Number Theory Algorithms: Zeph Grunschlag

Number Theory Algorithms
Zeph Grunschlag
Copyright Zeph Grunschlag, 2001-2002.
Announcements
Midterm coming up on Monday Practice midterms and study guide available from the Homework page. Will add last semesters midterm 1 as well after class. Midterm Review: Probably Sunday night. Lookout for announcement. MIDTERM 2 DATE: Considering moving it from Monday 4/8 to Wednesday 4/10. Please tell me ASAP if this is problematic.
L11 2
Agenda
Euclidean Algorithm for GCD Number Systems
Decimal numbers (base-10) Binary numbers (base-2)
Ones complement Twos complement
General base-b number systems
Arithmetic Algorithms
Addition Multiplication Subtraction 1s and 2s complement
L11 3
Euclidean Algorithm
m,n
Euclidean Algorithm
gcd(m,n)
integer euclid(pos. integer m, pos. integer n) x = m, y = n while(y > 0) r = x mod y x=y y=r return x
L11 4
Euclidean Algorithm. Example

gcd(33,77):
Step 0
r = x mod y
-
x
33
y
77
L11

gcd(33,77):
Step 0 1
r = x mod y
-
x
33 77
y
77 33
33 mod 77 = 33
L11

gcd(33,77):
Step 0 1 2
r = x mod y
-
x
33 77 33
y
77 33 11
33 mod 77 = 33 77 mod 33 = 11
L11

gcd(33,77):
Step 0 1 2 3
L11
r = x mod y
-
x
33 77 33 11
y
77 33 11 0
8
33 mod 77 = 33 77 mod 33 = 11 33 mod 11 =0

gcd(244,117):
Step 0
r = x mod y
-
x
244
y
117
L11

gcd(244,117):
Step 0 1
r = x mod y
-
x
244 117
y
117 10
244 mod 117 = 10
L11
10

gcd(244,117):
Step 0 1 2
r = x mod y
-
x
244 117 10
y
117 10 7
244 mod 117 = 10 117 mod 10 = 7
L11
11

gcd(244,117):
Step 0 1 2 3
r = x mod y
-
x
244 117 10 7
y
117 10 7 3
244 mod 117 = 10 117 mod 10 = 7 10 mod 7 = 3
L11
12

gcd(244,117):
Step 0 1 2 3 4
r = x mod y
-
x
244 117 10 7 3
y
117 10 7 3 1
244 mod 117 = 10 117 mod 10 = 7 10 mod 7 = 3 7 mod 3 = 1
L11
13

gcd(244,117): Step 0 1 2 3 4 5 By definition
L11
r = x mod y
-
x
244 117 10 7 3 1
y
117 10 7 3 1 0
14
244 mod 117 = 10 117 mod 10 = 7 10 mod 7 = 3 7 mod 3 = 1 3 mod 1=0
244 and 117 are rel. prime.
Euclidean Algorithm Correctness

The reason that Euclidean algorithm works is gcd(x,y ) is not changed from line to line. If x, y denote the next values of x , y then: gcd(x,y) = gcd(y, x mod y) (the useful fact) = gcd(y, x + qy) = gcd(y, x ) (subtract y -multiple) = gcd(x,y)
L11 15
Euclidean Algorithm Running Time

EX: Compute the asymptotic running time of the Euclidean algorithm in terms of the number of mod operations:
L11
16

Assuming mod operation is O (1): integer euclid(m, n) x = m, y = n while( y > 0) r = x mod y x=y y=r return x
L11
O (1) + ? ( O (1) + O (1) + O (1) + O (1) ) + O (1) = ? O(1)

17
Where ? is the number of while-loop iterations.

Facts: (x = next value of x, etc. ) 1. x can only be less than y at very beginning of algorithm once x > y, x = y > y = x mod y 2. When x > y, two iterations of while loop guarantee that new x is < original x because x = y = x mod y. Two cases:
I. II.
y>x y! x
x mod y = x y < x x mod y < y ! x

