Objectives

CS 312: Algorithm Analysis

ƒ Remember the form of a generic greedy
algorithm
ƒ Define the Knapsack Problem
ƒ Learn how to formulate the problem in
algebraic terms
Lecture #9: Knapsack ƒ Get exposed to the formulation as a linear
program (for future reference)
ƒ Work through an example

Generic Greedy Alg. The Knapsack Problem

Function greedy (C:set): set
{C is the set of candidates} ƒ Given n objects with
S ← ∅ {S will be the solution} weight wi and value vi
while (C != ∅ & !solution(S))
ƒ Knapsack can only hold a
x ← select(C)
total weight of W
C ← C \ {x}
if feasible(S∪{x}) then
ƒ Fill Knapsack
ƒ Maximize total value
S ← S ∪ {x}
if solution(S) then return S ƒ Don’t exceed W
else “there are no solutions”

The Knapsack Problem The Knapsack Problem

Mathematically, represent a selection Mathematically, represent a selection
from n objects by a vector x from n objects by a vector x

⎡1 ⎤ ⎡1 ⎤ ⎡.75⎤ ⎡ x1 ⎤ ⎡ w1 ⎤ ⎡ v1 ⎤
⎢0 ⎥ ⎢0 ⎥ ⎢0⎥ ⎢x ⎥ ⎢w ⎥ ⎢v ⎥
⎢ ⎥ ⎢ ⎥ ⎢ ⎥ ⎢ 2⎥ ⎢ 2⎥ ⎢ 2⎥
⎢0 ⎥ and ⎢0 ⎥ and ⎢0⎥ ⎢x ⎥ ⎢w ⎥ ⎢v ⎥
⎢ ⎥ ⎢ ⎥ ⎢ ⎥ x = ⎢ 3⎥ w = ⎢ 3⎥ v = ⎢ 3⎥
⎢0 ⎥ ⎢0 ⎥ ⎢0⎥ ⎢ x4 ⎥ ⎢ w4 ⎥ ⎢v4 ⎥
⎢0 ⎥ ⎢1 ⎥ ⎢0⎥ ⎢ x5 ⎥ ⎢ w5 ⎥ ⎢v5 ⎥
⎢ ⎥ ⎢ ⎥ ⎢ ⎥ ⎢ ⎥ ⎢ ⎥ ⎢ ⎥
⎣⎢0⎦⎥ ⎣⎢0⎦⎥ ⎣⎢ 0 ⎦⎥ ⎣⎢ x6 ⎦⎥ ⎣⎢ w6 ⎦⎥ ⎣⎢v6 ⎦⎥
Fix Order
of Objects means means
{ , } means
Fix Order
of Objects 0 ≤ xi ≤ 1
Corresponding
Object Weights
Corresponding
Object Values

1
 The Knapsack Problem The Knapsack Problem
⎡ x1 ⎤
The problem is written mathematically as ⎢x ⎥ The problem is written mathematically as
⎢ 2⎥
T ⎢x ⎥ n
max v x [v1 v2 v3 v4 v5 v6 ]⎢ 3 ⎥ = ∑ vi xi max vT x
⎢ x4 ⎥ i =1
x ⎢ x5 ⎥ x
⎢ ⎥ ⎡ x1 ⎤
⎣⎢ x6 ⎦⎥ ⎢x ⎥
subject to subject to
⎢ 2⎥
⎢x ⎥ n
wT x ≤ W w x ≤W
T [w1 w2 w3 w4 w5 w6 ]⎢ 3 ⎥ = ∑ wi xi
⎢ x4 ⎥ i =1
⎢ x5 ⎥
where vi>0, wi>0, and 0<=xi<=1 for all 1<= i<=n. where vi>0, wi>0, and 0<=xi<=1 for all 1<= i<=n. ⎢⎢ x ⎥⎥
⎣ 6⎦

The Knapsack Problem The Knapsack Problem

The problem is written mathematically as Problems of this form are called Linear Programs.
n

∑ vi xi
T
max v x max max vT x
x x i =1 x
subject to
= n subject to
subject to ∑ wi xi ≤ W
wT x ≤ W i =1 wT x ≤ W
where vi>0, wi>0, and 0<=xi<=1 for all 1<= i<=n.

