简介:Storingthewholedatabaseinthemain-memoryisacommonmethodtoprocessreal-timetransactioninreal-timedatabasesystems.TherecoverymechanismofMain-memoryReal-timeDatabaseSystems(MMRTDBS)shouldreflectthecharacteristicsofthemain-memorydatabaseandreal-timedatabasebecausetheirstructuresarequitedifferentfromotherconventionaldatabasesystems.Inthispaper,therefore,weproposeamulti-levelrecoverymechanismformain-memoryreal-timedatabasesystemswithExtendibleChainedBucketHashing(ECBH).Owingtotheoccurrenceofreal-timedatainreal-timesystems,weshouldalsoconsideritinourrecoverymechanism.Accordingtoourperformancetest,thismechanismcanimprovethetransactionconcurrency,reducingtransactions'deadlinemissingrate.
简介:在这份报纸,一个分布式的一致协议在一般固定指导拓扑学下面与测量噪音为分离时间的单个整数的多代理人系统被建议。变化时间的控制获得令人满意随机的近似条件被介绍稀释噪音,这样靠近环的多代理人系统是内在地一个线性变化时间的随机的差别系统。然后,吝啬的方形的一致集中分析基于Lyapunov技术被开发,并且Lyapunov函数的构造特别不要求为二次的Lyapunov功能的存在假定的典型平衡网络拓扑学条件。因此,建议一致协议能对更一般的联网的多代理人系统适用,特别地当在代理人之间的双向或平衡的信息交换没被要求时。在建议协议下面,每个代理人的状态在吝啬的平方收敛到其数学期望是起始的状态珍视的代理人的加权的一般水准的一个普通随机的变量,这被证明;同时,随机的变量变化被围住。
简介:OneofthekeychallengesforimplementingRFIDsystemsinsupplychainmanagementisthedifficultyineconomicjustification.Suchdifficultyisfurtheramplifiedbyitspublicparticipationnatureasmultipleself-interestbeneficiariesmayreceivediversepaybacks,andtheirincentivestojointhesystemaredifficulttoalign.Thispaperaimstoaddresstheseproblemsbyacollaborativedesignfromtwoaspects.First,weproposetointroduceacentralizedplanningmechanisminthechaintofacilitatetheparticipation,sothatthecostoftheoverallsystemcanbeminimized.Second,weproposetoanalyzethemulti-faceteconomicreturnfrommulti-purposeapplicationstoachievethefullpotentialofRFIDsystems.Toillustrateourapproach,itsapplicationforinventoryinaccuracyandproductrecallinRFIDsystemispresented.
简介:Inthispaper,alocal-learningalgorithmformulti-agentispresentedbasedonthefactthatindividualagentperformslocalperceptionandlocalinteractionundergroupenvironment.Asforin-dividual-learning,agentadoptsgreedystrategytomaximizeitsrewardwheninteractingwithenvi-ronment.Ingroup-learning,localinteractiontakesplacebetweeneachtwoagents.Alocal-learningalgorithmtochooseandmodifyagents'actionsisproposedtoimprovethetraditionalQ-learningalgorithm,respectivelyinthesituationsofzero-sumgamesandgeneral-sumgameswithuniqueequi-libriumormulti-equilibrium.Andthislocal-learningalgorithmisprovedtobeconvergentandthecomputationcomplexityislowerthantheNash-Q.Additionally,throughgrid-gametest,itisindicatedthatbyusingthislocal-learningalgorithm,thelocalbehaviorsofagentscanspreadtoglobe.
简介:Inthisstudyamathematicalexpressionoftraceelementabundancerelationshipforthemul-ti-stagecomagmaticfractionalcrystallizationhasbeenestablished,basedongeochemicalstudiesoftheEmeishanbasalt-trachyteseriesandadjacentmafic-ultramaficlayeredintrusions,aswellasontheavail-abledataforbasalt,andesite,daciteandrhyoliteseriesinsouthernAndes,Chile,whichhavebeenwelldocumented.Itisdemonstratedthattheabundanceconstant(R)foragiventraceelementatdif-ferentstagesoffractionalcrystallizationofaparentalmagmaishighlyvariable,whichcanbeusedasacriteriontodividefractionalcrystallizationstages.
简介:
简介:在黑色基于Blandford-Znajek和磁性的联合进程的共存洞(BH)生长圆盘,为喷气力量的一个多参数模型和活跃星群的原子核(AGN)的收音机大声被学习。结果,AGN的收音机大声能被五个参数管理:(ⅰ)BH旋转,(ⅱ)变化的一个幂定律索引磁性在圆盘上删除;(ⅲ)决定圆盘的内部边的位置的一个参数,(ⅳ)到生长流动的最深处的部分的公羊压力的领带地平线上的磁场的压力的比率,并且(ⅴ)到地平线的户外Reldlines的尖速度的比率。观察两分在之间收音机大声并且收音机安静AGN被上述参数的效果很好解释。而且,我们独立关于每个参数讨论AGN的收音机大声的衍生物。另外,AGN的收音机大声上的螺丝钉不稳定性的效果被讨论。
简介:MetropolitanManila,thecapitalregionofthePhilippines,initiatedurbanforestryprojectstoshowcaseitsenvironmentalbenefits.MetroManilawasonceapartofatropicalmangroveforestdominatedbysmalltreeslocallycallednilad(Scyphiphorahydrophyllaceae).Thefast-pacedurbanizationtransformedtheoriginalecosystemintoabustlingmetropolissubjectedtoenvironmentaldecay.Theestablishmentofurbanforestryisoneoftheinterventionssoughtbythegovernmenttomitigatetheenvironmentalconditions.Thus,theconceptofparticipatoryurbanforestrywasformalizedin1988bythenationalgovernment.However,itwasduringtheearlyyearsofthe1970sthattheideaofexertingcollectiveeffortingreeningMetroManilastartedthroughtheGreenRevolution,withtheprimarymotiveofupliftingtheeconomicconditionofthepeopleratherthanonecologicalenhancement.Theprojectwasforcedonthepeopletoplantvegetablesandfruittreeswithlessemphasisonforesttreesspecies.Althoughtheprojectwasnotsuccessful,thegovernmentrealizedthatthecitizenrycouldbeagreatpartnerforacollectiveworthwhileundertaking.