18
L11

(1&2) After first iteration, size of x decreases by factor > 2 every two iterations. I.e. after 2m+1 iterations, x < original_x / 2m Q: When in terms of m does this process terminate?
L11 19

After 2m+1 steps, x < original_x / 2m A: While-loop exits when y is 0, which is right before would have gotten x = 0. Exiting while-loop happens when 2m > original_x, so definitely by: m = log2 ( original_x ) Therefore running time of algorithm is: O(2m+1) = O(m) = O (log2 (max(a,b)) )
L11 20

Measuring input size in terms of n = number of digits of max(a,b): n = (log10 (max(a,b)) ) = (log2 (max(a,b)) ) Therefore running time of algorithm is: O(log2 (max(a,b)) ) = O(n) (we assumed naively that mod is an O(1) operation, so this estimate only holds for fixed-size integers such as ints and longs)
L11 21
Number Systems
Q: What does the string of symbols 2134 really mean as a number and why?
L11
22
Number Systems
A: 2 thousands 1 hundreds 3 tens and 4 = 2 103 + 1 102 + 3 101 + 4 100 But on the planet Ogg, the intelligent life forms have only one arm with 5 fingers.
L11
23
Number Systems
So on Ogg, numbers are counted base 5. I.e. on Ogg 2134 means: 2 53 + 1 52 + 3 51 + 4 50 To distinguish between these systems, subscripts are used: (2134)10 for Earth (2134)5 for Ogg
L11
24
Number Systems
DEF: A base b number is a string of symbols u = ak ak-1 ak-2 a2 a1 a0 With the ai taken from {0,1,2,3,,b-2,b-1}. The string of symbols u represents the number (u )b = ak bk + ak-1 bk-1 + . . . + a1 b + a0 NOTE: When b > 10, run out of decimal number symbols so after 7, 8, 9 use capital letters, starting from A, B, C, EG: base-2 (binary) 101, 00010 base-8 (octal ) 74, 0472 base-16 (hexadecimal ) 12F, ABCD L11 25 Q: Compute the base 10 version of these.
Number Systems
A: base-2 (binary) 101, 00010 (101)2 = 1 22 + 0 21 + 1 20 = 5 (00010)2 = 0(24+23+22+20) + 121 = 2 base-8 (octal ) 74, 0472 (74)8 = 7 81 + 4 80 = 60 (0472)8 = 4 82 + 7 81 + 2 80 = 314 base-16 (hexadecimal ) 12F, ABCD (12F)16 = 1162+2161+15160 = 303 (ABCD)16=10163+11162+12161+13160 L11 26 = 43981
Number Systems
The binary and hexadecimal systems are especially important in computer science since binary is the most natural system for bit-strings and hexadecimal for bytestrings (1 byte = 2 hexadecimals) EG in HTML: <font color="ff00ff"> Nice Color </font> Q: What color will this become?
L11 27
Number Systems
A: "ff00ff" represents the rgb value: The first byte is for redness, the second byte is for green-ness, and the last for blue-ness. The HTML command above specifies that there should be 1516 + 15 = 255 redness and blueness values, but no green-ness at all. 255 is the biggest possible byte value so this is bright purple.
L11 28
Number Systems Reverse Conversion

Would also like to convert arbitrary decimal numbers into various bases, (without using calculator-function [see Windows Calculator]). EG: Back at Ogg. Convert 646 to base-5. Try to do all operations base-5: (646)10 = (11)5(20)52 + (4)5(20)5+ (11)5 = (11)5(400)5 + (130)5+ (11)5 = (4400)5 + (141)5 = (10041)5 Thinking like an Oggian hurts brain too much
L11 29

Systematic approach to converting bases. Given an integer n and a base b find the string u such that (u )b = n. Pseudocode: string represent(pos. integer n, pos. integer b) q = n, i = 0 while( q > 0 ) ui = q mod b q = q/b i = i +1 L11return ui ui-1 ui-2 u2 u1 u0 30

