Kevin's Social Blog: December 2012

Notes

1.1 Algorithms

Algorithm = Inputs -> Outputs (Specific procedure)

Computational problem = desired I/O relationship (general)...an algorithm defines a specific relationship

Nondecreasing = same or greater

Define computational problem by specifiying I/O

Instance = One input

Correct algorithm = halts on all inputs with correct output

Correct algorithm solves a problem

Specification = Precise description of guts

"Incorrect algorithms can sometimes be useful"

Problems solved by algorithms:

Determining sequences (Human Genome Project)
Finding good routes; quickly find Internet pages
Negotiate and exchange electronic commerce
Allocate scarce resources

2-D -> 3-D mapping?

Given = inputs

Wish to find = outputs

Two common characteristics of algorithms:

Many candidate solutions
Practical applications

Data structure = store and organize data to facilitate access and modifications

Problems without published algorithms

Efficiency

Interesting NP-complete properties:

it is unknown whether or not efficient algorithms exist
If an efficient algorithm exists for any one of them, then efficient algorithms exist for all of them
Several NP-complete problems are similar, but not identical, to problems for which we do not know of efficient algorithms

Real world NP-complete problems arise in real world

Traveling salesman

1.2 Algorithms as a technology

Terminates with correct answer

What if infinite speed and free memory?

Good software engineering practice = well designed and documented

Bounded resources (like computing time) should be used wisely

Algorithms are greater than hardware and software considerations

Insertion sort

Merge sort

Constant factors versus input sizes

Running time

Crossover point

Algorithms are important

Algorithms = core of contemporary technology

Algorithm knowledge technique = truly skilled programmer

Definitions

Concave polygon (one of the angles surpasses 180-degrees)
Permutation (a rearranging of terms)
Linear programming (Linear programming (optimization) is a specific case of mathematical programming (mathematical optimization).
Graph
Product
Vertex
Associative
Dynamic programming (a method for solving complex problems by breaking them down into simpler subproblems...The word dynamic was chosen by Bellman to capture the time-varying aspect of the problems, and because it sounded impressive.[3] The word programming referred to the use of the method to find an optimal program, in the sense of a military schedule for training or logistics. This usage is the same as that in the phrases linear programming and mathematical programming, a synonym for mathematical optimization.)
Constant not dependent on n
Pipelining (a set of data processing elements connected in series, so that the output of one element is the input of the next one)
Superscalar (a form of parallelism called instruction level parallelism within a single processor. It therefore allows faster CPU throughput than would otherwise be possible at a given clock rate. A superscalar processor executes more than one instruction during a clock cycle by simultaneously dispatching multiple instructions to redundant functional units on the processor.)

Exercises 1.1

1.1-1 Give a real world example in which one of the following computational problems appears: sorting, determining the best order for multiplying matrices, or finding the convex hull.

Sorting

Sorting files in a directory by creation date
Sorting various paper denominations of money
Sorting tax return forms by form ID number

Determining the best order for multiplying matrices

Finding the convex hull

Identifying area of forest burned by fire

1.1-2 Other than speed, what other measures of efficiency might one use in a real-world setting?

Storage requirements during runtime
Power consumption
Size of hardware
Size of software at rest

1.1-3 Select a data structure you have seen previously and discuss its strengths and limitations.
Stack

Strengths

Simple to implement: two operations, push and pop
Push and pop operate in constant time

Limitations

Not elegant way to traverse elements, since have to pop and store elsewhere in order to inspect and preserve order

1.1-4 How are the shortest-path and traveling-salesman problems given above similar? How are they different?

Same

Both want shortest path, per constraints
Both seem like NP-complete problems

Differences

Traveling-salesman requires begin and end at same node
Shortest-path goes point-to-point

1.1-5 Come up with a real-world problem in which only the best solution will do. Then come up with one in which a solution that is "approximately" the best is good enough.

