You are on page 1of 2

Homework #1 solutions / IEOR4405 1

1. Problem 2.3
(a) The optimal schedule has ΣUj = 2. The following schedules are optimal {1, 3, 2, 4}, {1, 3, 4, 2},
{3, 4, 1, 2}, {3, 4, 2, 1}, {2, 3, 1, 4}, {2, 3, 4, 1}, {2, 4, 3, 1}, {2, 4, 1, 3} Direct computation reveals that
all the above schedules have ΣUj = 2.
(b) The optimal schedule is discussed in Algorithm 3.3.1 in chapter 3. It will be discussed in the
following week during lecture hours.

2. Problem 2.5
The total processing time of 11 jobs is 75. Since there are 5 machines working in parallel. We

need at least 15 time units to process all the jobs. That is Cmax ≥ 15. Now, consider the
following schedule, The schedule has a completion time of 15. This is exactly the lower bound
that we computed. Since a feasible solution equals the lower bound, it is the optimal solution.

3. If we have a preemptive policy with all the jobs having the same release date. Then it is no dif-
ferent from the problem without a preemptive policy∗ . Hence, the optimal way would be to sort
all the jobs in the ascending order of the job lengths and then process in that order. This is quite
intuitive. A mathematical proof based on exchange arguments was illustrated in last lecture hour

If we have a preemptive policy with specified release dates for the jobs. Then, for 1||ΣCj ,
Shortest Remaining Processing time (SRPT) is optimal. It is very similar to the aforementioned
policy. Here, at each time instant the jobs are arranged in ascending order according to the
remaining service time and the first element is processed. Note: The completed part of the job
is not lost due to preemption.

Proof(*): Suppose the jobs are not arranged in the ascending order of the processing times.
Then 3 at least 1 pair of jobs (j, k) such that pj > pk and job j is placed ahead of k in the queue.
In this configuration,
Let C = Sum of completion times of all the jobs except jobs j&k
The objective value ΣCj = C + (t + pj ) + (t + pj + pk )
where t is the completion of all the jobs before job j
This configuration is clearly not optimal because by processing k before j, the objective value is
ΣCj = C + (t + pk ) + (t + pk + pj )
which is smaller than the earlier configuration since we have pj > pk
Note: C and t does not change

This study source was downloaded by 100000831952622 from CourseHero.com on 02-06-2022 02:05:57 GMT -06:00

https://www.coursehero.com/file/7111094/sol1/
Homework #1 solutions / IEOR4405 2

Hence, by induction over all the pair of jobs we can conclude that the jobs have to be placed
in the ascending order of the processing times. Thus, we don’t need to preempt the jobs after
structuring the initial queue. If the jobs have different release dates, follow similar arguments to
show that, SRPT (as defined earlier) is optimal.

This study source was downloaded by 100000831952622 from CourseHero.com on 02-06-2022 02:05:57 GMT -06:00

https://www.coursehero.com/file/7111094/sol1/
Powered by TCPDF (www.tcpdf.org)

You might also like