min-max vs. min-sum vehicle routing: a worst-case analysis
DESCRIPTION
Min-Max vs. Min-Sum Vehicle Routing: A Worst-Case Analysis. L. Bertazzi , B. Golden, and X. Wang Route 2014 Denmark June 2014 . Introduction. In the min-sum VRP, the objective is to minimize the total cost incurred over all the routes - PowerPoint PPT PresentationTRANSCRIPT
![Page 1: Min-Max vs. Min-Sum Vehicle Routing: A Worst-Case Analysis](https://reader035.vdocuments.us/reader035/viewer/2022062502/568163d3550346895dd51c53/html5/thumbnails/1.jpg)
1
Min-Max vs. Min-Sum Vehicle Routing:
A Worst-Case AnalysisL. Bertazzi, B. Golden, and X. Wang
Route 2014Denmark
June 2014
![Page 2: Min-Max vs. Min-Sum Vehicle Routing: A Worst-Case Analysis](https://reader035.vdocuments.us/reader035/viewer/2022062502/568163d3550346895dd51c53/html5/thumbnails/2.jpg)
2
Introduction In the min-sum VRP, the objective is to minimize the
total cost incurred over all the routes
In the min-max VRP, the objective is to minimize the maximum cost incurred by any one of the routes
Suppose we have computer code that solves the min-sum VRP, how poorly can it do on the min-max VRP?
Suppose we have computer code that solves the min-max VRP, how poorly can it do on the min-sum VRP?
![Page 3: Min-Max vs. Min-Sum Vehicle Routing: A Worst-Case Analysis](https://reader035.vdocuments.us/reader035/viewer/2022062502/568163d3550346895dd51c53/html5/thumbnails/3.jpg)
3
Introduction Applications of the min-max objective
Disaster relief efforts Serve all victims as soon as possible
Computer networks Minimize maximum latency between a server and a
client
Workload balance Balance amount of work among drivers and/or across
a time horizon
![Page 4: Min-Max vs. Min-Sum Vehicle Routing: A Worst-Case Analysis](https://reader035.vdocuments.us/reader035/viewer/2022062502/568163d3550346895dd51c53/html5/thumbnails/4.jpg)
4
An Instance of the VRP The min-max solution The min-sum solution
24.553
47.10526
83.6224
24.10236
Max load = 2
Max # vehicles = 2
total cost total costmin-max cost min-max cost
![Page 5: Min-Max vs. Min-Sum Vehicle Routing: A Worst-Case Analysis](https://reader035.vdocuments.us/reader035/viewer/2022062502/568163d3550346895dd51c53/html5/thumbnails/5.jpg)
5
Motivation behind our Worst-Case Study Observation: The min-max solution has a slightly
higher (2.2%) total cost, but it has a much smaller (23.3%) min-max costAlso, the routes are better balanced
Is this always the case? What is the worst-case ratio of the cost of the
longest route in the min-sum VRP to the cost of the longest route in the min-max VRP?
What is the worst-case ratio of the total cost of the min-max VRP to the total cost of the min-sum VRP?
![Page 6: Min-Max vs. Min-Sum Vehicle Routing: A Worst-Case Analysis](https://reader035.vdocuments.us/reader035/viewer/2022062502/568163d3550346895dd51c53/html5/thumbnails/6.jpg)
6
Variants of the VRP Studied Capacitated VRP with infinitely many vehicles
(CVRP_INF)
Capacitated VRP with a finite number of vehicles (CVRP_k)
Multiple TSP (MTSP_k)
Service time VRP with a finite number of vehicles (SVRP_k)
![Page 7: Min-Max vs. Min-Sum Vehicle Routing: A Worst-Case Analysis](https://reader035.vdocuments.us/reader035/viewer/2022062502/568163d3550346895dd51c53/html5/thumbnails/7.jpg)
7
CVRP_INF Capacitated VRP with an infinite number of vehicles
: the cost of the longest route of the optimal min-max solution
: the cost of the longest route of the optimal min-sum solution
: the total cost of the optimal min-max solution
: the total cost of the optimal min-sum solution
The superscript denotes the variant
MSr
MMr
MSz
MMz
![Page 8: Min-Max vs. Min-Sum Vehicle Routing: A Worst-Case Analysis](https://reader035.vdocuments.us/reader035/viewer/2022062502/568163d3550346895dd51c53/html5/thumbnails/8.jpg)
8
A Preview of Things to Come For each variant, we present worst-case
bounds
In addition, we show instances that demonstrate that the worst-case bounds are tight
![Page 9: Min-Max vs. Min-Sum Vehicle Routing: A Worst-Case Analysis](https://reader035.vdocuments.us/reader035/viewer/2022062502/568163d3550346895dd51c53/html5/thumbnails/9.jpg)
9
CVRP_INF
MMMS rr