Only the best solution

Note: best means no cutting corners, not necessarily "perfection"
Everything comes down to acceptable tolerances, in the real world
Typically, anything involving human safety in life and death situations

Space station airlocks (opening and closing)
Airplane auto-pilot systems
Traffic control systems (for example, traffic lights)

Financial transactions

Approximately the best is good enough

Pretty much everything else ; o )
Point-to-point driving when time not of the essence...probably OK to get reasonably close to a location and then figure out parking, eating, and so forth
Pouring a beer...OK to have a bit of foam at the top

Exercises 1.2

1.2-1 Give an example of an application that requires algorithmic content at the application level, and discuss the function of the algorithms involved.

Section 1.1 of this book defines an algorithm as "any well-defined computational procedure that takes some value, or set of values, as input and produces some value, or set of values, as output." Given this broad definition, I suggest the example of a web-based SQL formatter. As input, it takes a sequence of characters representing a SQL query. As output, it produces formatted SQL code, per constraints specified by the user. The algorithm functions as a set of rules which transform the input into the output based on the constraints specified by the user. For example, if the user selects "DB2" SQL as the input and C# as the output, the algorithm logic would transform the inputs into outputs conforming to C# rules.

1.2-2 Suppose we are comparing implementations of insertion sort and merge sort on the same machine. For inputs of size n, insertion sort runs in 8n^2 steps, while merge sort runs in 64nlgn steps. For which values of n does insertion sort beat merge sort?

We wish to find the greatest value of n for which 8n^2 < 64nlgn. Since this seems to represent a transcendental function, I have elected to plug and chug to arrive at the answer. I used OpenOffice.org Calc to calculate the results below:

When n = 2, we have 8(4) < 64(2)(1) => 32 < 128, which is true.

When n = 4, we have 8(16) < 64(4)(2) => 128 < 512, which is true.

and so forth, until n = 44, when we have 15,488 < 15373.8, which is false

Therefore, in this particular implementation, insertion sort beats merge sort for values of 2 <= n <= 43.

1.2-3 What is the smallest value of n such that an algorithm whose running time is 100n^2 runs faster than an algorithm whose running time is 2^n on the same machine?

We wish to find the smallest value of n for which 100n^2 < 2^n. Since this seems to represent a transcendental function, I have elected to plug and chug to arrive at the answer. I used OpenOffice.org Calc to calculate the results below:

When n = 1, we have 100(1) < 2^1 => 100 < 2, which is false.

When n = 2, we have 100(4) < 2^2=> 400 < 4, which is false.

and so forth, until n = 15, when we have 22,500 < 32,768, which is true

Therefore, in this particular implementation, the algorithm with running time 100(n^2) runs faster than the algorithm with running time 2^n for values of n >= 15.

Problems

1-1 Comparison of running times
For each function f(n) and time t in the following table, determine the largest size n of a problem that can be solved in time t, assuming that the algorithm to solve the problem takes f(n) microseconds.

Note: I found this problem statement confusing, initially. The authors treat the computational problem and algorithm as a black box. The black box can solve an input of n items in f(n) microseconds. The authors ask the reader to calculate the largest size n the black box can solve on or before various times (t). For example, for f(n) = n, the black box can solve an input of 1 item in f(1) = 1 microsecond and 2 items in f(2) = 2 microseconds.

We first convert the times in the header row into microseconds (one microsecond = 1/1,000,000 of a second):

1 second = 1 * 10^6 or 1,000,000 microseconds
1 minute = 6 * 10^7 or 60,000,000 microseconds
1 hour = 3.6 * 10^9 or 3,600,000,000 microseconds
1 day = 8.64 * 10^10 or 86,400,000,000 microseconds
1 month = 2.592 * 10^12 or 2,592,000,000,000 microseconds (assuming 30 days per month, on average)
1 year = 3.15576 * 10^13 or 31,557,600,000,000 microseconds (assuming 365.25 days per year)
1 century = 3.15576 * 10^15 or 3,155,760,000,000,000 microseconds

