12
15] Invited Review An introduction to timetabling D. de WERR/I Swiss Federal h~stitute of Technology. Department of Matllematics, CH-IOIS, Lausanne, Switzerland Abstract. A huge variety of timetabling models have been described in the OR literature; they range from the weekly timetable of a school to the scheduling of courses or exams in a university. Graphs and networks have proven to be useful in the formulation and solution of such problems. Various models wilt be described with an e.nphasis on graph theoretical models. Keywords: Timetabling, education networks, graphs, heuristic,,: 1, Intro~tuetion During the last twenty years many contribu- tions related to timetabling have appeared and it will probaN/ condnue with the same ra'e for years. One reason for this may be the huge variety of problems which are included in the field of timetabling; another reason is the fact that educa- tional methods are changing, so that the models have to be modified. Fina!ly. ~,._,mputiag facilities are now available in most schools and, zs ,; conse quence, the approach to timembling has to take this phenomenon into account. This means in par- ticular that interactive methods are now becoming more important and, furthermore, they have to be adapted to microcom[~'uters C.,,uaily one considers dne ;.ic,eutbfiag proce.as, t~t:~ consisl.~ c,f 2 distinct phases [71: a) First. the curricula are defined for each class Recciveg Aug=5: ~.')S ;t N.;,r',h- H,',~!~,~ cl Eu:ope~ Journ?.! of Operat:onal R-zsea~c0 19 ..i9,35) !51-162 or for each group of ~.tudents and one ha.; to assign the various resources (in manpower or ,n equipment) to the classes. b) Second, when an agreement has been ,e- ached concerning these assignments of resour s, then one tries to see whether a workab!e deta -d timetable can be worked out which is compati le with all the previously defined requirements. (3eneral/y the role of computers is to handle the second phase and we shall concentrate on this type of problem. However, one shouid mention t, at some attention has also been given to the first phase, both from a theoretical and from a comou- tational point of view [15,20]. In this survey we cannot describe all special types of timetables which have been mentioned up to now. W,= shall refer the reader to the annotated bibliograp!~y [29] which covers the contributions wbhch appeared before 1980 and to general surve,js [5,7,18]. Our pui~ose will be t,.~ introduce the reader tc some basic problem:, which occur ir~ rues,', o. ~ the real cases; we shall first describe the cla~-tc~c);c. ~ problem and its vanatic.ns. "then we ,,-All d~,,cus; the course scheduling rr.odel toge~',er with .,,o~e closely related prc:blems. For both typ.~s of timeta- bling problems, "re shall mainly corc~ntrate on models based on graph; and netwo:lcs. As ob- served by Mulvey {24], such models are iateresting. since network problems are efficiently solved even when their size is large; although, generally, reai timetab!ing problems cannot be formuiated with ware network mo:!els, in the man-maczine iter~- tVe soiufion , -,,,l. ,.- ,,r~ netwc,,k '- ~'' ~ [ " ~ I c.ccur a! '.?rio'.:s 5~ag..?s. O~t~ .=v,:' .,hertZ'ore tc:.::e ::d,,a~_-'.tag,~ :~I th~ fact :!'.z:~ :.[..ey arc r,.!a iveiy easy t:) handle [o devise fast heuristic m~,thrads for Another argument lot j,_,.~tifyiag eer ~nt~.re~:t i~: =-,od~ls .%~ :i'npe~, timetab~.i:~g prob2cms i~ tne ~c.'. ;an: heuristic :nethcds ffcr '.,-anJ!i~ :ea~ "..irc.e'z.-. i-.lino r~,roble,'rc~) are often d'~ri,c ,~ a~d adavtec2 0377-22!7/85/$3.30 'b 1985, E!sz'Aer Sci,.'nze Pub2:,he:~ p.V. (Nor~.h-[-ioiiand)

(Werra, 1985)

Embed Size (px)

