[PDF] The Concurrency Column work on type systems for

View - Download The Concurrency Column

Previous PDF

Next PDF

TheConcurrencyColumn

by

NobukoYoshida

Departmentof Computing

ImperialCollege London

180Queen's Gate,London,SW72AZ

Interviewswiththe2020CONCUR

Test-of-TimeAwardRecipients

LucaAceto

ICE-TCS,Schoolof ComputerScience,

ReykjavikUniversity

GranSassoScience Institute,L'Aquila

luca@ru.is,luca.aceto@gssi.it Earlierthisyear ,theInternational ConferenceonConcurrencyTheory(CON- CUR)andthe IFIP1.8 WorkingGroup onConcurrency Theorydecidedto estab- lishtheCONCUR Test-of-Time Award torecognizeimportantachievementsin ConcurrencyTheorythatwerepublished attheCONCUR conferenceandha ve stoodthetest oftime.All paperspublishedat CONCURbetween1990 and1995 wereeligiblefor thefirstinstallment ofthea ward,which willbepresented at the31stInternational ConferenceonConcurrenc yTheory(CONCUR 2020).The conferencewill beheldon lineinthe period1-4September 2020,andhas Igor KonnovandLauraKov acsaschairs oftheprogram committee. Theaw ardwinnersfortheCONCURTest-of-Time Awards 2020,together withthea ward citations,maybefoundat Theywereselectedby ajurycomposed ofJosBaeten, PatriciaBouyer -Decitre, HolgerHermanns, AlexandraSilv aandmyself(aschairof thejury). Inorderto giv evisibilityto theCONCURTest-of-TimeAward,and toinspire youngresearchersto buildon thework oftheawardrecipients andto learnfrom theirexperiences, Idecidedtointerview allthea wardrecipients, namelySøren Christensen,Robv anGlabbeek, HansHüttel,NancyLynch, DavideSangior gi, RobertoSeg alaandColinStirling.Ireceived positive answersfromall ofthem, apartfromSøren Christensen,whohas worked inthepri vate sectorsincehe fin- ishedhisPhD, andColinStirling, whohasretired lastspring. Theinterviews canbeaccessedfrommy professionalblogat https://processalgebra.blogspot.com/ andfromthe above-mentio nedconferenceweb page.Thiswrite-upcollectsthem underoneroof andIhope thatitwill makethem easilyaccessibleto ev eryone.I encourageyoungresearchers ofall agestoread themandbe inspiredbythe work ofthose distinguishedcolleaguesof ours.