The Knapsack Problem The Knapsack Problem

Problems of this form are called Linear Programs. Problems of this form are called Linear Programs.
Objective Function Objective Function
T
max v x max f ( x ) max vT x max f ( x )
x Candidates x x Candidates x
subject to subject to subject to subject to

wT x ≤ W Feasibility
Check
g ( x )≤ W wT x ≤ W Feasibility
Check
g ( x )≤ W
They are a special case of general mathematical programs, If we can introduce a meaningful Selection Function, we
or optimization problems. can design Greedy Algorithms to try to solve the
optimization problem!

2
 Knapsack Knapsack Selection Fn.
ƒ What is the selection function if…
function knapsack (w[1…n], v[1…n], W) : array [1…n]
for i = 1 to n do x[i] ← 0 ƒ want to maximize profit?
weight ← 0
while weight < W do
i ← the best remaining object ƒ want to minimize profit (cost)?
if weight + w[i] <= W then x[i] ← 1
weight ← weight + w[i]
else x[i] ← (W - weight)/w[i]
weight ← W ƒ What is the complexity of the algorithm?
return x

The Knapsack Problem The Knapsack Problem

Greedy Algorithm: function knapsack(w[1..n],v[1..n],W):array [1..n] Greedy Algorithm: function knapsack(w[1..n],v[1..n],W):array [1..n]
{initialization} {initialization}
T
max v x for i = 1 to n do x[i] ← 0
weight ← 0
for i = 1 to n do x[i] ← 0
weight ← 0
x {Greedy Loop} {Greedy Loop}
Selection Function Ideas while weight < W do
while weight < W do
• Most valuable
subject to i ← the best remaining object • Lightest weight i ← the best remaining object
if weight+w[i]<=W then • Highest value/unit weight if weight+w[i]<=W then

wT x ≤ W x[i] ← 1
weight ← weight + w[i]
x[i] ← 1
weight ← weight + w[i]
else x[i] ← (W – weight)/w[i] else x[i] ← (W – weight)/w[i]
weight ← W weight ← W
return x return x

The Knapsack Problem The Knapsack Problem

Example: Let W=100 and consider Example: Let W=100 and consider
⎡10 ⎤ ⎡20⎤ ⎡2⎤ ⎡10 ⎤ ⎡20⎤ ⎡2⎤
⎢20⎥ ⎢30 ⎥ ⎢1.5 ⎥ ⎢20⎥ ⎢30 ⎥ ⎢1.5 ⎥
⎢ ⎥ ⎢ ⎥ ⎢ ⎥ ⎢ ⎥ ⎢ ⎥ ⎢ ⎥
w = ⎢30⎥ v = ⎢66⎥ v / w = ⎢2.2⎥ w = ⎢30⎥ v = ⎢66⎥ v / w = ⎢2.2⎥
Selection Function Ideas ⎢ ⎥ ⎢ ⎥ ⎢ ⎥ Selection Function Ideas ⎢ ⎥ ⎢ ⎥ ⎢ ⎥
• Most valuable ⎢40⎥ ⎢40⎥ ⎢1⎥ • Most valuable ⎢40⎥ ⎢40⎥ ⎢1⎥
• Lightest weight ⎣⎢50⎦⎥ ⎣⎢60⎦⎥ ⎣⎢1.2 ⎦⎥ • Lightest weight ⎣⎢50⎦⎥ ⎣⎢60⎦⎥ ⎣⎢1.2 ⎦⎥
• Highest value/unit weight • Highest value/unit weight
⎡0⎤
⎢0⎥
⎢ ⎥
x = ⎢0⎥ weight = 0
⎢ ⎥
⎢0⎥
⎢⎣0⎥⎦

