调整,使其逼近w(n)。

 

It is a computable function of εby Lemma 6.So we use θand fto compute δ,then use δand kto find the ε0to ensurefor all ε≤ε0.We have

 

The following result can be viewed as an effective version of Egorov Theorem.

Lemma 8(Lemma 3.6,[7]) Letand E=[-N,N].fkis a sequence of uniformly computable rational polynomials on Esuch that

 

then there is a Schnorr test Umon Esuch that for any m,fkuniformly on E-Um.Moreover,the speed of this convergence is uniform on m.

ByLemma3,weknowthatlimforSchnorrrandomx.Thefollowing lemma is very useful in our proof.

Lemma 9(Lemma 3.3,[7]) Let Sbe a bounded set in R such that Sis first-order definableinthefieldofR.ThenthemeasureofSisacomputablerealnumber.Moreover,this holds uniformly on the first-order definition of S.

Theorem 5 Let kεbe a computable dominated approximate identity.Then for all f∈and Schnorr random xwe have

 

Proof Notice that the value kε∗f(x)ignores fon a null set of R,so we can assume f= without losing generality.The main idea is to prove the Lemma 8 also holds for kε∗f(x)and kε∗fk(x)for all ε.We assume x<N.Let Ebe[-N,N]and fkbe the sequence of polynomials as in Lemma 8(fk=0 outsides E).

Our idea is to construct such a chain as ε→0 and n→∞:

 

Firstly,let us deal with the first.Notice that

 

Since kεis a computable dominated approximate identity,it has a L1-computable radial decreasing majorant K.Since Kcan be arbitrary large we can choose such Kthat‖K‖L1(R)=ais a rational.By Lemma 1 we have

 

Define AiasBy the definition we know

 

It is easy to see that Aiis uniformly first-order definable on i,so by Lemma 9µ(Ai)is uniformly computable on i.By Lemma 2 we have

 

Let.It follows from Lemma 4 thatµ(Vs)is computable.We also have

 

So Vsis a Schnorr test.If x/∈Vs,then for all i≥2s:

 

Combining with the definition of Mand triangle inequality we can have

 

Let Gm=Vm∪Umwith Umas in Lemma 8.Vmand Umboth have measure limits,so by Lemma 5 Gmis a Schnorr test.Suppose x/∈Gm.Fix δ,search a sufficient large nsuch that|f-fn|(x)<δ/3 and supε>0|(kε(f-fn))(x)|<δ/3.Then by Lemma 7 there is a sufficient small εsuch that|(kε∗fn-fn)(x)|<δ/3.Then we have

 

This δcan be arbitrary small so we conclude lim.This proves the required conclusion. □

4 The points satisfy differentiation theorem for approximate identities are Schnorr

In this section we will prove the converse part of the main theorem.Actually,we will show that if xis not Schnorr random,the limit of kε∗f(x)may not exist.

Lemma 10 Given a computable dominated approximate identity kε,kε∗χA0 asµ(A)0.Moreover,the speed of this convergence does not depend on the specific A.

Proof Sincekεisacomputabledominatedapproximateidentity,thereisaL1-computable radial decreasing Ksuch that K≥k.Note that Kε(x)≥Kε(y)when|x|≤|y|.We have

 

Hence,kε∗χA(x)0 asµ(A)0.And the speed of this convergenceis uniform on ε. □

We will show that our result hold for computable xfirst.

Theorem 6 Given a computable dominated approximate identity kεand a computable real r,there is ansuch thatdiverges as ε→0.

We will use the similar technic used in Lemma 4.5 in[7]to prove our theorem.

Proof Suppose r<Nfor some natural number N.Set E=[-N,N].We will construct a uniformly computable sequence of 0,1-valued functions fnon R such that f=limfn.First,Let fnEc=0 for all n.So we only need to define fnE.Let δbe a small positive rational.We will also construct an computable sequence εnand an sequence of neighborhood Anof rs.t.:When nis even,kεn∗fn(r)<δand kεn∗fn+1(r)<δ;when nis odd,kεn∗fn(r)1and kεn∗fn+1(r)1.

