如何做研究经验分享科学网.docx

如何做研究经验分享科学网.docx

《如何做研究经验分享科学网.docx》由会员分享，可在线阅读，更多相关《如何做研究经验分享科学网.docx（5页珍藏版）》请在冰豆网上搜索。

如何做研究经验分享科学网

MyStoryatPrinceton

IjustcompletedthePARSECprojectandstartedtolookfornewresearchprojects.IcameupwithanideathatIplannedtoworkonfor2~3yearswithagroupof3~5students.Idiscussedtheideawith Prof.KaiLi andaskedforhisadvice.  

IoriginallythoughtthatKaiwouldliketheidea,buthegavemeunexpectedfeedback.Hesaidthattheprojectistooambitioustobefinishedbyjustonepersoninashorttime.Furthermore,fromthemethodologyperspective,usually,thereisonlyonebestsolutionforonespecificgoal.Myidea violatedthenaturalrulethatachievingthreegoalsbydoingjustonething.HesuggestedmetofirstlylookforanideathatIcanmakeprogressoninthreemonths.

IhadtoadmitthatIwasstunnedbyhearingthecommentthatwasnotsupposedtobespokenoutfromKai'smouth.IusedtohearhimtalkabouthowtoidentifyaresearchtopicwhenIwasinICTseveralyearsago.HedisagreedwiththewayICTersdoresearchprojectswhosegoalsaretoprudentlycompletetherequirementsofproposals.HetoldICTersthatyouneedtosetupambitiousresearchprojectsaimingatglobaldemandsand youalsowanttotoleratefailuresincegoodresearchimplieshighrisk.

Inhisoffice,hekepttalkingabouthisopinion,experienceaswellassomeconcreteexamples.Hesaid,"ifyoudevoteyourselfonalong-termprojectwhenyouarejustajuniorfaculty,youwouldprobablynotgetanypublicationsforawhile,thensomeissueswillshowoff.Youmightencounterdoubt.Peoplewilldoubtyou.EspeciallyinChina,thereisperformancerevieweveryyear.Soyouneedtoescalateyourconfidencethroughtheprogressofaseriesofshort-termprojects.".Headded,"thisproposalissortofinappropriatebut I'mnotgonnatocriticizeyou.Actually thisisagoodexercise.Youcanlearnhowtoidentifyproperprojectsfromtheselessons."

AfterIwalkedoutfromKai'soffice,Ireflectedonmyoriginalwayofdoingresearch.ItseemsthatIalwayscomeupwithabunchofroughideasbutneverrefinethoseideastoseveralconcretechallenges.Thus,Ialwaysproposekindofbigprojects.Howtoidentifyaproperidea?

Infact,findinganideameansfindingagoodproblem.Ithinkthereshouldbeseveralcommonsteps.

AnoverviewofResearchFlow

Thefollowingchartillustratesmyunderstandingofhowtocarryonresearch.

CaseStudies

Let'slookattwocasestudies.

1.Dr. XinyuZhang wroteanarticleonhowhedidtheresearchthatendedupwithACMMobiCom2011bestpaperaward（"获得ACMMobiCom2011最佳论文奖",科学网）.Idrewthediagramtodescribehisexperienceaccordingtothearticle.

In2008,XinyutookacourseonrealsystemandlearnedtheideaofDVFSthatiswidelyusedinmicroprocessor.Heaskedaquestionthat"whynotapplyDVFStowirelessreceiver?

".Afterdoingsurveyandthinking,hefoundthereasonthatwirelessreceiverisrequiredtoobeyNyquist-ShannonSamplingTheoreminordertocorrectlyreceiveanddecodesignals.Itseemsimpossibletobreakthetheorem,sohethoughttheideawasnotdoable.Butin2010,onedaywhenhewasreadingpapers,hesuddenlycameupwithasolutionthatreceivinganddecodingcanbedecoupled.Then,hequicklyimplementedhisideaandverifieditsfeasibility.Finally,thisworkwontheACMMobicom2011bestpaperawardwhichledhimtobeafacultyofUniversityofWisconsinMadison.

InXinyu'sstory,Ithinkthekeyreasonhesucceededisthathehaddistilledthekeychallengein2008.

2.Prof.KaiLiwaselectedasamemberoftheNationalAcademyofEngineeringin2012forhiscontributionsinthefieldsofdatastorageanddistributedcomputersystems.Kaipioneeredthedistributedsharedmemory（DSM）techniquesthatallowuserstoprogramusingashared-memoryprogrammingmodelonclustersofcomputers.

IusedtoaskKaihowhecameupwiththegreatidea,andheansweredthatoriginallyhewasworkingonimplementingamessage-basedprogrammingmodelbyRPCthatwasjustproposedintheearly1980s.Buthestruggledwithhandlingpointersbetweendifferentmachines.Soheaskedaquestionthat"whynotuseglobaladdressspaceinthedistributedsystem?

". Hedelvedintothequestionandfoundthatthekeychallengewashowtomaintaindatacoherenceamongmultiplemachines.HeborrowedtheideaofCPUcachecoherenceprotocolandappliedittoDSM.Actually,hisprestigiouspaperonDSMisentitled"memorycoherenceinsharedvirtualmemorysystems",whichhasbeencitedbymorethan1600times.

Kai'sstoryexemplifiestheimportanceofdistillingkeychallengesaftercomingupwitharoughnewidea.

Actually,"howtodogoodresearch?

"Perhapsthisisthequestionthateveryresearcherusedtoaskhimselforherself.Thisarticlepresentsmythoughtandunderstandingonhowtodogoodresearch.BecautiousthatIamagoodresearcheryetbutjustajuniorguy.Youcommentsarewelcome.