3
 The Knapsack Problem The Knapsack Problem
Example: Let W=100 and consider Example: Let W=100 and consider
⎡10 ⎤ ⎡20⎤ ⎡2⎤ ⎡10 ⎤ ⎡20⎤ ⎡2⎤
⎢20⎥ ⎢30 ⎥ ⎢1.5 ⎥ ⎢20⎥ ⎢30 ⎥ ⎢1.5 ⎥
⎢ ⎥ ⎢ ⎥ ⎢ ⎥ ⎢ ⎥ ⎢ ⎥ ⎢ ⎥
w = ⎢30⎥ v = ⎢66⎥ v / w = ⎢2.2⎥ w = ⎢30⎥ v = ⎢66⎥ v / w = ⎢2.2⎥
Selection Function Ideas ⎢ ⎥ ⎢ ⎥ ⎢ ⎥ Selection Function Ideas ⎢ ⎥ ⎢ ⎥ ⎢ ⎥
• Most valuable ⎢40⎥ ⎢40⎥ ⎢1⎥ • Most valuable ⎢40⎥ ⎢40⎥ ⎢1⎥
• Lightest weight ⎣⎢50⎦⎥ ⎣⎢60⎦⎥ ⎣⎢1.2 ⎦⎥ • Lightest weight ⎣⎢50⎦⎥ ⎣⎢60⎦⎥ ⎣⎢1.2 ⎦⎥
• Highest value/unit weight • Highest value/unit weight
⎡0⎤ ⎡0⎤
⎢0⎥ ⎢0⎥
⎢ ⎥ ⎢ ⎥
x = ⎢0⎥ weight = 0 x = ⎢0⎥ weight = 0
⎢ ⎥ ⎢ ⎥
⎢0⎥ ⎢0⎥
⎢⎣0⎥⎦ ⎢⎣0⎥⎦

The Knapsack Problem The Knapsack Problem

Example: Let W=100 and consider Example: Let W=100 and consider
⎡10 ⎤ ⎡20⎤ ⎡2⎤ ⎡10 ⎤ ⎡20⎤ ⎡2⎤
⎢20⎥ ⎢30 ⎥ ⎢1.5 ⎥ ⎢20⎥ ⎢30 ⎥ ⎢1.5 ⎥
⎢ ⎥ ⎢ ⎥ ⎢ ⎥ ⎢ ⎥ ⎢ ⎥ ⎢ ⎥
w = ⎢30⎥ v = ⎢66⎥ v / w = ⎢2.2⎥ w=⎢−⎥ v=⎢−⎥ v / w = ⎢2.2⎥
Selection Function Ideas ⎢ ⎥ ⎢ ⎥ ⎢ ⎥ Selection Function Ideas ⎢ ⎥ ⎢ ⎥ ⎢ ⎥
• Most valuable ⎢40⎥ ⎢40⎥ ⎢1⎥ • Most valuable ⎢40⎥ ⎢40⎥ ⎢1⎥
• Lightest weight ⎣⎢50⎦⎥ ⎣⎢60⎦⎥ ⎣⎢1.2 ⎦⎥ • Lightest weight ⎣⎢50⎦⎥ ⎣⎢60⎦⎥ ⎣⎢1.2 ⎦⎥
• Highest value/unit weight • Highest value/unit weight
⎡0⎤ ⎡0⎤
⎢0⎥ ⎢0⎥
⎢ ⎥ ⎢ ⎥
x = ⎢1⎥ weight = 30 x = ⎢1⎥ weight = 30
⎢ ⎥ ⎢ ⎥
⎢0⎥ ⎢0⎥
⎢⎣0⎥⎦ ⎢⎣0⎥⎦