EG: Convert 646 to Oggian (base-5):
i ui = q mod b q = q/b 646
L11
31

i 0 ui = q mod b 646 mod 5 = 1 q = q/b 646 646/5=129
L11
32

i 0 1 ui = q mod b 646 mod 5 = 1 129 mod 5 = 4 q = q/b 646 646/5=129 129/5=25
L11
33

i 0 1 2 ui = q mod b 646 mod 5 = 1 129 mod 5 = 4 25 mod 5 = 0 q = q/b 646 646/5=129 129/5=25 25/5=5
L11
34

i 0 1 2 3 ui = q mod b 646 mod 5 = 1 129 mod 5 = 4 25 mod 5 = 0 5 mod 5 = 0 q = q/b 646 646/5=129 129/5=25 25/5=5 5/5= 1
L11
35

i 0 1 2 3 4
L11
ui = q mod b 646 mod 5 = 1 129 mod 5 = 4 25 mod 5 = 0 5 mod 5 = 0 1 mod 5 = 1
Reading last column in reverse: 10041
q = q/b 646 646/5=129 129/5=25 25/5=5 5/5= 1 1/5=0

36
Number Systems Blackboard Exercise

Some number-theory facts are base-dependent. For example First-Grade Teachers Rule: A base-10 number is divisible by 3 iff the sum of its digits are. Formally, let n = (uk uk-1 uk-2 u2 u1 u0)10. Then:
EG: 3|12135 because 3|(1+2+1+3+5 = 12)
k n mod 3 = ui mod 3 i =0
L11
37
Arithmetical Algorithms
Lets write down some familiar arithmetical algorithms. Conveniently, they run the same in any number base. In some cases, such as addition, there are asymptotically faster approaches, but these are the simplest procedures and tend to be fastest for relatively small (e.g. < 1000 bits) number sizes.
L11 38
Arithmetical Algorithms Addition

Numbers are added from least significant digit to most, while carrying any overflow resulting from adding a column:
base-10 Carry:
1 1 1 1
base-16
x
+y
L11
7 + 1 2 0
4 9 3
6 0 7
3 9 2 + 1
A C 6
4 B F
F 0 F
0 9 9
39
Arithmetical Algorithms Addition of Positive Numbers

