ie 501: optimization models
TRANSCRIPT
1
IE 501: Optimization Models
Autumn 2009
Vishnu Narayanan
Industrial Engineering and Operations ResearchIndian Institute of Technology Bombay
Lecture 1, 23rd July, 2009
Vishnu Narayanan IE 501: Optimization Models
2
General InfoSlot 4: M 1135–1230, Tu 0830–0925, Th 0930–1025
At your service: Vishnu Narayanan
Email: [email protected]: x-7878Office: F20, Old CSEPlease contact me in this order!
Teaching Assistant: Virendra Patidar([email protected])
Course webpage:http://www.ieor.iitb.ac.in/∼vishnu/teaching/ie501/Check this page and moodle regularly.
Vishnu Narayanan IE 501: Optimization Models
2
General InfoSlot 4: M 1135–1230, Tu 0830–0925, Th 0930–1025
At your service: Vishnu Narayanan
Email: [email protected]: x-7878Office: F20, Old CSEPlease contact me in this order!
Teaching Assistant: Virendra Patidar([email protected])
Course webpage:http://www.ieor.iitb.ac.in/∼vishnu/teaching/ie501/Check this page and moodle regularly.
Vishnu Narayanan IE 501: Optimization Models
2
General InfoSlot 4: M 1135–1230, Tu 0830–0925, Th 0930–1025
At your service: Vishnu Narayanan
Email: [email protected]: x-7878Office: F20, Old CSEPlease contact me in this order!
Teaching Assistant: Virendra Patidar([email protected])
Course webpage:http://www.ieor.iitb.ac.in/∼vishnu/teaching/ie501/Check this page and moodle regularly.
Vishnu Narayanan IE 501: Optimization Models
2
General InfoSlot 4: M 1135–1230, Tu 0830–0925, Th 0930–1025
At your service: Vishnu Narayanan
Email: [email protected]: x-7878Office: F20, Old CSEPlease contact me in this order!
Teaching Assistant: Virendra Patidar([email protected])
Course webpage:http://www.ieor.iitb.ac.in/∼vishnu/teaching/ie501/Check this page and moodle regularly.
Vishnu Narayanan IE 501: Optimization Models
2
General InfoSlot 4: M 1135–1230, Tu 0830–0925, Th 0930–1025
At your service: Vishnu Narayanan
Email: [email protected]: x-7878Office: F20, Old CSEPlease contact me in this order!
Teaching Assistant: Virendra Patidar([email protected])
Course webpage:http://www.ieor.iitb.ac.in/∼vishnu/teaching/ie501/Check this page and moodle regularly.
Vishnu Narayanan IE 501: Optimization Models
2
General InfoSlot 4: M 1135–1230, Tu 0830–0925, Th 0930–1025
At your service: Vishnu Narayanan
Email: [email protected]: x-7878Office: F20, Old CSEPlease contact me in this order!
Teaching Assistant: Virendra Patidar([email protected])
Course webpage:http://www.ieor.iitb.ac.in/∼vishnu/teaching/ie501/Check this page and moodle regularly.
Vishnu Narayanan IE 501: Optimization Models
2
General InfoSlot 4: M 1135–1230, Tu 0830–0925, Th 0930–1025
At your service: Vishnu Narayanan
Email: [email protected]: x-7878Office: F20, Old CSEPlease contact me in this order!
Teaching Assistant: Virendra Patidar([email protected])
Course webpage:http://www.ieor.iitb.ac.in/∼vishnu/teaching/ie501/Check this page and moodle regularly.
Vishnu Narayanan IE 501: Optimization Models
3
Textbooks/References
1 Wayne L. Winston, Operations Research: Applicationsand Algorithms, 4th edition, Thomson Learning, 2004.
2 Wayne L. Winston, Introduction to MathematicalProgramming: Applications and Algorithms, 4th edition,Duxbury, 2003.
3 H. Paul Williams, Model Building in MathematicalProgramming, 4th edition, John Wiley and Sons, 1999.
4 Ashok D. Belegundu and Tirupathi R. Chandrupatla,Optimization Concepts and Applications in Engineering,Pearson Education India, 1999.