The Knapsack Problem The Knapsack Problem

Example: Let W=100 and consider Example: Let W=100 and consider
⎡10 ⎤ ⎡20⎤ ⎡2⎤ ⎡10 ⎤ ⎡20⎤ ⎡2⎤
⎢20⎥ ⎢30 ⎥ ⎢1.5 ⎥ ⎢20⎥ ⎢30 ⎥ ⎢1.5 ⎥
⎢ ⎥ ⎢ ⎥ ⎢ ⎥ ⎢ ⎥ ⎢ ⎥ ⎢ ⎥
w=⎢−⎥ v=⎢−⎥ v / w = ⎢2.2⎥ w=⎢−⎥ v=⎢−⎥ v / w = ⎢2.2⎥
Selection Function Ideas ⎢ ⎥ ⎢ ⎥ ⎢ ⎥ Selection Function Ideas ⎢ ⎥ ⎢ ⎥ ⎢ ⎥
• Most valuable ⎢40⎥ ⎢40⎥ ⎢1⎥ • Most valuable ⎢40⎥ ⎢40⎥ ⎢1⎥
• Lightest weight ⎣⎢50⎦⎥ ⎣⎢60⎦⎥ ⎣⎢1.2 ⎦⎥ • Lightest weight ⎣⎢50⎦⎥ ⎣⎢60⎦⎥ ⎣⎢1.2 ⎦⎥
• Highest value/unit weight • Highest value/unit weight
⎡0⎤ ⎡0⎤
⎢0⎥ ⎢0⎥
⎢ ⎥ ⎢ ⎥
x = ⎢1⎥ weight = 30 x = ⎢1⎥ weight = 80
⎢ ⎥ ⎢ ⎥
⎢0⎥ ⎢0⎥
⎢⎣0⎥⎦ ⎢⎣1⎥⎦

4
 The Knapsack Problem The Knapsack Problem
Example: Let W=100 and consider Example: Let W=100 and consider
⎡10 ⎤ ⎡20⎤ ⎡2⎤ ⎡10 ⎤ ⎡20⎤ ⎡2⎤
⎢20⎥ ⎢30 ⎥ ⎢1.5 ⎥ ⎢20⎥ ⎢30 ⎥ ⎢1.5 ⎥
⎢ ⎥ ⎢ ⎥ ⎢ ⎥ ⎢ ⎥ ⎢ ⎥ ⎢ ⎥
w=⎢−⎥ v=⎢−⎥ v / w = ⎢2.2⎥ w=⎢−⎥ v=⎢−⎥ v / w = ⎢2.2⎥
Selection Function Ideas ⎢ ⎥ ⎢ ⎥ ⎢ ⎥ Selection Function Ideas ⎢ ⎥ ⎢ ⎥ ⎢ ⎥
• Most valuable ⎢40⎥ ⎢40⎥ ⎢1⎥ • Most valuable ⎢40⎥ ⎢40⎥ ⎢1⎥
• Lightest weight ⎣⎢ − ⎦⎥ ⎣⎢ − ⎦⎥ ⎣⎢1.2 ⎦⎥ • Lightest weight ⎣⎢ − ⎦⎥ ⎣⎢ − ⎦⎥ ⎣⎢1.2 ⎦⎥
• Highest value/unit weight • Highest value/unit weight
⎡0⎤ ⎡0⎤
⎢0⎥ ⎢0⎥
⎢ ⎥ ⎢ ⎥
x = ⎢1⎥ weight = 80 x = ⎢1⎥ weight = 80
⎢ ⎥ ⎢ ⎥
⎢0⎥ ⎢0⎥
⎢⎣1⎥⎦ ⎢⎣1⎥⎦