InterviewwithRobv anGlabbeek

Luca:Youreceiveone ofthetwoCONCURTest-of-TimeA wardsfor thepe- riod1990-1993for yourcompanionpapers on"TheLinear Time-BranchingT ime Spectrum",publishedat CONCUR1990and 1993.Couldyou tellusbriefly what spurredyou toembarkon theencyclopaedic studyofprocess semanticsyoude- velopedinthosetwo papersandwhat wereyourmain sourcesofinspiration? Rob:MyPhDsupervisors, JanWillem KlopandJan Bergstra,were examining bisimulationsemantics, butalso failuressemantics,incollaborationwith Ernst- RüdigerOlderog,and atsome point,togetherwith JosBaeten,wrote apaperon ready-tracesemanticsto dealwith apriorityoperator .Ihad aneedto seethese activities,andalsothoseof Tony HoareandRobin Milner,as revealing di!erent aspectsof thesamegreater whole,ratherthan asisolatedresearch e!orts.Thisled naturallytoan abstractionoftheir workin whichthecrucial di!erencewas the semanticequiv alenceemployed.Orderingthesesemanticsinalattice wasnever optionalforme, butnecessary toseethe largerpicture.Itmaybe inmy nature torelatedi !erentsolutionsto similarproblemsI encounterbyplacing themin lattices. Luca:Didyouimagine atthetime thatyoura ward-winningpapers wouldha ve somuchimpact? Rob:Ididn't thinkmuchaboutimpactinthose days.Iwrote thepapersbecause Ifeltthe needtodo so,andgi venthat, couldjustas wellpublishthem. HadI madeanestimate, Iw ouldhav eguessedthat mypaperswouldhav eimpact,as Iimaginedthat thingsthatI findnecessaryto knoww ouldinterestsome others too.ButI wouldha veunderestimated theamountofimpact,inpart becauseIhad grosslyunderestimatedthe sizeofthe researchcommunitye ventuallyw orkingon relatedmatters. Luca:Inyourpapers, youpresentcharacterizations ofprocesssemantics inrela- tionalform,via testingscenarios,via modallogics andinterms of(conditional) equationalaxiomatizations.Of allthose characterizations,whichones doyou thinkhav ebeenmostusefulinfuturework?T oyourmind, whatarethe most interestingorune xpectedusesin theliteratureofthenotionsand techniquesyou developedinthe"TheLinearTime-Branching TimeSpectrum"? Rob:Ithinkthat allthesecharacterizations togethershedthe lightonprocess se- manticsthat makesthem understandableanduseful.Relationalcharacterizations, i.e.,of(bi)simulations, areoftenthe mostusefulto graspwhatkind ofrelation one has,andthis isessentialfor anyuse. Butmodalcharacterizations areintimately Equationalcharacterizationsalso findpracticaluse inverification, e.g.,in normal- ization.Yet, theyaremostusefulin illustratingtheessentialdi!erencebetween processsemantics, andthusform aguidein choosingtheright oneforan appli- cation.To me,thetestingscenariosarethe mostusefulin weighinghow realistic certainsemanticidentifications areincertain applicationcontexts. Luca:Howmuchofyourlater workhas builton youraw ard-winningpapers? Whatfollow-up resultsofyoursareyoumost proudofand why?

Rob:Aboutone third,Ithink.