Conduct the induction on n.

 

The next stage is to find a rational neighborhood A0⊂Eof rsuch thatµ(A0)2-1and|kε0∗χA0|(r)<δ.We define f1=f0+χA0.So kε0∗f1(r)<δ.f1(x)=1 for all x∈A0,by Lemma 7 there is a ε1such that kε1∗f1(r)1.

Stage n+1.

·Assumeniseven.ApplyLemma10toobtainarationalneighborhoodAn⊆An-1 of rsuch thatµ(An)2-n+1and|kεn∗fn|(r)+|kεn∗χAn|(r)<δ.Let fn+1=fn+χAn.So we still have kεn∗fn+1(r)<δ.Note that fn+1(x)=1 for x∈An,wecanapplyLemma7toobtainanεn+1suchthatkεn+1∗fn+1(r)1.

· If nis odd,then we need the Ansatisfy|kεn∗fn|(r)-|kεn∗χAn|(r)1.Let fn+1=fnAn.Here we have fn+1(x)=0 on An,so we can obtain an εn+1 such that kεn+1∗fn+1(r)<δ.

So actually for every n,we define fn+1An=1-fnand fn+1Acn=fn.By this construction we have

 

which means the kεn∗fnis always close enough to fnand Anis too small to change it.

ItiseasytocheckthatfnisuniformlyL1-computable.Definefasf(x)=limn→∞fn(x).We need f∈L1compso it does not matter fis undefined on a null set.Since‖f-fnL1<µ(An-1)2-n,we know fis a L1-computable function.This sequence kεn∗f(r)diverges since for all nit is always between kεn∗fn(r)and kεn∗fn+1(r),that is,kεn∗f(r)<δwhen nis even,and kεn∗f(r)1when nis odd. □

A finite decimal is a real with the form 2-nkand a decimal interval Iis an interval on R with the form[2-nk,2-n(k+1)]where nis a natural number and kis a integer.A decimal interval can be viewed as a interval on Cantor space.Two intervals are said to be almost disjointif their intersection has at most one element.It is known that for any Σ1set U,there is a uniformly computable decimal interval Iisuch thatandare almost disjoint.

Lemma 11 Let kεbe a computable dominated approximate identity.Eis a compact set and Fis a neighborhood of E,B(E,θ)⊆F.fis constant on Fand 0,1-valued on R.Then there is a computable function S:R+×R+R+such that for all ε≤S(θ,δ)we have

 

on E.

Proof It is a straight forward conclusion from the proof of Lemma 7.Note that this function Sonly depends on the majorant K. □

Theorem 7 Given a computable dominated approximate identity kεand a Schnorr testGm,we can construct asuch thadiverges for alwhere xis not finite decimal.

We will modify the technic used in theorem 6 to construct the desired f.

Proof SetE=[-N,N],clearlyGm∩EisaSchnorrtest.Sowithoutlosinggenerality we can assume that x∈Gm⊆Efor all m.We will construct a sequence of 0,1-valued function fnand f=limfn.Let fnEc=0 for all n.Let δbe a small positive real.

Stagesuch that I0,iare almost disjoint decimal interval.fn(x)=0 for all x∈E.

Just like the proof of previous theorem we need to find an εsuch that kε∗fis close enough to fon an interval I,but this general εdoes not exist.Actually,we need to cut each Iinto infinite parts,and find a specific εfor each part.

Stagesuch that Iis a computable sequence of pairwise n,ialmost disjoint decimal intervals.Fix i,let l,k∈N and In,i=[2-lk,2-l(k+1)].We define aj∈In,ifor j∈Z as follows:

·a0is the midpoint of In,i:

 

·For j>0,ajis the midpoint of[aj-1,2-l(k+1)]:

 

·For j<0,ajis the midpoint of[2-lk,aj+1]:

 

Then we define Jj=[aj,aj+1].To each jwe define an εj.By the property of approximate identity,find εjs.t. for all x∈Jjandεjsatisfies εj≤S(µ(Jj)/2/2)with Sdefined in Lemma 11.