1122 nrMS2
MMr
# customers: n =1+1/ε
capacity = 1/ε
![Page 10: Min-Max vs. Min-Sum Vehicle Routing: A Worst-Case Analysis](https://reader035.vdocuments.us/reader035/viewer/2022062502/568163d3550346895dd51c53/html5/thumbnails/10.jpg)
10
CVRP_INF
MSMM zz
524 nzMS
# customers: n =1+1/ε
capacity = 1/ε
112 MMz
![Page 11: Min-Max vs. Min-Sum Vehicle Routing: A Worst-Case Analysis](https://reader035.vdocuments.us/reader035/viewer/2022062502/568163d3550346895dd51c53/html5/thumbnails/11.jpg)
11
CVRP_k Capacitated VRP with at most k vehicles
available
kMM
kMM
kMS
kMS krzzr
kMS
kMS
kMM
kMM kzkrkrz
krr kMM
kMS
kzz kMS
kMM
![Page 12: Min-Max vs. Min-Sum Vehicle Routing: A Worst-Case Analysis](https://reader035.vdocuments.us/reader035/viewer/2022062502/568163d3550346895dd51c53/html5/thumbnails/12.jpg)
12
CVRP_k
krr kMM
kMS
12 kkr kMS 22kMMr
![Page 13: Min-Max vs. Min-Sum Vehicle Routing: A Worst-Case Analysis](https://reader035.vdocuments.us/reader035/viewer/2022062502/568163d3550346895dd51c53/html5/thumbnails/13.jpg)
13
CVRP_k
kzz kMS
kMM
12 kz kMS kz kMM 2
![Page 14: Min-Max vs. Min-Sum Vehicle Routing: A Worst-Case Analysis](https://reader035.vdocuments.us/reader035/viewer/2022062502/568163d3550346895dd51c53/html5/thumbnails/14.jpg)
14
MTSP_k Multiple TSP with k vehicles
The customers just have to be visitedExactly k routes have to be defined
MMM
MMM
MMS
MMS krzzr
MMS
MMS
MMM
MMM kzkrkrz
krr MMM
MMS
kzz MMS
MMM
![Page 15: Min-Max vs. Min-Sum Vehicle Routing: A Worst-Case Analysis](https://reader035.vdocuments.us/reader035/viewer/2022062502/568163d3550346895dd51c53/html5/thumbnails/15.jpg)
15
MTSP_k
krr MMM
MMS
2MMMr
12212
kkkr MMS
![Page 16: Min-Max vs. Min-Sum Vehicle Routing: A Worst-Case Analysis](https://reader035.vdocuments.us/reader035/viewer/2022062502/568163d3550346895dd51c53/html5/thumbnails/16.jpg)
16
MTSP_k
kzz MMS
MMM
kz MMM 2 132
1122
k
kkz MMS
![Page 17: Min-Max vs. Min-Sum Vehicle Routing: A Worst-Case Analysis](https://reader035.vdocuments.us/reader035/viewer/2022062502/568163d3550346895dd51c53/html5/thumbnails/17.jpg)
17
SVRP_k Service time VRP with at most k vehicles
Customer demands are given in terms of service times
Cost of a route = travel time + service time Routing of the min-sum solution is not
affected by service times Routing of the min-max solution may be
affected by service times
SMM
SMM
SMS
SMS krzzr krr S
MMSMS
![Page 18: Min-Max vs. Min-Sum Vehicle Routing: A Worst-Case Analysis](https://reader035.vdocuments.us/reader035/viewer/2022062502/568163d3550346895dd51c53/html5/thumbnails/18.jpg)
18
SVRP_kMin-max solution without service times
Min-max solution with service times
![Page 19: Min-Max vs. Min-Sum Vehicle Routing: A Worst-Case Analysis](https://reader035.vdocuments.us/reader035/viewer/2022062502/568163d3550346895dd51c53/html5/thumbnails/19.jpg)
19
SVRP_k
krr SMM
SMS
122 kktkr SMS 222 tr SMM
![Page 20: Min-Max vs. Min-Sum Vehicle Routing: A Worst-Case Analysis](https://reader035.vdocuments.us/reader035/viewer/2022062502/568163d3550346895dd51c53/html5/thumbnails/20.jpg)
20
SVRP _k The bound
is still valid, but no longer tight
We prove the tight bound , where T = total
service time in our paper
SMS
SMS
SMM
SMM kzkrkrz kzz S
MSSMM
Tkkzz SMS
SMM 1
![Page 21: Min-Max vs. Min-Sum Vehicle Routing: A Worst-Case Analysis](https://reader035.vdocuments.us/reader035/viewer/2022062502/568163d3550346895dd51c53/html5/thumbnails/21.jpg)
21
SVRP_k
Tkkzz SMS
SMM 1
Tkz SMM 2 Tkz SMS 12
![Page 22: Min-Max vs. Min-Sum Vehicle Routing: A Worst-Case Analysis](https://reader035.vdocuments.us/reader035/viewer/2022062502/568163d3550346895dd51c53/html5/thumbnails/22.jpg)
22
A SummaryRatio of the cost of the longest route Ratio of the total cost
CVRP_INF
CVRP_k
MTSP_k
SVRP_k Tkkzz SMS
SMM 1 krr S
MMSMS
MMMS rr
MSMM zz kzz k
MMkMS kzz k
MSkMM
krr MMM
MMS kzz M
MSMMM
![Page 23: Min-Max vs. Min-Sum Vehicle Routing: A Worst-Case Analysis](https://reader035.vdocuments.us/reader035/viewer/2022062502/568163d3550346895dd51c53/html5/thumbnails/23.jpg)
23
Conclusions If your true objective is min-max, don’t use
the min-sum solution
If your true objective is min-sum, don’t use the min-max solution