•Mywork onactionrefinement,togetherwithUrsula Goltz,extends theclas- sificationof processsemanticsin adirectionorthogonal tothelinear time -branchingtime spectrum.Iam proudof thiswork mostlybecause,for adistinctclass ofapplications,action refinementisa greatcriterionto tell whichsemanticequi valencesone oughttopreferoverothers.Thesepro- cesssemanticsha vealso beenusedinlaterworkondistributability with Ursulaandwith Jens-WolfhardSchick eU!mann.Distributability tellsus whichdistributed systemscanbecastas sequentialsystemscooperating •Iwrotese veral papersonStructuralOperationalSemantics,manyjointly withWan Fokkink,aimingtoobtain congruenceresultsforsemanticequiv- alences.Thesew orkscarefullyemplo yedthemodalcharacterizationsfrom myspectrum papers.Iconsider thiswork importantbecausecongruence propertiesareessential inverification, asweneed compositionalitytoa void stateexplosions. •Iuseda versionof failuresimulation semanticsfrommysecondspectrum paperin workwith YuxinDeng,MatthewHennessy ,CarrollMor ganand ChenyiZhang,characterizingmay-and must-testingequiv alencesforprob- abilisticnondeterministicprocesses. Probabilityand nondeterminisminter- actinman yapplications,and Ithinkthatmay-andmust-testing aretheright toolstosingle outtheessence oftheirsemantics. Luca:IfIremember correctly,your masterthesisw asinpuremathematics.Why didyoudecide tomov etoa PhDincomputer science?Whatwasitliketow ork asaPhD studentatCWI inthelate 1980s?How didyoustart yourcollaborations withJosBaeten andfellow PhDstudentFrits Vaandrager, amongstothers? Rob:Iwas interestedinmathematicallogicalreadysince highschool,after win- ninga bookonthe foundationsoflogic andsettheory inamathematics Olympiad. Ifoundthe materialinmy logiccourse,thought attheUni versityof Leidenby JanBergstra, fascinating,andwhendoinga relatedexamatCWI,where Janhad movedtoatthattime,heask edmeto doaPhD onrelatedsubjects, Isa wnorea- sonnot to.Buthe hadtocon vincemefirst thatmyw orkwould notinv olveany computerscience, asmystudent advisoratThe University ofLeiden,Professor Claas,hadtold usthat thiswas notareal science.Fortunately ,Janhad noproblem passingthathurdle. Iwas thethirdmemberoftheteam ofJanBer gstraandJan Willem Klop,after JosBaetenhad beenrecruited afew monthsearlier, butin amoresenior position. Fritsarriv edamonthlaterthanme.Aswewere exploringsimilar subjects,and sharedano "ce,itw asnaturalto collaborate.Asthesecollaborationswork edout verywell,they becamequitestrong. CWIwas amarvelousenvironment inthelate 1980s.Ourgroupgradually grewbeyonda dozenmembers,butthereremained amuchgreater cohesionthan Iha veseenanywhereelse.Wehad weeklyprocessalgebra meetings,whereeach ofusshared ourrecentideas withothersin thegroup.At thosemeetings,if one memberpresenteda proof,theentire audiencefollowed itstepby step,contrib ut- ingmeaningfullywhere needed,andnot asingleerror wentunnoticed. Luca:AlreadyasPhD student,you tookupleadership roleswithinthe commu- nity.Forinstance,I rememberattendingaworkshopon "CombiningComposi- tionalityandConcurrenc y"inMarch 1988,whichyouco-organized withUrsula GoltzandErnst-Rüdiger Olderog.Would yourecommendto aPhDstudent that he/shebecomein volved inthatkindofactivitiesearlyonintheircareer?Do you thinkthatdoing sohad apositiv ee!ectonyour futurecareer? Rob:GenerallyIrecommend this.AlthoughI co-organized thisworkshop be- causethe circumstanceswereright forit,and theneedgreat; notbecauseI feltit necessarytoor ganizethings. MyadvicetoPhDstudentswouldnotbe specifically toadda leadershipcomponentto theirCVre gardlessof