简介:我们考虑被随机的震动和生产进程本身败坏的一个生产系统的预防维护。恶化的度由分离、有限的状态被建模。震动根据一个泊松过程到达并且败坏由随机的数量的系统。每当它生产一个项目时,系统可以败坏。系统连续地被监视并且修理系统状态是否在以上是为维护的预定水平。我们由于安装,操作,和修理从产品和费用分析一生,产品数量,平均支出,和平均利润考虑收入。假定系统参数和费用的结构,用数字例子,我们在平均利润和最大化平均利润的最佳的维护水平上调查生产和吃惊到达的影响。建议模型是适用的,转弯,打,并且钻例如,机器由于生产在穿的生产任务压过程,milling。
简介:Ageneralmethodofconstructingproxyblindsignatureisproposedbasedonmultilineartransform.Basedonthismethod,thefourproxyblindsignatureschemesarecorrespondentlygeneratedwithfourdifferentsignatureequations,andeachofthemhasfourformsofvariationsofsigns.Hencetherearesixteensignaturesinall,andallofthemareproxystronglyblindsignatureschemes.Furthermore,thetwodegeneratedsituationsofmulti-lineartransformarediscussed.Theircorrespondingproxyblindsignatureschemesareshown,too.Butsomeschemescomefromoneofthesedegeneratesituationsareproxyweakly-blindsignaturescheme.Thesecurityforproposedschemeisanalyzedindetails.Theresultsindicatethatthesesignatureschemeshavemanygoodpropertiessuchasunforgeability,distinguish-abilityofproxysignature,non-repudiationandextensivevalueofapplicationetc.
简介:Propertiesofdampingelectronsincollisionwithphotonsbasedonmulti-photonnonlinearComptoneffectareinvestigated.Theexpressionsofthedifferentialscatteringsectionarederived.Severalusefulconclusionsaredrawn.
简介:Ahierarchicalparticlefilter(HPF)frameworkbasedonmulti-featurefusionisproposed.TheproposedHPFeffectivelyusesdifferentfeatureinformationtoavoidthetrackingfailurebasedonthesinglefeatureinacomplicatedenvironment.Inthisapproach,theHarrisalgorithmisintroducedtodetectthecornerpointsoftheobject,andthecornermatchingalgorithmbasedonsingularvaluedecompositionisusedtocomputethefirstorderweightsandmakeparticlescentralizeinthehighlikelihoodarea.Thenthelocalbinarypattern(LBP)operatorisusedtobuildtheobservationmodelofthetargetbasedonthecolorandtexturefeatures,bywhichthesecond-orderweightsofparticlesandtheaccuratelocationofthetargetcanbeobtained.Moreover,abacksteppingcontrollerisproposedtocompletethewholetrackingsystem.Simulationsandexperimentsarecarriedout,andtheresultsshowthattheHPFalgorithmwiththebacksteppingcontrollerachievesstableandaccuratetrackingwithgoodrobustnessincomplexenvironments.
简介:photonic水晶的热性质有18个循环地分布式的核心的纤维(PCF)激光被使用完整向量的有限元素方法调查(女性)。结果证明18核心PCF与单个核心PCF和19核心PCF相比举办更有效的热分散建设。另外,有不同热负担的18核心PCF激光的温度分发被模仿。当热负担在近似对应于600W的抽的力量的80W/m上面时,结果证明核心温度接近近似拉1800K的价值的纤维,当当热负担在近似对应于110W的抽的力量的15W/m上面时,涂层温度接近大约550K的损坏价值时。因此冷却的纤维是必要的完成力量可伸缩。与另外的不同冷却系统相比,使计划凉下来的铜被发现是一个有效方法减少热效果。
简介:时间的不平等的一种新类型被讨论,它适用于算法的进程,包含处理联合起来的有限记忆。他们是Leggett-Grag,以及到由Brukner等的修改的一种选择。如果一个人考虑包含有限存储器的系统的量和古典进程的比较(在两个盒子中的一样的能力),不平等给我们能期望量加速的一条清楚的消息。在古典进程,一个人清楚地总是定义可能的大小的值,或以信息处理语言如果我们有一取决于到达各个的数据的某功能的顺序的计算走在一个算法上,函数总是有清楚地定义的值。在算法的结束以后,在量盒子中,仅仅最后的价值被定义。与Bohrs补充一致,所有中间的价值不能被归功于明确的价值。