5 H. Taha, Operations Research: An Introduction, 8thedition, Prentice Hall India, 2002.
Vishnu Narayanan IE 501: Optimization Models
4
Grading
Regular homework assignments (40%). Modeling, solution,theory, computation. Will be posted on the webpage everymonday!
Project (10%). Formulate and solve a real-world optimizationproblem.
Mid-sem exam (20%) and Final exam (30%).
Audit students: do all of the above and get a passing grade(undergrads: DD, PhD/MSc/M.Tech/Dual degree/rest: CC) foraward of AU.
Cheating⇒ Severe penalty
Vishnu Narayanan IE 501: Optimization Models
4
Grading
Regular homework assignments (40%). Modeling, solution,theory, computation. Will be posted on the webpage everymonday!
Project (10%). Formulate and solve a real-world optimizationproblem.
Mid-sem exam (20%) and Final exam (30%).
Audit students: do all of the above and get a passing grade(undergrads: DD, PhD/MSc/M.Tech/Dual degree/rest: CC) foraward of AU.
Cheating⇒ Severe penalty
Vishnu Narayanan IE 501: Optimization Models
4
Grading
Regular homework assignments (40%). Modeling, solution,theory, computation. Will be posted on the webpage everymonday!
Project (10%). Formulate and solve a real-world optimizationproblem.
Mid-sem exam (20%) and Final exam (30%).
Audit students: do all of the above and get a passing grade(undergrads: DD, PhD/MSc/M.Tech/Dual degree/rest: CC) foraward of AU.
Cheating⇒ Severe penalty
Vishnu Narayanan IE 501: Optimization Models
4
Grading
Regular homework assignments (40%). Modeling, solution,theory, computation. Will be posted on the webpage everymonday!
Project (10%). Formulate and solve a real-world optimizationproblem.
Mid-sem exam (20%) and Final exam (30%).
Audit students: do all of the above and get a passing grade(undergrads: DD, PhD/MSc/M.Tech/Dual degree/rest: CC) foraward of AU.
Cheating⇒ Severe penalty
Vishnu Narayanan IE 501: Optimization Models
4
Grading
Regular homework assignments (40%). Modeling, solution,theory, computation. Will be posted on the webpage everymonday!
Project (10%). Formulate and solve a real-world optimizationproblem.
Mid-sem exam (20%) and Final exam (30%).
Audit students: do all of the above and get a passing grade(undergrads: DD, PhD/MSc/M.Tech/Dual degree/rest: CC) foraward of AU.
Cheating⇒ Severe penalty
Vishnu Narayanan IE 501: Optimization Models
5
What we will learn
Modeling fundamentals
Linear programming
Network models
Combinatorial optimization
Mixed-integer programming
Vishnu Narayanan IE 501: Optimization Models
5
What we will learn
Modeling fundamentals
Linear programming
Network models
Combinatorial optimization
Mixed-integer programming
Vishnu Narayanan IE 501: Optimization Models
5
What we will learn
Modeling fundamentals
Linear programming
Network models
Combinatorial optimization
Mixed-integer programming
Vishnu Narayanan IE 501: Optimization Models
5
What we will learn
Modeling fundamentals
Linear programming
Network models
Combinatorial optimization
Mixed-integer programming
Vishnu Narayanan IE 501: Optimization Models
5
What we will learn
Modeling fundamentals
Linear programming
Network models
Combinatorial optimization
Mixed-integer programming
Vishnu Narayanan IE 501: Optimization Models
6
Modeling fundamentals
Problems from science, engineering, business, and socialsectors
Continuous and discrete problems
Constrained and unconstrained problems
Single- and multi-stage models
Formulations and equivalences
Vishnu Narayanan IE 501: Optimization Models
6
Modeling fundamentals
Problems from science, engineering, business, and socialsectors
Continuous and discrete problems
Constrained and unconstrained problems
Single- and multi-stage models
Formulations and equivalences
Vishnu Narayanan IE 501: Optimization Models
6
Modeling fundamentals
Problems from science, engineering, business, and socialsectors
Continuous and discrete problems
Constrained and unconstrained problems
Single- and multi-stage models
