Class 5 - 2D Maxima Sweep-Line Algorithm

2D Maxima Sweep-Line
Algorithm
(Class 5)
From Book’s Chapter 2

• We want to improve the efficiency of the algorithm we developed in
previous class.
• For example, one algorithm has 𝑂𝑂(𝑛𝑛2 ) running time.
• And then we improved the algorithm that has 𝑂𝑂(𝑛𝑛) running time.
• We introduced a brute-force algorithm that ran in 𝑂𝑂(𝑛𝑛2 ) time.
• It operated by comparing all pairs of points.
• Is there an approach that is significantly better?
• The problem with the brute-force algorithm is that it uses no
intelligence in pruning out decisions.
• For example, once we know that a point 𝑝𝑝𝑖𝑖 is dominated by another
point 𝑝𝑝𝑗𝑗 , we do not need to use 𝑝𝑝𝑖𝑖 for eliminating other points.
• For example, if 𝑝𝑝𝑖𝑖 < 𝑝𝑝𝑗𝑗 then there is no possibility that 𝑝𝑝𝑖𝑖 will appear
as maximal point.
• So, we should exclude 𝑝𝑝𝑖𝑖 in the future comparisons inside that
algorithm.
• This follows from the fact that dominance relation is transitive.
• If 𝑝𝑝𝑗𝑗 dominates 𝑝𝑝𝑖𝑖 , and 𝑝𝑝𝑖𝑖 dominates 𝑝𝑝ℎ then 𝑝𝑝𝑗𝑗 also dominates
𝑝𝑝ℎ ; 𝑝𝑝𝑖𝑖 is not needed.
𝑝𝑝ℎ < 𝑝𝑝𝑖𝑖 < 𝑝𝑝𝑗𝑗
𝑝𝑝ℎ < 𝑝𝑝𝑗𝑗
• Hence, both 𝑝𝑝ℎ and 𝑝𝑝𝑖𝑖 can be excluded from the future comparisons.
Plane-Sweep Algorithm
• We will sweep a vertical line across the 2D plane from left to right.
• As we sweep, we will build a structure holding the maximal points
lying to the left of the sweep line.
• Sweep from minimum 𝑥𝑥 value to maximum 𝑥𝑥 value.
• And update the maximal set accordingly.
• The maximal set will keep updating while we sweep.
• When the sweep line reaches the rightmost point of 𝑃𝑃, we will have the
complete set of maximal points.
• This approach of solving geometric problems is called the plane sweep.
• Geometric algorithm is another domain of algorithms.
• First, we will sort all the points in ascending order with respect to 𝑥𝑥
coordinate.
(2,5)
Stack
(3,13)
Stack
(4,11)
(3,13)
Stack
(5,1)
(4,11)
(3,13)
Stack
(7,7)
(4,11)
(3,13)
Stack
(9,10)
(4,11)
(3,13)
Stack
(10,5)
(9,10)
(4,11)
(3,13)
Stack
(12,12)
(3,13)
Stack
(13,3)
(12,12)
(3,13)
Stack
(14,10)
(12,12)
(3,13)
Stack
(15,7)
(14,10)
(12,12)
(3,13)
Stack
• At this point, the algorithm will stop, and we get the maximal points
in the stack.
• We can also perform this whole operation on 𝑦𝑦 axis.
• How we got this idea?
• This is the designing of the algorithm.
• Algorithm designing is an art.
• To master it we have to study existing examples, with experience and
by solving different problems in different domains.
Analysis of Plane-Sweep Algorithm
PLANE-SWEEP-MAXIMA (n, Point P[1 … n])
1 sort P in increasing order by x coordinate
2 stack s
3 For i ← 1 to n
4 do
5 while (s is not empty AND s.top().y ≤ P[i].y)
6 s.pop()
7 s.push(P[i])
8 output the contents of stack s
• Sorting takes 𝑂𝑂 𝑛𝑛 log 2 𝑛𝑛 ; we will show this later when we discuss
sorting.
• The for loop executes 𝑛𝑛 times.
• The inner loop (seemingly) could be iterated (𝑛𝑛 − 1) times.
• It seems we still have an 𝑛𝑛(𝑛𝑛 − 1) or 𝑂𝑂(𝑛𝑛2 ) algorithm.
• Is it same as previous 2D maxima algorithm?
• We got confused by simple minded loop counting.
• We pop an element off the stack each time we go through the inner
while-loop.
• It is not possible to pop more elements than are ever pushed on the
stack.
• Therefore, the inner while-loop cannot execute more than 𝑛𝑛 times
over the entire course of the algorithm.
Worst-Case Scenario of Plane-Sweep
Algorithm
• Worst-case scenario is when all the points are same.
• All the points will be pushed to the stack and no point will pop.
• It is not the case that a point will be pushed then pop then pushed again.
• Total 𝑛𝑛 points will be pushed.
• No point will pop.
• The for loop iterates 𝑛𝑛 times and the inner while loop also iterates 𝑛𝑛
time for a total of 2𝑛𝑛 ≈ 𝑂𝑂(𝑛𝑛).
• Combined with the sorting, the runtime of entire plane-sweep
algorithm is:
𝑛𝑛 log 2 𝑛𝑛 + 2𝑛𝑛
= 𝑶𝑶 𝒏𝒏 𝒍𝒍𝒍𝒍𝒍𝒍𝟐𝟐 𝒏𝒏
Comparison
• How much of an improvement is plane-sweep over brute-force?
• Ratio of running times:
𝑛𝑛2
𝑛𝑛 log 2 𝑛𝑛
𝑛𝑛
=
log 2 𝑛𝑛
n log n Ratio (n/log n)
100 7 15
1000 10 100
10000 13 752
100000 17 6021
1000000 20 50171
• For example, if we have n = 1000000,
• The plane-sweep algorithm will take 1 second.
• The brute-force algorithm will take about 14 hours.

Class 5 - 2D Maxima Sweep-Line Algorithm

Uploaded by

Class 5 - 2D Maxima Sweep-Line Algorithm

Uploaded by

2D Maxima Sweep-Line

From Book’s Chapter 2

You might also like