thecircumstances, butin- steaddotak esuchan opportunitywhenthetimeisright, andnotbe shybecause oflackof experience. Thisworkshop definitivelyhada positivee!ectonmy futurecareer .When finishingmyPhD Iinquiredat StanfordandMIT forapossible post-docposition, andthegood impressionsIhad leftatthis workshopwere afactor ingetting o!ers frombothinstitutes. Infact, VaughanPratt, whowas deeplyimpressedwithwork Ihadpresented atthisw orkshop,convinced metoapply straightaw ayforan assistantprofessorship,and thisledto myjobat Stanford. Luca:Istillha vevi vidmemoriesofyourtalkatthefirstCONCUR conference in1990in Amsterdam,whereyou introducednotionsof black-boxexperiments onprocessesusing anactual boxandbeer -glassmats.Y ouhav eattendedand contributedtomany oftheeditions oftheconferencesincethen.T oyourmind, howhasthefocusof CONCURchangedsince itsfirstedition in1990? Rob:Ithink thefocuson foundationshasshifted somewhatto moreappliedtop- ics,and thescopeof"concurrencytheory"hasbroadenedsignificantly.Still,many topicsfromthe earlierCONCURsare stillcelebratedtoday . Luca:Thelastforty yearshav eseena hugeamountof workonprocessalge- braandprocess calculi.Howe ver, theemphasison thatfieldofresearchwithin concurrencytheoryseemstoha vediminished over thelastfe wyears,eveninEu- rope.What advicewould yougiv etoayoungresearcherinterested inworking on processcalculitoday? Rob:Until10years ago,many peopleImet inindustryv oicedtheopinionthat formalsmethodsin general,andprocess algebrainparticular ,hasbeen triedin theprevious century,andhadbeenfound tonotscaletotacklepractical problems. Butmorerecently thisopinionis onthew ayout,in partbecause itbecameclear thatthekind ofproblemssolv edbyformal methodsaremuch morepressingthan originallybeliev ed,andinpartbecausemanyofthe applicationsthatwere deemed toohardin thepast,are nowbeing addressedquiteadequately .For thesereasons, Ithinkw orkonprocess calculiisstillagoodbet foraPhD study, althoughareal- worldapplicationmotiv atingthetheory studiedisperhapsharderneededthanin thepast,and toye xamplesjustto illustratethetheoryarenotal waysenough. Luca:Whatarethe researchtopicsthat currentlyexcite youthemost? Rob:Iamv eryexcited aboutshowingliv enessproperties,inLamport'ssense, tellingthata modeledsystem eventually achieves adesirableoutcome. Andeven moreincreating processalgebraicv erificationframew orksthatare inprinciple abletodo this.Iha vecome tobeliev ethattraditional approachespopularin processalgebraand temporallogiccan onlydealwith liveness propertieswhen makingfairness assumptionsthataretoostrongand leadtounw arrantedconclu- sions.Together withPeterHoefnerIhav ebeenadv ocatingaweak erconcept of fairness,thatwecalljustness, thatismuch moresuitableas abasisfor formulat- ingliv enessproperties.Embeddingthisconceptintotheconceptualfoundations ofprocessalgebra andconcurrenc ytheory, insucha waythatitcanbe e!ectively usedinv erification,isfor ameachallengingtask,in volvingman yexciting open questions. Avaguely relatedsubjectthatexcitesme sincetheend oflastyear ,is theex- tensionofstandard processalgebra withatime-out operator,while stillabstaining fromquantifyingtime. Ibeliev ethatsuch anextension reachesexactlytheright levelofexpressivenessnecessary formany applications. Bothtopicsalso relatewith theimpossibilityof correctlyexpressing mutual exclusioninstandardprocessalgebras, adisco verythat calledforthman yphilo- sophicalquestions, suchasunder whichassumptionson ourhardware ismutual exclusion,whenfollowingthe formaldefinitions ofDijkstraandothers,ev enthe- oreticallyimplementable.