Formulations and equivalences
Vishnu Narayanan IE 501: Optimization Models
6
Modeling fundamentals
Problems from science, engineering, business, and socialsectors
Continuous and discrete problems
Constrained and unconstrained problems
Single- and multi-stage models
Formulations and equivalences
Vishnu Narayanan IE 501: Optimization Models
6
Modeling fundamentals
Problems from science, engineering, business, and socialsectors
Continuous and discrete problems
Constrained and unconstrained problems
Single- and multi-stage models
Formulations and equivalences
Vishnu Narayanan IE 501: Optimization Models
7
Linear Programming
Geometry and algebra of Simplex method
Duality and complementary slackness
Sensitivity analysis
Post-optimal analysis
Vishnu Narayanan IE 501: Optimization Models
7
Linear Programming
Geometry and algebra of Simplex method
Duality and complementary slackness
Sensitivity analysis
Post-optimal analysis
Vishnu Narayanan IE 501: Optimization Models
7
Linear Programming
Geometry and algebra of Simplex method
Duality and complementary slackness
Sensitivity analysis
Post-optimal analysis
Vishnu Narayanan IE 501: Optimization Models
7
Linear Programming
Geometry and algebra of Simplex method
Duality and complementary slackness
Sensitivity analysis
Post-optimal analysis
Vishnu Narayanan IE 501: Optimization Models
8
Network models
Decision problems involving network flows (shortest path, . . . )
Modeling problems as shortest paths, maximum flow, minimumcost flow, etc.
Integrality of solutions
Matching, assignment, and transportation problems
Multi-stage flows
Vishnu Narayanan IE 501: Optimization Models
8
Network models
Decision problems involving network flows (shortest path, . . . )
Modeling problems as shortest paths, maximum flow, minimumcost flow, etc.
Integrality of solutions
Matching, assignment, and transportation problems
Multi-stage flows
Vishnu Narayanan IE 501: Optimization Models
8
Network models
Decision problems involving network flows (shortest path, . . . )
Modeling problems as shortest paths, maximum flow, minimumcost flow, etc.
Integrality of solutions
Matching, assignment, and transportation problems
Multi-stage flows
Vishnu Narayanan IE 501: Optimization Models
8
Network models
Decision problems involving network flows (shortest path, . . . )
Modeling problems as shortest paths, maximum flow, minimumcost flow, etc.
Integrality of solutions
Matching, assignment, and transportation problems
Multi-stage flows
Vishnu Narayanan IE 501: Optimization Models
8
Network models
Decision problems involving network flows (shortest path, . . . )
Modeling problems as shortest paths, maximum flow, minimumcost flow, etc.
Integrality of solutions
Matching, assignment, and transportation problems
Multi-stage flows
Vishnu Narayanan IE 501: Optimization Models
9
Combinatorial optimization models
Examples: knapsack, set cover, set packing, . . .
Large feasible space and neighbourhood solutions
Representation of solution space
Search tree
Search techniques, branch-and-bound
Vishnu Narayanan IE 501: Optimization Models
9
Combinatorial optimization models
Examples: knapsack, set cover, set packing, . . .
Large feasible space and neighbourhood solutions
Representation of solution space
Search tree
Search techniques, branch-and-bound
Vishnu Narayanan IE 501: Optimization Models
9
Combinatorial optimization models
Examples: knapsack, set cover, set packing, . . .
Large feasible space and neighbourhood solutions
Representation of solution space
Search tree
Search techniques, branch-and-bound
Vishnu Narayanan IE 501: Optimization Models
9
Combinatorial optimization models
Examples: knapsack, set cover, set packing, . . .
Large feasible space and neighbourhood solutions
Representation of solution space
Search tree
Search techniques, branch-and-bound
Vishnu Narayanan IE 501: Optimization Models
9
Combinatorial optimization models
Examples: knapsack, set cover, set packing, . . .
