You are on page 1of 15

31-MM34 Metaheuristics

Video 1: Operations Research and Optimization Problems

Prof. Dr. Kevin Tierney

kevin.tierney@uni-bielefeld.de
Professor for Decision and Operation Technologies
Bielefeld University

Last updated: Summer 2022

This work is licensed under a Creative Commons Attribution-NonCommercial-ShareAlike 4.0 International license.
Outline

Outline

I What is OR?
I Types of optimization problems
I Formalizing optimization problems
I Why OR problems are hard (P vs. NP)
I Methods for solving optimization problems

31-MM34 MH: Video 1: Operations Research and Optimization Problems 2/13


OR Background

What is operations research?

Operations Research (O.R.) . . . is a discipline that deals with the application of


advanced analytical methods to help make better decisions. – INFORMS

Neighboring disciplines that overlap with OR:


I Management science
I Operations management
I Business analytics
More information: (From the OR Society (UK))
I What is OR: https://youtu.be/0oMVVx81kCs
I History of OR: https://youtu.be/ILWbaWrjgU4

31-MM34 MH: Video 1: Operations Research and Optimization Problems 3/13


OR Background

The OR process

31-MM34 MH: Video 1: Operations Research and Optimization Problems 4/13


OR Background

Problem classes

I Optimization
“What is the best possible solution to this problem?”
I Satisfaction
“What is a solution to this problem?”
Examples:
I Can these packages be delivered with my fleet of vehicles? (Satisfaction)
I ⇒ What is the lowest cost for delivering these packages? (Optimization)
I Can this wood be cut such that all customers’ orders are fulfilled? (Satisfaction)
I ⇒ Can I fulfill customer orders while minimizing waste?
I Can I store all of these containers on the train cars? (Satisfaction)
I ⇒ Can I minimize the overall storage time? (Optimization)

31-MM34 MH: Video 1: Operations Research and Optimization Problems 5/13


OR Background

Problem Formalization (Linear Model)

We can formulate a (linear) optimization problem as:

min c| x (1)
subject to Ax ≤ b (2)
x≥0 (3)

Where:
I x is a set of decision variables
I c is a set of costs for each decision
I A is a set of coefficients.
I b is called the right hand side (RHS).

31-MM34 MH: Video 1: Operations Research and Optimization Problems 6/13


OR Background

Problem Formalization (General Model)

Given
I A set of decision variables {x1 , . . . , xn }
I A set of domains {D1 , . . . , Dn }, xi ∈ Di
I A set of constraints {C1 , . . . , Cm }, Cj : D1 × · · · × Dn → B
I An objective function f : D1 × · · · × Dn → R

31-MM34 MH: Video 1: Operations Research and Optimization Problems 7/13


OR Background

Motivating Heuristics

In this course: Mainly NP-Hard (or harder) problems.

NP-Hard (and NP-complete) problems exhibit larger than polynomial growth in


runtime as the input size increases.

31-MM34 MH: Video 1: Operations Research and Optimization Problems 8/13


OR Background

Example: Traveling Salesperson Problem (TSP)


Given:
I n cities
I Distance matrix dij (distance between i and j)
Goal: Find a minimal distance path that visits every city exactly once.

31-MM34 MH: Video 1: Operations Research and Optimization Problems 9/13


OR Background

The growth of P and NP problems


Shortest path O(n2 ) TSP O(n!)
n |S| |S|
5 25 120
10 100 3628800
15 225 1.3×1012
20 400 2.4×1018
25 625 1.5×1025
30 900 2.6×1032

31-MM34 MH: Video 1: Operations Research and Optimization Problems 10/13


OR Background

The growth of P and NP problems

Shortest path O(n2 ) TSP O(n!)


n |S| |S|
5 25 120
10 100 3628800
15 225 1.3×1012
20 400 2.4×1018
25 625 1.5×1025
30 900 2.6×1032

Given a computer that can process 200,000 solutions per second, you would need over
42000000000000000000 years to solve an n = 30 traveling salesman problem!

31-MM34 MH: Video 1: Operations Research and Optimization Problems 10/13


OR Background

The growth of P and NP problems

Shortest path O(n2 ) TSP O(n!)


n |S| |S|
5 25 120
10 100 3628800
15 225 1.3×1012
20 400 2.4×1018
25 625 1.5×1025
30 900 2.6×1032

Despite the size of these problems, we have techniques to solve them!

31-MM34 MH: Video 1: Operations Research and Optimization Problems 10/13


OR Background

P vs. NP

NP-complete problems are the focus of one of computer science’s biggest questions:

Are hard problems really hard?


For more information, see these videos:

Big O notation: https://youtu.be/MyeV2_tGqvw


P vs. NP: https://youtu.be/OY41QYPI8cw

31-MM34 MH: Video 1: Operations Research and Optimization Problems 11/13


Methods for solving problems

Methods for solving optimization problems

When solving optimization problems, you must make a choice:


Do you want a fast answer with no guarantee of optimality?

or

do you want a optimal answer with a mathematical proof that may require
significant time to obtain?

31-MM34 MH: Video 1: Operations Research and Optimization Problems 12/13


Summary

Summary

Summarizing this video:


1. We learned about what OR is
2. We examined types of optimization problems and formalized them
3. We found out that OR problems are computationally difficult to solve; but that
with the right tricks, it’s possible.

31-MM34 MH: Video 1: Operations Research and Optimization Problems 13/13

You might also like