string add(strings xk xk-1x1x0, yk yk-1y1y0 , int base) carry = 0, xk+1 = yk+1 = 0 for(i = 0 to k+1) digitSum = carry + xi + yi zi = digitSum mod base carry = digitSum /base return zk+1zk zk-1z1z0
L11
40
1s Complement 2s Complement
The binary number system makes some operations especially simple and efficient under certain representations. Two such representations are
1s complement 2s complement
Each makes subtraction much simpler. Each has disadvantage that number length is pre-determined. Whats actually done in practice.
L11 41
1s Complement
Fix k bits. (EG, k = 8 for bytes) Represent numbers with |x | < 2k-1 Most significant bit tells the sign
0 positive 1 negative
Positive numbers the same as standard binary expansion Negative numbers gotten by taking the boolean complement, hence nomenclature
L11
42
1s Complement Examples
k = 8: 00010010 represents 18 11101101 represents -18 Notice that adding these two representations as usual results in the number 11111111, which by definition is negative 00000000 or 0. This leads to: Guess: adding numbers with mixed sign is the same as adding positive numbers Unfortunate: 0 is not represented uniquely
L11 43
1s Complement Addition
Addition is the same as usual binary addition except: if the final carry is 1, cycle the carry to the least significant digit: 00010010 represents 18, 11110011 represents -12 Sum 00000110 represents 6: Carry: x +y pre-sum overflow answer L11
0 1
0 1
0 1
1 1
0 0
0 0
1 1
0 1
44
0 1
0 1
0 1
1 1
0 0
0 0
1 1
0 1 1
45
1
0 1
0 1
0 1
1 1
0 0
0 0
1 1 0
0 1 1
46
1
0 1
0 1
0 1
1 1
0 0
0 0 1
1 1 0
0 1 1
47
1
0 1
0 1
0 1
1 1
0 0 0
0 0 1
1 1 0
0 1 1
48
1 1
0 1
0 1
0 1
1 1 0
0 0 0
0 0 1
1 1 0
0 1 1
49
1 1 1
0 1
0 1
0 1 0
1 1 0
0 0 0
0 0 1
1 1 0
0 1 1
50
1 1 1 1
0 1
0 1 0
0 1 0
1 1 0
0 0 0
0 0 1
1 1 0
0 1 1
51
1 1 1 1
0 1 0
0 1 0
0 1 0
1 1 0
0 0 0
0 0 1
1 1 0
0 1 1
52
1 1 1 1
0 1 0
0 1 0
0 1 0
1 1 0
0 0 0
0 0 1
1 1 0
0 1 1
1
53
1 1 1 1
0 1 0
0 1 0
0 1 0
1 1 0
0 0 0
0 0 1
1 1 0
1
0 1 1
1
54
1 1 1 1
0 1 0
0 1 0
0 1 0
1 1 0
0 0 0
0 0 1
1 1 0
1
0 1 1
1
55
1 1 1 1
0 1 0
0 1 0
0 1 0
1 1 0
0 0 0
0 0 1 1
1 1 0
1
0 1 1
1
56
1 1 1 1
0 1 0 0
0 1 0 0
0 1 0 0
1 1 0 0
0 0 0 0
0 0 1 1
1 1 0
1
0 1 1
1
57
2s Complement
Fixes the non-uniqueness of zero problem Adding mixed signs still easy No cycle overflow Javas approach (under the hood) Same fixed length k , sign convention, and definition of positive numbers as with 1s complement Represent numbers with -2(k-1) x < 2(k-1)
EG. Javas byte ranges from -128 to +127
L11 58
2s Complement
Negative numbers: 1. Compute 1s complement 2. Add 1
Not as simple as negating 1s complement
Summarize: -x = x + 1. EG: 00010010 represents 18, 11101101 + 1 = 11101110 represents -18. Add together: 00000000 (over-flow is lost). Q: What are the ranges of Javas 32-bit int and 64-bit long? (All of Javas integer types use 2s complement)
L11 59
2s Complement
A:
1) 32-bit ints: Largest int =

011111.1 = 231-1 = 2,147,483,647 Smallest int = 100000.0 = -231 = -2,147,483,648 2) 64-bit longs: Largest long = 011111.1 = 263-1 = 9,223,372,036,854,775,807 Smallest int = 100000.0 = -262 = -9,223,372,036,854,775,808
L11 60
Addition is the same as usual binary addition no exceptions!! 11101110 represents -18, 11110100 represents -12 Sum 11111010 represents -30:
1 Carry: x 1 1 1 1 1
+y
1 1
1 1 1
1 1 1
0 1 0
1 0 0
1 1 0
1 0 1
0 0 0
As a final check take the negative to see if get 30: (11100010+1) = (00011101+1) = 00011110. YES!
L11 61
Arithmetical Algorithms Positive Binary Multiplication

Long multiplication simplifies in binary because multiplying by 2k amounts to left-shifting k-places (<<k), and each time multiply either by 02k or 12k. EG:
x y 1(x<<0) 0(x<<1) 0(x<<2) 1(x<<3) Add rows:
1 1
0 0
1 0
1 1
L11
62

x y 1(x<<0) 0(x<<1) 0(x<<2) 1(x<<3) Add rows:
1 1 1
0 0 0
1 0 1
1 1 1
L11
63

x y 1(x<<0) 0(x<<1) 0(x<<2) 1(x<<3) Add rows:
1 1 1 0
0 0 0 0
1 0 1 0
1 1 1
L11
64

x y 1(x<<0) 0(x<<1) 0(x<<2) 1(x<<3) Add rows:
0 0
1 1 1 0 0
0 0 0 0 0
1 0 1 0
1 1 1
L11
65

x y 1(x<<0) 0(x<<1) 0(x<<2) 1(x<<3) Add rows:
0 0
0 0 1
1 1 1 0 0 1
0 0 0 0 0
1 0 1 0
1 1 1
L11
66

x y 1(x<<0) 0(x<<1) 0(x<<2) 1(x<<3) Add rows:
1 1
0 0 1
0 0 1 0
1 1 1 0 0 1 0
0 0 0 0 0 0
1 0 1 0
1 1 1
1
67
L11
Arithmetical Algorithms Binary Multiplication

bitstring multiply(bitstrings xk xk-1x1x0, yk yk-1y1y0) x = xk xk-1x1x0 p = 0 // the partial product for(i = 0 to k+1) if(yi == 1) p = add(p , x << i ) // prev. algorithm return p
L11
68