Large feasible space and neighbourhood solutions
Representation of solution space
Search tree
Search techniques, branch-and-bound
Vishnu Narayanan IE 501: Optimization Models
10
Mixed-integer programming
Problems with integer variables
Use of binary variables in modeling alternative decisions
Difficulty of solution
Formulations of combinatorial optimization problems asmixed-integer programs
Vishnu Narayanan IE 501: Optimization Models
10
Mixed-integer programming
Problems with integer variables
Use of binary variables in modeling alternative decisions
Difficulty of solution
Formulations of combinatorial optimization problems asmixed-integer programs
Vishnu Narayanan IE 501: Optimization Models
10
Mixed-integer programming
Problems with integer variables
Use of binary variables in modeling alternative decisions
Difficulty of solution
Formulations of combinatorial optimization problems asmixed-integer programs
Vishnu Narayanan IE 501: Optimization Models
10
Mixed-integer programming
Problems with integer variables
Use of binary variables in modeling alternative decisions
Difficulty of solution
Formulations of combinatorial optimization problems asmixed-integer programs
Vishnu Narayanan IE 501: Optimization Models
11
IIT Gandhinagar example
New IIT at Gandhinagar, mentored by IIT Bombay. IITBfaculty shuttle between BOM and AHD to teach courses.
I fly out from BOM on Mondays and fly back from AHD onThursdays for the next five weeks.
Regular round trip fare is Rs. 6,000. One-way ticket costs Rs.4,500.
20% discount if the ticket dates span a weekend (e.g., if I fly outFriday and return Thursday).
How should I buy my tickets to minimize the money spent?
Vishnu Narayanan IE 501: Optimization Models
11
IIT Gandhinagar example
New IIT at Gandhinagar, mentored by IIT Bombay. IITBfaculty shuttle between BOM and AHD to teach courses.
I fly out from BOM on Mondays and fly back from AHD onThursdays for the next five weeks.
Regular round trip fare is Rs. 6,000. One-way ticket costs Rs.4,500.
20% discount if the ticket dates span a weekend (e.g., if I fly outFriday and return Thursday).
How should I buy my tickets to minimize the money spent?
Vishnu Narayanan IE 501: Optimization Models
11
IIT Gandhinagar example
New IIT at Gandhinagar, mentored by IIT Bombay. IITBfaculty shuttle between BOM and AHD to teach courses.
I fly out from BOM on Mondays and fly back from AHD onThursdays for the next five weeks.
Regular round trip fare is Rs. 6,000. One-way ticket costs Rs.4,500.
20% discount if the ticket dates span a weekend (e.g., if I fly outFriday and return Thursday).
How should I buy my tickets to minimize the money spent?
Vishnu Narayanan IE 501: Optimization Models
11
IIT Gandhinagar example
New IIT at Gandhinagar, mentored by IIT Bombay. IITBfaculty shuttle between BOM and AHD to teach courses.
I fly out from BOM on Mondays and fly back from AHD onThursdays for the next five weeks.
Regular round trip fare is Rs. 6,000. One-way ticket costs Rs.4,500.
20% discount if the ticket dates span a weekend (e.g., if I fly outFriday and return Thursday).
How should I buy my tickets to minimize the money spent?
Vishnu Narayanan IE 501: Optimization Models
11
IIT Gandhinagar example
New IIT at Gandhinagar, mentored by IIT Bombay. IITBfaculty shuttle between BOM and AHD to teach courses.
I fly out from BOM on Mondays and fly back from AHD onThursdays for the next five weeks.
Regular round trip fare is Rs. 6,000. One-way ticket costs Rs.4,500.
20% discount if the ticket dates span a weekend (e.g., if I fly outFriday and return Thursday).
How should I buy my tickets to minimize the money spent?
Vishnu Narayanan IE 501: Optimization Models
11
IIT Gandhinagar example
New IIT at Gandhinagar, mentored by IIT Bombay. IITBfaculty shuttle between BOM and AHD to teach courses.
I fly out from BOM on Mondays and fly back from AHD onThursdays for the next five weeks.