InterviewwithHansHüttel

Luca:Youreceiveone ofthetwoCONCURTest-of-TimeA wardsfor theperiod

1990-1993forthe paper"BisimulationEqui valenceis Decidableforall Context-

FreeProcesses"you publishedwithSøren ChristensenandColin StirlingatCON- CUR1992.Could youtellus brieflywhatthe contextfor thatwork wasand how itcameabout? Hans:Iwas workingonmyPhD whichwasonthetopic ofdecidabilityof be- haviouralequivalencesfor processcalculithatallowprocessestohav einfinite statespaces.Baeten, Bergstraand Klophadpro vedthatstrongbisimilarity is infact decidablefortheBPAcalculus even thoughtheclass oftrace languages forBPA isexactlytheclassof context-freelanguages. Theresultonlyheldfor normedBPA processes,thatis,processesthatcould always terminate.Colinw as mysupervisorin Edinburgh, andhesuggested thatItrytofindasimplerproof ofthatresult (muchcanbe saidaboutthe originalproof,b utsimpleit was not!). Thisturnedout tobe reallyinteresting;I metDidierCaucal fromIRISAwho had founda nice,muchshorter proofthatrelied onfinitecharacterizations andIw as abletouse hisnotionof self-bisimilaritytopro vethat atableausystem forbisim- ilaritywas soundandcompletewithrespectto bisimilary.Jan FrisoGroote,who wasaPhDstudentat theCWIat thetime,and myselfprov edthatall otherkno wn equivalencesareundecidablefornormedBP A(andtherefore afortiorialsofor full,unnormedBP A).Butone problemthatColinandIwere never ableto solve wasifthedecidabilityresult alsoheld forthefull BPAcalculus. Sørenarriv edinEdinburghin1990and weshareda flatthere,inStLeonards Bankjustacross fromHolyroodP ark.We hungoutwith thesamepeople inthe LFCS,many ofwhomwereclever Italianssuchas DavideSangior giand clever quasi-Italianssuch asMarceloFiore. Itisno surprisethatSøren andIoften dis- cussedour workor thathe,giventhat Colinwas alsohissupervisor ,mov edon tostudydecidability problemsas well,theonly di!erencebeing thatSørenw as mostlyinterestedin theBPPcalculus thathasparallel compositioninsteadof se- quentialcomposition. WhenIleft Edinburgh attheend ofAugustof1991,ColinandIhadman- agedtomak esomeprogress onthedecidabilityproblemforunnormed BPA.W e realisedthata finitecharacterizationof themaximalbisimulation álaCaucal was theway aheadbuttheactualcharacterization escapedus.When Ireturnedformy vivainDecember,Sørenhadfound animportantlemma onfinitecharacteriza- tionsande verythingfell intoplace.Thedecidabilityresultreliedonproving that bisimilarityissemi-decidable usingthefinite characterization;non-bisimilarity wasalreadyknownto besemi-decidable. Luca:BothSørenChristensen andyouwrote thea ward-winningpaper asPhD studentsinEdinb urgh.Could youtellusabouttheresearchenvironmentin Ed- inburghatthattime,andtheinfluence ithadon youthenand intherest ofyour career? Hans:TheLFCSin Edinburgh wasa greatplacetobeatthetime.RobinMilner wasstillaroundandb usyworking onthepi-calculus togetherwithDa videSan- giorgi,whowashis studentatthe time.Watchingthematthe blackboardwas interesting,ev enthoughtherestofusoftencouldnot followtheir discussions! GordonPlotkinw asthere(and stillis,fortunately).RodBurstallw asstillacti ve, andsow asMike Fourman.Plusmany, manyothers.Therewas alwayssomeone totalkto, andit wasperfectly legitimateto have aninterestinbasicresearch.Un- liketheotherprofessors,Colin hadnobackground inmaths- hewas originallya philosopher!