Using Lagrangean relaxation to minimize the weighted number of late jobs on a single machine
read more
Citations
Mixed integer programming formulations for single machine scheduling problems
Evaluation of mixed integer programming formulations for non-preemptive parallel machine scheduling problems
Genetic algorithms to minimize the weighted number of late jobs on a single machine
Flexibility and Robustness in Scheduling
A survey of scheduling with parallel batch (p-batch) processing
References
Complexity of machine scheduling problems
An n Job, One Machine Sequencing Algorithm for Minimizing the Number of Late Jobs
A Solvable Case of the One-Machine Scheduling Problem with Ready and Due Times
Algorithms for scheduling a single machine to minimize the weighted number of late jobs
Minimizing late jobs in the general one machine scheduling problem
Related Papers (5)
Frequently Asked Questions (9)
Q2. What is the eciency of Lagrangean relaxation approaches?
One condition that is often associated to the e ciency of Lagrangean relaxation approaches is to relax as few constraints as possible, in order to obtain good bounds when solving the relaxed problem.
Q3. Why is it important to remove as many positions from the master sequence?
Because the size of the model is directly linked to the length of the master sequence, it is interesting to remove as many positions as possible from .
Q4. What is the dominance rule for the master sequence?
In the master sequence, if Conditions (1) ri < rj , (2) ri+pi rj+pj, (3) ri+pi+pj > dj , (4) rj + pj + pi > di, (5) di pi dj pj, and (5) wj wi hold, then Jj dominates Ji and all positions of job Ji can be removed from the master sequence.
Q5. What is the constraint for the job at position k?
By Constraint (2) the authors ensure that, if the job at the kth position in the master sequence is set on time (uk = 1), then the job at position k + 1 cannot start before the completion of the job at position k.
Q6. How many times faster are the bounds obtained when n is large?
When n is large, the bounds are obtained faster (184.66 seconds on average vs 359.49 for n = 140), and the average gap between the two bounds is also reduced.
Q7. What is the simplest way to compute the lower bound?
If coef < coef( (k)), then coef( (k)) = coef and pos( (k)) = k.Step 4 - Compute the lower bound:LB = n+ nX i=1 2 6664 PX k=1; (k)=i( r k+1 r k ) 0( rk+1r k)ri + ( r k+1pi 1)uk+ PXk=1; (k)=i ( r k+1 r k )<0( rk+1r k)(di piuk) + ( r kpi 1)uk 3 7775Moreover, the Kise et al.'s algorithm [6] can be used to compute the upper bound associated to the current value of the multipliers r in Step 6.
Q8. How can the Moore's algorithm be applied to a single machine?
Lawler [7] showed that the Moore's algorithm ([9]) could be applied when processing times and weights are aggeeable, i.e., pi < pj ) wi wj 8(Ji; Jj).
Q9. What is the constraint that is used to constrain the jobs sequenced before k?
Dk is chosen big enough to not constrain the jobs sequenced before k, for instanceDk = max r=1;::;k 1 d (r)>d (k) (d (r) d (k)) (= max r=1;::;k 1 (0; d (r) d (k))):