Regular round trip fare is Rs. 6,000. One-way ticket costs Rs.4,500.
20% discount if the ticket dates span a weekend (e.g., if I fly outFriday and return Thursday).
How should I buy my tickets to minimize the money spent?
Vishnu Narayanan IE 501: Optimization Models
12
Three possible alternatives
Buy five BOM–AHD–BOM tickets for departure on Mondaysand return on Thursdays.Cost: Rs. 5 × 6,000 = Rs. 30,000.
Buy one BOM–AHD for week 1, one AHD–BOM for week 5,and four AHD–BOM–AHD tickets that span weekends.Cost: Rs. 2 × 4,500 + 4 × 0.8 × 6,000 = Rs. 28,200.
Buy one BOM–AHD–BOM flying out in week 1 and returningin week 5, and four AHD–BOM–AHD tickets that spanweekends.Cost: Rs. 5 × 0.8 × 6,000 = Rs. 24,000.
Vishnu Narayanan IE 501: Optimization Models
12
Three possible alternatives
Buy five BOM–AHD–BOM tickets for departure on Mondaysand return on Thursdays.Cost: Rs. 5 × 6,000 = Rs. 30,000.
Buy one BOM–AHD for week 1, one AHD–BOM for week 5,and four AHD–BOM–AHD tickets that span weekends.Cost: Rs. 2 × 4,500 + 4 × 0.8 × 6,000 = Rs. 28,200.
Buy one BOM–AHD–BOM flying out in week 1 and returningin week 5, and four AHD–BOM–AHD tickets that spanweekends.Cost: Rs. 5 × 0.8 × 6,000 = Rs. 24,000.
Vishnu Narayanan IE 501: Optimization Models
12
Three possible alternatives
Buy five BOM–AHD–BOM tickets for departure on Mondaysand return on Thursdays.Cost: Rs. 5 × 6,000 = Rs. 30,000.
Buy one BOM–AHD for week 1, one AHD–BOM for week 5,and four AHD–BOM–AHD tickets that span weekends.Cost: Rs. 2 × 4,500 + 4 × 0.8 × 6,000 = Rs. 28,200.
Buy one BOM–AHD–BOM flying out in week 1 and returningin week 5, and four AHD–BOM–AHD tickets that spanweekends.Cost: Rs. 5 × 0.8 × 6,000 = Rs. 24,000.
Vishnu Narayanan IE 501: Optimization Models
12
Three possible alternatives
Buy five BOM–AHD–BOM tickets for departure on Mondaysand return on Thursdays.Cost: Rs. 5 × 6,000 = Rs. 30,000.
Buy one BOM–AHD for week 1, one AHD–BOM for week 5,and four AHD–BOM–AHD tickets that span weekends.Cost: Rs. 2 × 4,500 + 4 × 0.8 × 6,000 = Rs. 28,200.
Buy one BOM–AHD–BOM flying out in week 1 and returningin week 5, and four AHD–BOM–AHD tickets that spanweekends.Cost: Rs. 5 × 0.8 × 6,000 = Rs. 24,000.
Vishnu Narayanan IE 501: Optimization Models
12
Three possible alternatives
Buy five BOM–AHD–BOM tickets for departure on Mondaysand return on Thursdays.Cost: Rs. 5 × 6,000 = Rs. 30,000.
Buy one BOM–AHD for week 1, one AHD–BOM for week 5,and four AHD–BOM–AHD tickets that span weekends.Cost: Rs. 2 × 4,500 + 4 × 0.8 × 6,000 = Rs. 28,200.
Buy one BOM–AHD–BOM flying out in week 1 and returningin week 5, and four AHD–BOM–AHD tickets that spanweekends.Cost: Rs. 5 × 0.8 × 6,000 = Rs. 24,000.
Vishnu Narayanan IE 501: Optimization Models
12
Three possible alternatives
Buy five BOM–AHD–BOM tickets for departure on Mondaysand return on Thursdays.Cost: Rs. 5 × 6,000 = Rs. 30,000.