-and maybethatw aswhy hewas tryingto avoidheavymathemati- callingo,trying tobe simpleandprecise inhiswriting atthesame time.Ilearned alotfrom him,alsoin thatrespect. AfterSørenreturned toDenmark,he leftacademiaand gotajob inthe private sector.Hestillliv esnearAarhus. Sadlywelost touchwitheachother,inall likelihoodbecauseourliv esturned outsodi !erent.We gotbackintouchrecently , whenwewere toldabout therew ardandwe lookforward tofinallymeeting each otheragain. Colinretiredrecently,andI reallyhopeto seehimag ain,whenIam abletotra vel toScotlandagainsomeday. Luca:Howmuchofyourlater workhas builton youraw ard-winningpaper? Whatfollow-up resultsofyoursareyoumost proudofand why? Hans:Iwork edondecidabilityissuesforanotherfew yearsafterthat butthere wasno-onetotalk toinAalbor g,orrather ,therew asno-onethere thatsharedmy interestinthe areaatthe time.Whatis more,theopen problemsthatremained weremajorchallenges. Somewereonly solvedmuch later(suchas thedecidabil- ityofbisimilarity forpushdown automata,aresult duetoSénizer gues,theproof latergreatlysimplified byColin)or remainopen(such asthedecidability ofweak bisimilarityforBP A).Eventually Idrifteddowntheslippery slopeandbecamein- terestedinother ,relatedtopics, andfocusedsomewhatmoredirectly onprogram properties.Thefollo w-upresultthat mostdirectlyrelatestomypaper withSøren andColinis theresultfrom 1994thatbisimilarity isalsothe onlyequiv alence thatisdecidable forBPP. Thisisa resultthatI amalsoquiteproudof. Itgrew outof discussionswithJa vierEsparzaand YoramHirshfeld, whenIwasbackin Edinburghforawhilein1993.Ironically ,therew asasubtle flawin theproof thatNaokiK obayashidisco veredmanyyearslater .Naokiandoneofhisstudent foundoutho wtorepair theproof,andthethreeof usco-authoredthe journalver - sionthatcame outmany yearslater .Thereason whyNaokibecameinterestedin BPPw asthathewastrying tousethe calculusasa behaviouraltypesystemfora pi-calculus.Asit happened,thisw asalsothe routethatI hadtak en. Thoseofmy laterresultsthat Iammost proudofha veto dowith thisarea:My workontypesystems forpsi-calculus(CONCUR 2011andlater) andasession typesystemfor boundedname useinthe pi-calculus(ActaInformatica 2019). Thelatterpaper alsohasa CONCURconnection;it grewout ofattendinga talk byRolandMe yeratCONCUR 2013andwonderingwhy theywere nottrying to characterizenotionsof name-boundednessinthe pi-calculusbymeans ofa type system.Itturned outthatMe yeretal. werenotf amiliarwithtype systemsatall. Ittookme anawful longtimeto workout atype-basedcharacterization(using sessiontypes),as youcanprobably tell. Asyoucan tell,Iha venot worked onbisimulationfor quitesometime.Not thatthereis anythingwrong withbisimulation,of course,butifonew antsresults thatareapplicable forautomatedreasoning aboutprogrambeha viour,decidability isimportant.One cantheneither choosetogo foraless expressiv emodel of computation(whichis whatresearchersin themodelchecking communitydo)or keepthemodelof computationandgo forsound,b utincompletecharacterizations ofprogramproperties (whichis whatresearchersinterested intypesystems and relatedstaticanalysis methodsdo). Iendedup inthelatter camp. Luca:Toyourmind,whatare themostinteresting orunexpected usesinthe liter- atureofthe notionsandtechniques youdev elopedin"Bisimulation Equivalence isDecidablefor allContext-Free Processes"? Hans:BPA,BPPandsimilarprocess calculicanbe usedasthe basisofbe- haviouraltypesystems,andI amthrilledthat myoldresearch interestsandmy currentresearchinterests aresodirectly related. In2018I discovered thatVasco VasconcelosandPeterThiemannhaddevised asessiontype systemfora functionalprogramminglanguage inwhichsession typeswerenot regulare xpressions(whichis essentiallywhattheyarein theorig- inalwork byHonda,KuboandV asconcelos)but BPAprocesses. VascoandPeter knewthattypeequiv alenceshould bedecidableb uttheywerenotsofamiliarwith ourresultsfrom theearly1990s. AtPOPL2019 Iattendeda talkbyAndreia Mor- dido,oneof Vasco's collaborators,andshe mentionedourpaperfromCONCUR