The Knapsack Problem The Knapsack Problem

Example: Let W=100 and consider Example: Let W=100 and consider
⎡10 ⎤ ⎡20⎤ ⎡2⎤ ⎡10 ⎤ ⎡20⎤ ⎡2⎤
⎢20⎥ ⎢30 ⎥ ⎢1.5 ⎥ ⎢20⎥ ⎢30 ⎥ ⎢1.5 ⎥
⎢ ⎥ ⎢ ⎥ ⎢ ⎥ ⎢ ⎥ ⎢ ⎥ ⎢ ⎥
w=⎢−⎥ v=⎢−⎥ v / w = ⎢2.2⎥ w=⎢−⎥ v=⎢−⎥ v / w = ⎢2.2⎥
Selection Function Ideas ⎢ ⎥ ⎢ ⎥ ⎢ ⎥ Selection Function Ideas ⎢ ⎥ ⎢ ⎥ ⎢ ⎥
• Most valuable ⎢40⎥ ⎢40⎥ ⎢1⎥ • Most valuable ⎢−⎥ ⎢−⎥ ⎢1⎥
• Lightest weight ⎣⎢ − ⎦⎥ ⎣⎢ − ⎦⎥ ⎣⎢1.2 ⎦⎥ • Lightest weight ⎣⎢ − ⎦⎥ ⎣⎢ − ⎦⎥ ⎣⎢1.2 ⎦⎥
• Highest value/unit weight • Highest value/unit weight
⎡0⎤ ⎡0⎤
⎢0⎥ ⎢0⎥
⎢ ⎥ Score 1 = 146 ⎢ ⎥
x=⎢ 1 ⎥ weight = 100 x=⎢ 1 ⎥ weight = 100
⎢ ⎥ ⎢ ⎥
⎢0.5⎥ ⎢0.5⎥
⎢⎣ 1 ⎥⎦ ⎢⎣ 1 ⎥⎦

The Knapsack Problem The Knapsack Problem

Example: Let W=100 and consider Example: Let W=100 and consider
⎡10 ⎤ ⎡20⎤ ⎡2⎤ ⎡10 ⎤ ⎡20⎤ ⎡2⎤
⎢20⎥ ⎢30 ⎥ ⎢1.5 ⎥ ⎢20⎥ ⎢30 ⎥ ⎢1.5 ⎥
⎢ ⎥ ⎢ ⎥ ⎢ ⎥ ⎢ ⎥ ⎢ ⎥ ⎢ ⎥
w = ⎢30⎥ v = ⎢66⎥ v / w = ⎢2.2⎥ w = ⎢30⎥ v = ⎢66⎥ v / w = ⎢2.2⎥
Selection Function Ideas ⎢ ⎥ ⎢ ⎥ ⎢ ⎥ Selection Function Ideas ⎢ ⎥ ⎢ ⎥ ⎢ ⎥
• Most valuable ⎢40⎥ ⎢40⎥ ⎢1⎥ • Most valuable ⎢40⎥ ⎢40⎥ ⎢1⎥
• Lightest weight ⎣⎢50⎦⎥ ⎣⎢60⎦⎥ ⎣⎢1.2 ⎦⎥ • Lightest weight ⎣⎢50⎦⎥ ⎣⎢60⎦⎥ ⎣⎢1.2 ⎦⎥
• Highest value/unit weight • Highest value/unit weight
⎡0⎤ ⎡1⎤
⎢0⎥ ⎢1⎥
Score 1 = 146 ⎢ ⎥ Score 1 = 146 ⎢ ⎥
x = ⎢0⎥ weight = 0 x = ⎢1⎥ weight = 100
⎢ ⎥ ⎢ ⎥
⎢0⎥ ⎢1⎥
⎢⎣0⎥⎦ ⎢⎣0⎥⎦