Buy one BOM–AHD for week 1, one AHD–BOM for week 5,and four AHD–BOM–AHD tickets that span weekends.Cost: Rs. 2 × 4,500 + 4 × 0.8 × 6,000 = Rs. 28,200.
Buy one BOM–AHD–BOM flying out in week 1 and returningin week 5, and four AHD–BOM–AHD tickets that spanweekends.Cost: Rs. 5 × 0.8 × 6,000 = Rs. 24,000.
Vishnu Narayanan IE 501: Optimization Models
13
Terminology
In the above situation,
what are the decision alternatives?what are the constraints?what is a criterion for evaluating the alternatives?how many decision alternatives are there?
Another problem: Given a wire of length `, how would onemake a rectangle of maximum area with it?
Alternatives, constraints?
Note: As opposed to the flight problem, the possiblealternatives are uncountably many!
Vishnu Narayanan IE 501: Optimization Models
13
Terminology
In the above situation,
what are the decision alternatives?what are the constraints?what is a criterion for evaluating the alternatives?how many decision alternatives are there?
Another problem: Given a wire of length `, how would onemake a rectangle of maximum area with it?
Alternatives, constraints?
Note: As opposed to the flight problem, the possiblealternatives are uncountably many!
Vishnu Narayanan IE 501: Optimization Models
13
Terminology
In the above situation,
what are the decision alternatives?what are the constraints?what is a criterion for evaluating the alternatives?how many decision alternatives are there?
Another problem: Given a wire of length `, how would onemake a rectangle of maximum area with it?
Alternatives, constraints?
Note: As opposed to the flight problem, the possiblealternatives are uncountably many!
Vishnu Narayanan IE 501: Optimization Models
13
Terminology
In the above situation,
what are the decision alternatives?what are the constraints?what is a criterion for evaluating the alternatives?how many decision alternatives are there?
Another problem: Given a wire of length `, how would onemake a rectangle of maximum area with it?
Alternatives, constraints?
Note: As opposed to the flight problem, the possiblealternatives are uncountably many!
Vishnu Narayanan IE 501: Optimization Models
14
Maximum-area rectangle
Let w and h be the dimensions of the rectangle. (decisionvariables)
Constraints: 2(w + h) = `, w ≥ 0, h ≥ 0.
Objective function: area = wh
Optimization problem:
maximize z = whsubject to 2(w + h) = `
w, h ≥ 0.
Vishnu Narayanan IE 501: Optimization Models
14
Maximum-area rectangle
Let w and h be the dimensions of the rectangle. (decisionvariables)
Constraints: 2(w + h) = `, w ≥ 0, h ≥ 0.
Objective function: area = wh
Optimization problem:
maximize z = whsubject to 2(w + h) = `
w, h ≥ 0.
Vishnu Narayanan IE 501: Optimization Models
14
Maximum-area rectangle
Let w and h be the dimensions of the rectangle. (decisionvariables)
Constraints: 2(w + h) = `, w ≥ 0, h ≥ 0.
Objective function: area = wh
Optimization problem:
maximize z = whsubject to 2(w + h) = `
w, h ≥ 0.
Vishnu Narayanan IE 501: Optimization Models
14
Maximum-area rectangle
Let w and h be the dimensions of the rectangle. (decisionvariables)
Constraints: 2(w + h) = `, w ≥ 0, h ≥ 0.
Objective function: area = wh
Optimization problem:
maximize z = whsubject to 2(w + h) = `
w, h ≥ 0.
Vishnu Narayanan IE 501: Optimization Models
14
Maximum-area rectangle
Let w and h be the dimensions of the rectangle. (decisionvariables)
Constraints: 2(w + h) = `, w ≥ 0, h ≥ 0.
Objective function: area = wh
Optimization problem:
maximize z = whsubject to 2(w + h) = `
w, h ≥ 0.
Vishnu Narayanan IE 501: Optimization Models
14
Maximum-area rectangle
Let w and h be the dimensions of the rectangle. (decisionvariables)
Constraints: 2(w + h) = `, w ≥ 0, h ≥ 0.