Apply Lemma 10 to obtain αjsuch that|kεj∗χA|≤δ/4 ifµ(A)≤αj.Then we effectively find a large integer mjsuch that

 

Let.We construct such Hn,ifor all i∈N.Then define

Let

Due to the construction of Un+1we knowµ(Un+1)2-nfor all n>0.So by the same argument of the previous theorem we conclude f=limfnis L1-computable.

It remain to show that kε∗fdiverges a.SupposeUn.Let εjbe the real as in our construction.By the definition of function SWe have.Notice that,we then have

 

Combining the definition of εjit follows that

 

which means the kεj∗fnis always close enough to fnand Un+1is too small to change it.So

Thus for all,we have

 
 

Thus,kε ∗fdiverges at xas ε→0.□

Let the function kbe,we have this result directly:

Corollary 1

 

holds for all functionsif and only if xis Schnorr random.

This effective version of the Lebesgue differentiation theorem is slightly different from[7]:

Theorem 8 ([7])

 

holds for all functionsif and only if xis Schnorr random.

Compared with this theorem,our effective version of Lebesgue differentiation theorem requires the xto be the ‘center’of Q,and also extends the theorem from[0,1]to R.

References

[1]L.Bienvenu,A.R.Day,M.Hoyrup,I.Mezhirov and A.Shen,2012,“A constructive version of Birkhoff’s ergodic theorem for Martin-Löf random points”,Information and Computation,210:21-30.

[2]L.Bienvenu,R.Hölzl,J.S.Miller and A.Nies,2014,“Denjoy,Demuth and density”,Journal of Mathematical Logic,14(01):1450004.

[3]V.Brattka,J.Miller and A.Nies,2016,“Randomness and differentiability”,Transactions of the American Mathematical Society,368(1):581-605.

[4]J.Franklin,N.Greenberg,J.Miller and K.M.Ng,2012,“Martin-Löf random points satisfy Birkhoff’s ergodic theorem for effectively closed sets”,Proceedings of the American Mathematical Society,140(10):3623-3628.

[5]L.Grafakos,2008,Classical Fourier Analysis,New York:Springer.

[6]M.Hoyrup and C.Rojas,2009,“Computability of probability measures and Martin-Löf randomness over metric spaces”,Information and Computation,207(7):830-847.

[7]N.Pathak,C.Rojas and S.Simpson,2014,“Schnorr randomness and the Lebesgue differentiation theorem”,Proceedings of the American Mathematical Society,142(1):335-349.

[8]M.B.Pour-El and J.I.Richards,1989,Computability in Analysis and Physics,Berlin:Springer-Verlag.

[9]J.Rute,2012,“Algorithmicrandomness,martingalesanddifferentiability”:http://www.personal.psu.edu/jmr71/preprints/RMD1_paper_draft.pdf.

 
陈超 中山大学逻辑与认知研究所 中山大学哲学系 chench53.logic@gmail.com
《逻辑学研究》 2018年第02期
《逻辑学研究》2018年第02期文献

服务严谨可靠 7×14小时在线支持 支持宝特邀商家 不满意退款

本站非杂志社官网,上千家国家级期刊、省级期刊、北大核心、南大核心、专业的职称论文发表网站。
职称论文发表、杂志论文发表、期刊征稿、期刊投稿,论文发表指导正规机构。是您首选最可靠,最快速的期刊论文发表网站。
免责声明:本网站部分资源、信息来源于网络,完全免费共享,仅供学习和研究使用,版权和著作权归原作者所有
如有不愿意被转载的情况,请通知我们删除已转载的信息 粤ICP备2023046998号