5
 The Knapsack Problem The Knapsack Problem
Example: Let W=100 and consider Example: Let W=100 and consider
⎡10 ⎤ ⎡20⎤ ⎡2⎤ ⎡10 ⎤ ⎡20⎤ ⎡2⎤
⎢20⎥ ⎢30 ⎥ ⎢1.5 ⎥ ⎢20⎥ ⎢30 ⎥ ⎢1.5 ⎥
⎢ ⎥ ⎢ ⎥ ⎢ ⎥ ⎢ ⎥ ⎢ ⎥ ⎢ ⎥
w = ⎢30⎥ v = ⎢66⎥ v / w = ⎢2.2⎥ w = ⎢30⎥ v = ⎢66⎥ v / w = ⎢2.2⎥
Selection Function Ideas ⎢ ⎥ ⎢ ⎥ ⎢ ⎥ Selection Function Ideas ⎢ ⎥ ⎢ ⎥ ⎢ ⎥
• Most valuable ⎢40⎥ ⎢40⎥ ⎢1⎥ • Most valuable ⎢40⎥ ⎢40⎥ ⎢1⎥
• Lightest weight ⎣⎢50⎦⎥ ⎣⎢60⎦⎥ ⎣⎢1.2 ⎦⎥ • Lightest weight ⎣⎢50⎦⎥ ⎣⎢60⎦⎥ ⎣⎢1.2 ⎦⎥
• Highest value/unit weight • Highest value/unit weight
⎡1⎤ ⎡0⎤
⎢1⎥ ⎢0⎥
Score 1 = 146 ⎢ ⎥ Score 1 = 146 ⎢ ⎥
Score 2 = 156 x = ⎢1⎥ weight = 100 Score 2 = 156 x = ⎢0⎥ weight = 0
⎢ ⎥ ⎢ ⎥
⎢1⎥ ⎢0⎥
⎢⎣0⎥⎦ ⎢⎣0⎥⎦

The Knapsack Problem The Knapsack Problem

Example: Let W=100 and consider Example: Let W=100 and consider
⎡10 ⎤ ⎡20⎤ ⎡2⎤ ⎡10 ⎤ ⎡20⎤ ⎡2⎤
⎢20⎥ ⎢30 ⎥ ⎢1.5 ⎥ ⎢20⎥ ⎢30 ⎥ ⎢1.5 ⎥
⎢ ⎥ ⎢ ⎥ ⎢ ⎥ ⎢ ⎥ ⎢ ⎥ ⎢ ⎥
w = ⎢30⎥ v = ⎢66⎥ v / w = ⎢2.2⎥ w = ⎢30⎥ v = ⎢66⎥ v / w = ⎢2.2⎥
Selection Function Ideas ⎢ ⎥ ⎢ ⎥ ⎢ ⎥ Selection Function Ideas ⎢ ⎥ ⎢ ⎥ ⎢ ⎥
• Most valuable ⎢40⎥ ⎢40⎥ ⎢1⎥ • Most valuable ⎢40⎥ ⎢40⎥ ⎢1⎥
• Lightest weight ⎣⎢50⎦⎥ ⎣⎢60⎦⎥ ⎣⎢1.2 ⎦⎥ • Lightest weight ⎣⎢50⎦⎥ ⎣⎢60⎦⎥ ⎣⎢1.2 ⎦⎥
• Highest value/unit weight • Highest value/unit weight
⎡0⎤ ⎡0⎤
⎢0⎥ ⎢0⎥
Score 1 = 146 ⎢ ⎥ Score 1 = 146 ⎢ ⎥
Score 2 = 156 x = ⎢0⎥ weight = 0 Score 2 = 156 x = ⎢0⎥ weight = 0
⎢ ⎥ ⎢ ⎥
⎢0⎥ ⎢0⎥
⎢⎣0⎥⎦ ⎢⎣0⎥⎦