Second, for each cell, we need to calculate the largest size n the black box can solve on or before the specified number of microseconds:

	Notes	1 second	1 minute	1 hour	1 day	1 month (assume 30 days, on average)	1 year (assume 365.25 days)	1 century
lg(n)	Put both sides on base 2; this converts 2^lg(n) to n.	2^(1 * 10^6)	2^(6 * 10^7)	2^(3.6 * 10^9)	2^(8.64 * 10^10)	2^(2.592 * 10^12)	2^(3.15576 * 10^13)	2^(3.15576 * 10^15)
sqrt(n)	Squaring both sides solves for n.	(1 * 10^6)^2 = 1 * 10^12	(6 * 10^7)^2 = 3.6 * 10^15	(3.6 * 10^9)^2 = 1.296 * 10^19	(8.64 * 10^10)^2 = 7.46496 * 10^21	(2.592 * 10^12)^2 = 6.718464 * 10^24	(3.15576 * 10^13)^2 = 9.9588211776×10^26	(3.15576 * 10^15)^2 = 9.9588211776×10^30
n		1 * 10^6	6 * 10^7	3.6 * 10^9	8.64 * 10^10	2.592 * 10^12	3.15576 * 10^13	3.15576 * 10^15
nlg(n)	Put both sides as an exponent on base 2; since nlg(n) = lg(n^n), this converts 2^(lg(n^n)) to n^n. Transcendental...so we have to do things the hard way.	6.2746 * 10^4 (link)	2.801417 * 10^6 (link to get approximation, then manual refinement)	1.33378058 * 10^8 (link to get approximation, then manual refinement)	2.755147513 * 10^9 (link to get approximation, then manual refinement)	7.1870856404 * 10^10 (link to get approximation, then manual refinement)	7.98161 * 10^11 (link to get approximation, then manual refinement)	6.86565x10^13 (link to get approximation, then manual refinement)
n^2	Taking the square root of both sides solves for n.	sqrt(1 * 10^6) = 1 * 10^3 or 1,000	sqrt(6 * 10^7) = 7.745966692 * 10^3 or 7,745	sqrt(3.6 * 10^9) = 6 * 10^4 or 60,000	sqrt(8.64 * 10^10) = 2.939387691 * 10^5 or 293,938	sqrt(2.592 * 10^12) = 1.609968944 * 10^6 or 1,609,968	sqrt(3.15576 * 10^13) = 5.617615152357805 * 10^6 or 5,617,615	sqrt(3.15576 * 10^15) = 5.617615152357 × 10^7 or 56,176,151
n^3	Taking the cube root of both sides solves for n.	cuberoot(1 * 10^6) = 1 * 10^2 or 100	cuberoot(6 * 10^7) = 3.914867641×10^2 or 391	cuberoot(3.6 * 10^9) = 1.532618865×10^3 or 1,532	cuberoot(8.64 * 10^10) = 4.420837798×10^3 or 4,420	cuberoot(2.592 * 10^12) = 1.373657091×10^4 or 13,736	cuberoot(3.15576 * 10^13) = 3.1601×10^4 or 31,601	cuberoot(3.15576 * 10^15) = 1.46679335×10^5 or 146,679
2^n	Taking lg of both sides solves for n.	log(1 * 10^6) / log(2) = 1.993156857×10^1 or 19	log(6 * 10^7) / log(2) = 2.583845916×10^1 or 25	log(3.6 * 10^9) / log(2) = 3.174534976×10^1 or 31	log(8.64 * 10^10) / log(2) = 3.633031226×10^1 or 36	log(2.592 * 10^12) / log(2) = 4.123720286×10^1 or 41	log(3.15576 * 10^13) / log(2) = 4.484305×10^1 or 44.84305	log(3.15576 * 10^15) / log(2) = 5.1486909×10^1 or 51
n!	Plug and chug...Wolfram Alpha helps	9	11	12	13	15	16	17