1992!Laterthat day,I endeduptalking toAndreiaandVascoabout mywork on

BPAfromallthoseyears ago.

Luca:Thelastforty yearshav eseena hugeamountof workonprocessalge- braandprocess calculi.Howe ver, theemphasison thatfieldofresearchwithin concurrencytheoryseemstoha vediminished over thelastfe wyears,eveninEu- rope.What advicewould yougiv etoayoungresearcherinterested inworking on processcalculitoday? Hans:Thatthey shouldstillworkonprocess calculi!Thereremains alotof interestingwork tobedone.Onereasonwh yresearchtopics driftinand outof focusissimply thatresearcherslose interest;itis hardlyev erthe casethata topic runsoutof interestingresearch questions.Anotherreason israthergrim: Basic researchisno wherenearas well-respectedasitusedto be.Ifyou lookatthe fundingschemesthat wehav etoday, onlythev erysuccessfulfewcangetfunding forbasicresearch; Iamnot amongthemand itdoesget frustratingquiteoften. Mosttopicsthese daysdealwith appliedresearch.There isnothingwrong with appliedresearchper se;someof whatIdo isonthat sideofthings, but therehas tobemore toresearchthan that. Luca:Whatarethe researchtopicsthat currentlyexcite youthemost? Hans:Ihav eslowlybecomemoreandmoreinterested inprogramminglanguage theory,andsomeofmy currentcollaboratorsha vetak enasimilar route,be gin- ninginconcurrenc ytheory, driftingintotheworld ofbehaviouraltypesystems andfinallyw antingto applyalloftheirskillsto actualprogramminglanguages. Rightnow MarioBravetti,Adrian Francalanza,AntónioRa varaandmyselfare involvedinworkingonabehavioural typesystemrelated totheMungotoolfora subsetofJa va.I amfortunatelyenoughtohavehadthree exceptionallycle ver and productiveMScstudentsinvolv edaswell, andthishas beenextremelyhelpful. Mungoisin manyw aysthew orkofSimonGayandOrnela DardhaatGlasgow University,andthey,too,began theircareersw orkingonprocesscalculi. Luca:Youhavea keeninterestinteachingatuniversity level. Isthereanything youthinkwe shoulddo betterincon veying thebeautyand usefulnessoftheo- reticalcomputerscience ingeneral,and concurrencytheory inparticular, toour studentstoday?Do youhav eany thoughtsyou' dliketosharewithuson howto combinedigitalresources andin-personmeetings intheteaching ofsubjectsin thetheory ofcomputing? Hans:PersonallyIthink thereisa tendencyto presentany academictopic- andinparticular topicsinthe mathematicalsciences,broadly speaking-in such away thatthedefinitionsandtheoremsappear asifthe yfellfrom theheav ens, fullyformed.As any researcherwillkno w,thatiscertainlynothowtheycame about.Inthis way, wefocusa lotonwhatImreLakatosandKarlPoppercalled the contextofjustification.Awell-honed presentationofa theorymayappear tobe beautiful,but inmyopiniontheactual beautyisthe onethatone experienceswhen onehas finallyunderstoodthe theoryandunderstands whythe definitionsand theoremturnedout thew aythey did-that is,onemustunderstandthecontextof discovery.Ithink problem-basedlearning(which issomethingthat wetalkabout alot atAalborg University andatRoskildeUniversity) isthekeyhere,becauseit putsthefocus onstudent-centered active learning. Ihav electuredalotovertheyears butsince 2013Iha vedrifted awayfrom traditionallecturesto wardsflipped teachinginwhichIusepre-recordedpodcasts (ofmyo wnmaking)that studentscanwatchwhenev erthey want;I thenuse the plenarysessionsfor activitiesthat focusonacti velearning.Iby farprefer having adialoguewith studentsthat focusesonthe problemsthatthe yencounterto the monologuestyle ofalecture. Allmyteaching activitiesare nowflipped andIam happytosaythatsome ofmycolleagues arenow thinkingalongsimilar lines.It isalw aysgoodtohavesomeonetotalk to.

InterviewwithNancyLynch andRobertoSegala

Luca:Youreceiveone ofthetwoCONCURTest-of-TimeA wardsfor theperiod

1992-1995foryour paper"Probabilisticsimulations forprobabilisticprocesses",

presentedatCONCUR 1994.Could youtellus brieflywhatspurred youtode- velopthe"simple"probabilistic automatonmodelintroduced inthatpaper and howtheideasunderlyingthat notioncame about? Roberto:Wewerestudyingrandomizeddistrib utedalgorithms andwenoted that severalalgorithmssu!eredfromsubtle errors.We werelookingfor aformal,quotesdbs_dbs22.pdfusesText_28

[PDF] Rapport du jury Filière PSI 2013 - concours Centrale-Supélec

[PDF] Modalités d 'hébergement ? CESAL pour l 'année - CentraleSupelec

[PDF] Rapport du jury Filière TSI 2012 - concours Centrale-Supélec

[PDF] NOTE energie thermique - EDF

[PDF] Étude préparatoire du projet de construction de la centrale

[PDF] Rapport TSI - 2004 - concours Centrale-Supélec

[PDF] Rapport du jury Filière TSI 2015 - concours Centrale-Supélec

[PDF] Centrale MP PC PSI TSI 2015: Corrigé - AL9AHIRA

[PDF] calendrier des séminaires spécialisés programmés ? abidjan - campc

[PDF] CODE DE VIE DU CENTRE CHAMPAGNAT

[PDF] Script d 'appelS téléphoniqueS pour l 'adjoint en marketing

[PDF] 105 centres d 'enfouissement technique ? travers le territoire - CDER

[PDF] 6-17-2-2015-2 centres_epreuves_ecrites,de_la_bel_2015 - ENS

[PDF] rapport annuel - ONCFma

[PDF] Centres d 'impulsion et inégale intégration - mediaeduscoleducationfr