Objective function: area = wh
Optimization problem:
maximize z = whsubject to 2(w + h) = `
w, h ≥ 0.
Vishnu Narayanan IE 501: Optimization Models
14
Maximum-area rectangle
Let w and h be the dimensions of the rectangle. (decisionvariables)
Constraints: 2(w + h) = `, w ≥ 0, h ≥ 0.
Objective function: area = wh
Optimization problem:
maximize z = whsubject to 2(w + h) = `
w, h ≥ 0.
Vishnu Narayanan IE 501: Optimization Models
14
Maximum-area rectangle
Let w and h be the dimensions of the rectangle. (decisionvariables)
Constraints: 2(w + h) = `, w ≥ 0, h ≥ 0.
Objective function: area = wh
Optimization problem:
maximize z = whsubject to 2(w + h) = `
w, h ≥ 0.
Vishnu Narayanan IE 501: Optimization Models
14
Maximum-area rectangle
Let w and h be the dimensions of the rectangle. (decisionvariables)
Constraints: 2(w + h) = `, w ≥ 0, h ≥ 0.
Objective function: area = wh
Optimization problem:
maximize z = whsubject to 2(w + h) = `
w, h ≥ 0.
Vishnu Narayanan IE 501: Optimization Models
15
More terminology
problem: min{f(x) : x ∈ S}
If x satisfies all constraints (i.e., x ∈ S), then it is a feasiblesolution. Otherwise, it is infeasible.
The set of all feasible solutions (in this case, S) is called thefeasible region.
x is called optimal if x ∈ S, and f(y) ≥ f(x) for all y ∈ S.
If x is feasible but not optimal, it is called suboptimal.
Vishnu Narayanan IE 501: Optimization Models
15
More terminology
problem: min{f(x) : x ∈ S}
If x satisfies all constraints (i.e., x ∈ S), then it is a feasiblesolution. Otherwise, it is infeasible.
The set of all feasible solutions (in this case, S) is called thefeasible region.
x is called optimal if x ∈ S, and f(y) ≥ f(x) for all y ∈ S.
If x is feasible but not optimal, it is called suboptimal.
Vishnu Narayanan IE 501: Optimization Models
15
More terminology
problem: min{f(x) : x ∈ S}
If x satisfies all constraints (i.e., x ∈ S), then it is a feasiblesolution. Otherwise, it is infeasible.
The set of all feasible solutions (in this case, S) is called thefeasible region.
x is called optimal if x ∈ S, and f(y) ≥ f(x) for all y ∈ S.
If x is feasible but not optimal, it is called suboptimal.
Vishnu Narayanan IE 501: Optimization Models
15
More terminology
problem: min{f(x) : x ∈ S}
If x satisfies all constraints (i.e., x ∈ S), then it is a feasiblesolution. Otherwise, it is infeasible.
The set of all feasible solutions (in this case, S) is called thefeasible region.
x is called optimal if x ∈ S, and f(y) ≥ f(x) for all y ∈ S.
If x is feasible but not optimal, it is called suboptimal.
Vishnu Narayanan IE 501: Optimization Models
15
More terminology
problem: min{f(x) : x ∈ S}
If x satisfies all constraints (i.e., x ∈ S), then it is a feasiblesolution. Otherwise, it is infeasible.
The set of all feasible solutions (in this case, S) is called thefeasible region.
x is called optimal if x ∈ S, and f(y) ≥ f(x) for all y ∈ S.
If x is feasible but not optimal, it is called suboptimal.
Vishnu Narayanan IE 501: Optimization Models
15
More terminology
problem: min{f(x) : x ∈ S}
If x satisfies all constraints (i.e., x ∈ S), then it is a feasiblesolution. Otherwise, it is infeasible.
The set of all feasible solutions (in this case, S) is called thefeasible region.
x is called optimal if x ∈ S, and f(y) ≥ f(x) for all y ∈ S.
If x is feasible but not optimal, it is called suboptimal.
Vishnu Narayanan IE 501: Optimization Models