The Knapsack Problem The Knapsack Problem

Example: Let W=100 and consider Example: Let W=100 and consider
⎡10 ⎤ ⎡20⎤ ⎡2⎤ ⎡10 ⎤ ⎡20⎤ ⎡2⎤
⎢20⎥ ⎢30 ⎥ ⎢1.5 ⎥ ⎢20⎥ ⎢30 ⎥ ⎢1.5 ⎥
⎢ ⎥ ⎢ ⎥ ⎢ ⎥ ⎢ ⎥ ⎢ ⎥ ⎢ ⎥
w=⎢−⎥ v = ⎢66⎥ v / w = ⎢2.2⎥ w=⎢−⎥ v = ⎢66⎥ v / w = ⎢2.2⎥
Selection Function Ideas ⎢ ⎥ ⎢ ⎥ ⎢ ⎥ Selection Function Ideas ⎢ ⎥ ⎢ ⎥ ⎢ ⎥
• Most valuable ⎢40⎥ ⎢40⎥ ⎢1⎥ • Most valuable ⎢40⎥ ⎢40⎥ ⎢1⎥
• Lightest weight ⎣⎢50⎦⎥ ⎣⎢60⎦⎥ ⎣⎢1.2 ⎦⎥ • Lightest weight ⎣⎢50⎦⎥ ⎣⎢60⎦⎥ ⎣⎢1.2 ⎦⎥
• Highest value/unit weight • Highest value/unit weight
⎡0⎤ ⎡0⎤
⎢0⎥ ⎢0⎥
Score 1 = 146 ⎢ ⎥ Score 1 = 146 ⎢ ⎥
Score 2 = 156 x = ⎢1⎥ weight = 30 Score 2 = 156 x = ⎢1⎥ weight = 30
⎢ ⎥ ⎢ ⎥
⎢0⎥ ⎢0⎥
⎢⎣0⎥⎦ ⎢⎣0⎥⎦

6
 The Knapsack Problem The Knapsack Problem
Example: Let W=100 and consider Example: Let W=100 and consider
⎡−⎤ ⎡20⎤ ⎡2⎤ ⎡−⎤ ⎡20⎤ ⎡2⎤
⎢20⎥ ⎢30 ⎥ ⎢1.5 ⎥ ⎢20⎥ ⎢30 ⎥ ⎢1.5 ⎥
⎢ ⎥ ⎢ ⎥ ⎢ ⎥ ⎢ ⎥ ⎢ ⎥ ⎢ ⎥
w=⎢−⎥ v = ⎢66⎥ v / w = ⎢2.2⎥ w=⎢−⎥ v = ⎢66⎥ v / w = ⎢2.2⎥
Selection Function Ideas ⎢ ⎥ ⎢ ⎥ ⎢ ⎥ Selection Function Ideas ⎢ ⎥ ⎢ ⎥ ⎢ ⎥
• Most valuable ⎢40⎥ ⎢40⎥ ⎢1⎥ • Most valuable ⎢40⎥ ⎢40⎥ ⎢1⎥
• Lightest weight ⎣⎢50⎦⎥ ⎣⎢60⎦⎥ ⎣⎢1.2 ⎦⎥ • Lightest weight ⎣⎢50⎦⎥ ⎣⎢60⎦⎥ ⎣⎢1.2 ⎦⎥
• Highest value/unit weight • Highest value/unit weight
⎡1⎤ ⎡1⎤
⎢0⎥ ⎢0⎥
Score 1 = 146 ⎢ ⎥ Score 1 = 146 ⎢ ⎥
Score 2 = 156 x = ⎢1⎥ weight = 40 Score 2 = 156 x = ⎢1⎥ weight = 40
⎢ ⎥ ⎢ ⎥
⎢0⎥ ⎢0⎥
⎢⎣0⎥⎦ ⎢⎣0⎥⎦

