Page - 54 - in Algorithms for Scheduling Problems
Image of the Page - 54 -
Text of the Page - 54 -
Algorithms 2018,11, 80
Phase2: If thefirstconditioninTheorem2isnotsatisfied, i.e., anIA(b2)at thefirstpassofPhase1
occurs, thenPhase2 is invoked. Letybe theearliestarisenType (b)y-job that couldnothavebeen
included inscheduleS(B,y). Ingeneral, job jmaybepushedbyoneormoreotherType (b)y-jobs.
Since the latenessof joby surpassesL(δ), a completeschedulerespecting this thresholdvaluecannot
becreatedunless the latenessof joby isdecreased. Thenextobservationfollows:
Observation6. SupposeanIA(b2)with(aType(b)y-job)yat thefirstpassofPhase1 inbinBarises. Then, the
resultantmaximumjob lateness inbinByieldedby joby cannotbedecreasedunless joby and/or otherType
(b)y-jobs currentlypushing joby inbinBare rescheduled to someearlier bin(s).
Accordingto theaboveobservation, inPhase2,either joby is toberescheduledtoanearlierbin
orasubsetof theaboveType(b)y-jobs (withanappropriate totalprocessing time) is tobe formedand
all jobs fromthatsubsetare toberescheduledtosomeearlierbin(s). Inparticular, if the latenessof job
y surpassesL(δ)byamountλ, then the totalprocessing timeof jobs insuchasubset is tobeat leastλ.
Bytheconstruction,nonewjobcanfeasiblybe introducedintoanyalreadyscheduledbinBunless
some jobs (withanappropriate total length) fromthatbinarerescheduledbehindall theaboveType
(b)y-jobs. Sucha job,whichwecallasubstitution jobfor joby, shouldclearlybeanemerging jobfor
joby, i.e.,ds> dy (asotherwise,oncerescheduled, its latenesswill surpassL(δ)).
Theorem3. If for an IA(b2)with a (Type (b) y-job) y, there exists no substitution job for job y, then there
existsno feasible schedule respecting thresholdvalueL(δ). Hence, there existsnoPareto-optimal solutionwith
maximumjob lateness L(δ).
Proof. DuetoObservation6, jobyorsomeotherType(b)y-job(s)pushing jobymustberescheduled
toabinprecedingbinB. By thewayweconstructourbins,nonewjobcan feasiblybe introduced
intoanybinprecedingbinBunlesssome(sufficiently long) job s fromthatbin is rescheduledbehind
all theaboveType (b)y-jobs. ds< dymustholdasotherwise job sor some joboriginally included
before jobywill surpass thecurrent threshold; i.e., s isasubstitution job. It followsthat if thereexists
nosubstitution job, the latenessofat leastone jobfrombinBor fromitsprecedingbinwill surpass
thresholdvalueL(δ), andhence, theremayexistno feasible schedule respectingL(δ). The second
claimimmediately follows.
Due to the above theorem, if these exists no substitution job, the call from the binary search
procedure for thecurrentδhaltswith the failureoutcome,andtheprocedureproceedswith thenext
(larger) trialvalue forδ; if thereremainsnosuchδ, thewholeschemehalts.
Thecasewhenthereexistsasubstitution jobfor joby isdiscussednow.We observe that there
mayexistaPareto-optimalsolution. Tovalidatewhether it exists,wefirstneedtoverify if thereexists
a feasible solution respecting threshold L(δ) (given that the call to thebinpackingprocedurewas
performedfromthe iteration in thebinarysearchprocedurewith that thresholdvalue). This task is
easilyverifiedif, for instance, there isasinglesubstitution job s. Indeed,wemayconstructanauxiliary
ED-schedule inwhichwe activate that substitution job for job y (similarly as before,we increase
artificially the release timeof job s to that of job y, so that once theEDheuristics isnewlyapplied
to thatmodifiedprobleminstance, thesubstitution job swillbe forcedtoberescheduledbehind job
y). In the resultantED-schedulewith theactivated job s, if the latenessofnoneof the rescheduled
y-jobs isnogreater thanL(δ),Phase1canberesumedforbinB. Otherwise, theremayexitnofeasible
solutionrespectingthresholdL(δ) (hence,noPareto-optimalsolutionwith themaximumjob lateness
notexceedingL(δ)mayexist), andtheschemecanproceedwith thenext trialδ.
In practice, with a known upper limit on the total number of substitution jobs, the above
activationprocedurecanbegeneralizedtoanotherprocedure thatenumeratesall thepossiblesubsets
of substitution jobs inpolynomial time. Ingeneral, if thenumberofsubstitution jobs isboundedonly
byn, thenumberofpossiblesubsetsof substitution jobswith totalprocessing timeλormoremightbe
54
back to the
book Algorithms for Scheduling Problems"
Algorithms for Scheduling Problems
- Title
- Algorithms for Scheduling Problems
- Authors
- Frank Werner
- Larysa Burtseva
- Yuri Sotskov
- Editor
- MDPI
- Location
- Basel
- Date
- 2018
- Language
- English
- License
- CC BY 4.0
- ISBN
- 978-3-03897-120-7
- Size
- 17.0 x 24.4 cm
- Pages
- 212
- Keywords
- Scheduling Problems in Logistics, Transport, Timetabling, Sports, Healthcare, Engineering, Energy Management
- Categories
- Informatik
- Technik