DESCRIPTION

dsadfsgdf

Citation preview

15] InvitedReview Ani ntroducti ontoti metabl i ng D. d e WE RR/ ISwissFederalh~stituteof Technology.Department ofMatllematics,CH-IOIS,Lausanne,Switzerland Abstract. Ahugevari et yoft i met abl i ngmodel s havebeendescri bedintheORliterature;t hey rangefromtheweekl yt i met abl eof aschool t othe schedulingof cour sesor examsinauniversity.Gr aphs andnet wor kshaveproventobeuseful in theformul at i onandsolutionof suchprobl ems.Vari ousmodel swiltbedescri bedwithane. nphasi s ongraphtheoreticalmodels.Keywords: Ti met abl i ng, educat i onnet wor ks,graphs, heuristic,,: 1,Intro~tuetion Duri ngthelastt went yyearsmanycont ri bu- tionsrelatedtot i met abl i nghaveappear edandit willp r o b a N/ c ondnue withthesamer a' efor years.Onereasonforthismaybethehugevariety of pr obl emswhi chareincludedint hefieldof timetabling;anot her reasonisthefactthateduca-tionalmet hodsarechanging,sothatthemodel s havetobemodi fi ed. Fina!ly.~,._,mputiagfacilities arenowavai l abl einmost schoolsand, zs,;cons e quence, theappr oachtot i membl i nghast otake thisphenomenoni nt oaccount. Thismeansinpar- ticularthati nt eract i vemet hodsarenowbecomi ng mor ei mport ant and, furthermore, t heyhavetobe adapt edtomicrocom[~'uters C.,,uailyoneconsi dersdne;.ic,eut bfi agproce.as, t~t:~consisl.~c,f2distinctphases[71: a)First.thecurri cul aaredefi nedforeachclass ReccivegAug=5:~.')S;t N.;,r',h- H,',~!~,~cl Eu : o p e ~ Journ?.!ofOperat : onal R-zsea~c0 19..i9,35)!51-162 or foreachgroupof ~.tudentsandoneha.;to assignthevari ousresources(inmanpower or ,n equi pment ) totheclasses. b) Second, whenanagr eement hasbeen, e- achedconcerni ngtheseassi gnment sof resours, t henonetriestoseewhet her aworkab!edet a-d t i met abl ecanbewor kedout whichiscompat i le wi t hallt heprevi ousl ydef i nedrequirements.(3eneral / ytheroleof comput er s ist ohandl ethe secondphaseandweshallconcent r at eonthist ype of pr obl em. However, oneshoui dment i ont,at someat t ent i onhasal sobeengivent ot hefirst phase, bot hfromat heoret i cal andfromacomou-t at i onal poi nt ofview[15,20]. Inthissurveywecannot descri beallspecial t ypesof t i met abl eswhichhavebeenment i onedup tonow. W,=shallrefert hereadertotheannot at ed bibliograp!~y[29]whi chcoversthecont r i but i ons wbhchappear edbef or e1980andtogeneralsurve,js [5,7,18]. Our pui ~osewillbet,.~ i nt roducethereadertc somebasi cproblem:,whichoccurir~rues,',o.~ the realcases;weshallfirstdescri bethecl a~- t c~c) ; c.~ pr obl emanditsvanatic.ns."thenwe,,-All d~,,cus; thecourseschedulingrr.odeltoge~',erwith.,,o~e closelyrelatedprc:blems.For bot htyp.~softimeta- bl i ngprobl ems, "reshallmai nl ycor c~nt r at eon model sbasedongr aph; andnetwo:lcs.Asob-servedbyMul vey{24], suchmodel sareiateresting.sincenet workpr obl emsareefficientlysolvedeven whentheirsizeislarge;al t hough, generally,reai t i met ab!i ngprobl emscannot beformui at edwith warenet wor kmo:!els,inthema n- ma c z i ne iter~- t Vesoi uf i on,-,,,l., . - ,,r~netwc,,k'-~''~["~I c.ccura!'.?rio'.:s5~ag..?s.O~t~.=v,:'.,hertZ'oretc:.::e ::d,,a~_-'.tag,~:~Ith~fact:!'.z:~ :.[..ey arcr,.!aiveiyeasy t:)handl e[odevisefastheuristicm~,thradsfor Anot her argument l ot j,_,.~tifyiage e r ~nt~.re~:t i~: =-,od~ls.%~:i'npe~,timetab~.i:~gprob2cmsi~tne~c.'. ; an: heuristic: net hcdsf f cr ' .,-anJ!i~:ea~"..irc.e'z.-. i-.linor~,roble,'rc~) areof t end'~ri,c ,~a~dadavtec2 0377-22!7/ 85/ $3. 30' b1985,E!sz'AerSci,.'nzePub2:,he:~p.V.(Nor~.h-[-ioiiand) 152D.deWerm /Ani n t ~ i o n to ametabling fromexactn~t hc, l s dew:lopedfort hesimplecases [30]. Inthistextweshallassumethatt hereaderhas someveryel ement aryknowledge of graph-theore' ;- icalconcepts. Alltermsrelatedt ographsand net workswhicharenot definedherecanbefot md in[t,111. 2.The-dass-teacher model . . ~ . As i mpl e mo a e lWefirstdescri bethebasiccl ass- t eacher model wi t hout includingalIconstraintswhichareusually presentinrealca.,es. Aclasswi l l camsistof asetof s~udentswho followexactlythesameprogramLet C={c 1. . . . . c,~}beasetofcl os es andT,--(t~ . . . . . G} asetofteachers.Wearegi,,enanmxnrequire- ment matrixR=(r~j)where%isthenumber oflecturesinvolvingclassqan, tteacherty. Weshallassumetkatalll ~t ur es have*.he same durat i on(say oneperiod).Gi wmasetofpperiods, theprobel m~stoassigneach!ecturet osome peric~'linsuchawaythatnoteacher(resp.no class)isinvolvedinmorethanonelectureata timeMoreprecisely,ifwedefinex,~ ktobe1if cl l ssc,andteachertjmeetatperi odkand0 olhelwise, wehavetosolve ~aroblemCTI :,.~V" x,i~=r,j( i = 1 . . . . . n . ; j = l , . . , n ) , ( 1):1 ~_,x, j ~~1( i - - 1 . . . . . m ; k =1 . . . . .p ) , (2) i ~ 1 E x , : < l ( : = 1 . . . . .n ; k = ! . . . . .p) , (3) .x~:~ =0or1(i =, 1 . . . . . m ; j =1 . . . . , n;(4) k = 1. . . ., p) ,Withthisformul at i onwemayassociateabi par- titemultigraphG=( C, T . . ~ ) : It.nodesarethe classesandtheteachers;nodec,andno.tetjare linkedbvr,,para!!,:!edges.Ifeachperi adcorre- spot~.ds :oacolor. ~heprobl emconsistsinfinding anassignmentcf or~e amongpcol orst oe,:tch edge ofGinsuchawa-,: thatnot woadjaceF,tedges havethesamecolor;sox,i ~. wiltbe1ifSO'heedge. [ c . t : ] getscolc.rk. Proposi t i on2.1.The r e e x : s t s asolutiont oCT 1 i f fm ~.,r, j