The Knapsack Problem The Knapsack Problem

Example: Let W=100 and consider Example: Let W=100 and consider
⎡−⎤ ⎡20⎤ ⎡2⎤ ⎡−⎤ ⎡20⎤ ⎡2⎤
⎢−⎥ ⎢30 ⎥ ⎢1.5 ⎥ ⎢−⎥ ⎢30 ⎥ ⎢1.5 ⎥
⎢ ⎥ ⎢ ⎥ ⎢ ⎥ ⎢ ⎥ ⎢ ⎥ ⎢ ⎥
w=⎢−⎥ v = ⎢66⎥ v / w = ⎢2.2⎥ w=⎢−⎥ v = ⎢66⎥ v / w = ⎢2.2⎥
Selection Function Ideas ⎢ ⎥ ⎢ ⎥ ⎢ ⎥ Selection Function Ideas ⎢ ⎥ ⎢ ⎥ ⎢ ⎥
• Most valuable ⎢40⎥ ⎢40⎥ ⎢1⎥ • Most valuable ⎢40⎥ ⎢40⎥ ⎢1⎥
• Lightest weight ⎣⎢50⎦⎥ ⎣⎢60⎦⎥ ⎣⎢1.2 ⎦⎥ • Lightest weight ⎣⎢50⎦⎥ ⎣⎢60⎦⎥ ⎣⎢1.2 ⎦⎥
• Highest value/unit weight • Highest value/unit weight
⎡1⎤ ⎡1⎤
⎢1⎥ ⎢1⎥
Score 1 = 146 ⎢ ⎥ Score 1 = 146 ⎢ ⎥
Score 2 = 156 x = ⎢1⎥ weight = 60 Score 2 = 156 x = ⎢1⎥ weight = 60
⎢ ⎥ ⎢ ⎥
⎢0⎥ ⎢0⎥
⎢⎣0⎥⎦ ⎢⎣0⎥⎦

The Knapsack Problem The Knapsack Problem

Example: Let W=100 and consider Example: Let W=100 and consider
⎡−⎤ ⎡20⎤ ⎡2⎤ ⎡−⎤ ⎡20⎤ ⎡2⎤
⎢−⎥ ⎢30 ⎥ ⎢1.5 ⎥ ⎢−⎥ ⎢30 ⎥ ⎢1.5 ⎥
⎢ ⎥ ⎢ ⎥ ⎢ ⎥ ⎢ ⎥ ⎢ ⎥ ⎢ ⎥
w=⎢−⎥ v = ⎢66⎥ v / w = ⎢2.2⎥ w=⎢−⎥ v = ⎢66⎥ v / w = ⎢2.2⎥
Selection Function Ideas ⎢ ⎥ ⎢ ⎥ ⎢ ⎥ Selection Function Ideas ⎢ ⎥ ⎢ ⎥ ⎢ ⎥
• Most valuable ⎢40⎥ ⎢40⎥ ⎢1⎥ • Most valuable ⎢40⎥ ⎢40⎥ ⎢1⎥
• Lightest weight ⎣⎢ − ⎦⎥ ⎣⎢60⎦⎥ ⎣⎢1.2 ⎦⎥ • Lightest weight ⎣⎢ − ⎦⎥ ⎣⎢60⎦⎥ ⎣⎢1.2 ⎦⎥
• Highest value/unit weight • Highest value/unit weight
⎡1⎤ ⎡1⎤
⎢1⎥ ⎢1⎥
Score 1 = 146 ⎢ ⎥ Score 1 = 146 ⎢ ⎥
Score 2 = 156 x=⎢ 1 ⎥ weight = 100 Score 2 = 156 x=⎢ 1 ⎥ weight = 100
⎢ ⎥ Score 3 = 164 ⎢ ⎥
⎢0⎥ ⎢0⎥
⎢⎣0.8⎥⎦ ⎢⎣0.8⎥⎦

7
 The Knapsack Problem Assignment
Theorem: If objects are selected in order of decreasing ƒ HW #6.5
vi/wi, then algorithm knapsack finds an optimal solution.
ƒ Question 6.18
Proof: Work through proof
Selection Function Ideas In book on page 204. ƒ Read 7.1, 7.2
• Most valuable
• Lightest weight
• Highest value/unit weight
Efficiency: Show that the
Algorithm exhibits time
Score 1 = 146
complexity that grows in
Score 2 = 156
O(nlogn).
Score 3 = 164