Number Theory Algorithms: Zeph Grunschlag

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Number Theory Algorithms: Zeph Grunschlag

Uploaded by

Copyright:

Available Formats

Number Theory Algorithms

Copyright Zeph Grunschlag, 2001-2002.

General base-b number systems

Euclidean Algorithm. Example

Euclidean Algorithm. Example

Euclidean Algorithm. Example

Euclidean Algorithm. Example

33 mod 77 = 33 77 mod 33 = 11 33 mod 11 =0

Euclidean Algorithm. Example

Euclidean Algorithm. Example

244 mod 117 = 10

Euclidean Algorithm. Example

244 mod 117 = 10 117 mod 10 = 7

Euclidean Algorithm. Example

244 mod 117 = 10 117 mod 10 = 7 10 mod 7 = 3

Euclidean Algorithm. Example

244 mod 117 = 10 117 mod 10 = 7 10 mod 7 = 3 7 mod 3 = 1

Euclidean Algorithm. Example

244 mod 117 = 10 117 mod 10 = 7 10 mod 7 = 3 7 mod 3 = 1 3 mod 1=0

244 and 117 are rel. prime.

Euclidean Algorithm Correctness

Euclidean Algorithm Running Time

Euclidean Algorithm Running Time

O (1) + ? ( O (1) + O (1) + O (1) + O (1) ) + O (1) = ? O(1)

Where ? is the number of while-loop iterations.

Euclidean Algorithm Running Time

x mod y = x y < x x mod y < y ! x

Euclidean Algorithm Running Time

Euclidean Algorithm Running Time

Euclidean Algorithm Running Time

Number Systems Reverse Conversion

Number Systems Reverse Conversion

Number Systems Reverse Conversion

Number Systems Reverse Conversion

Number Systems Reverse Conversion

Number Systems Reverse Conversion

Number Systems Reverse Conversion

Number Systems Reverse Conversion

ui = q mod b 646 mod 5 = 1 129 mod 5 = 4 25 mod 5 = 0 5 mod 5 = 0 1 mod 5 = 1

Reading last column in reverse: 10041

q = q/b 646 646/5=129 129/5=25 25/5=5 5/5= 1 1/5=0

Number Systems Blackboard Exercise

EG: 3|12135 because 3|(1+2+1+3+5 = 12)

Arithmetical Algorithms Addition

Arithmetical Algorithms Addition of Positive Numbers

1) 32-bit ints: Largest int =

Arithmetical Algorithms Positive Binary Multiplication

x y 1(x<<0) 0(x<<1) 0(x<<2) 1(x<<3) Add rows:

Arithmetical Algorithms Positive Binary Multiplication

x y 1(x<<0) 0(x<<1) 0(x<<2) 1(x<<3) Add rows:

Arithmetical Algorithms Positive Binary Multiplication

x y 1(x<<0) 0(x<<1) 0(x<<2) 1(x<<3) Add rows:

Arithmetical Algorithms Positive Binary Multiplication

x y 1(x<<0) 0(x<<1) 0(x<<2) 1(x<<3) Add rows:

Arithmetical Algorithms Positive Binary Multiplication

x y 1(x<<0) 0(x<<1) 0(x<<2) 1(x<<3) Add rows:

Arithmetical Algorithms Positive Binary Multiplication

x y 1(x<<0) 0(x<<1) 0(x<<2) 1(x<<3) Add rows:

Arithmetical Algorithms Binary Multiplication

You might also like