MATH 470 Lecture 2

From Notes
Jump to navigation Jump to search

« previous | Thursday, January 17, 2013 | next »


Fundamental Theorem of Arithmatic

Any n can be written as n=i=1kpiai for pairwise distinct primes p1,,pk and a1,,ak


Corollary

GCD(p1a1pkak,p1b1p1min(a1,b1)pkmin(ak,bk)


Corollary

If a,x,n with GCD(a,n) = 1 and n(ax),then<math>nx


Extended Euclidean Algorithm

Input: a,bZ
Output: g:=GCD(a,b) and (x,y)2 with ax+by=g

  1. Form matrix [1001ab]
  2. Use elementary column operations (only allowing divisions by ≠ 1) to reduce to [uxvy0g]
  3. You have your answers g, x, and y

Example

Solve 173x=3(mod256)

Rewrite as 173x+256y=3. Find x and y such that 173x+256y=1

[1001173256][110117383][3121783][33422376][3734252316][372562517310][256371732501]

Therefore, all of the following are true:

  • 173(37)+256(25)=1
  • 17337=1(mod256)
  • 1173=37(mod256)


1173173x=3(1173)(mod256)=337(mod256)=111(mod256)

Complexity: Lame's Theorem

Number of column operations needed is ≤ 5 × number of digits in smaller number

Theorem on Congruences

Given any a,b,n, the congruence ax=b(modn) has exactly g:=GCD(a,n) solutions when gb and no solutions otherwise

Example

27x=18(mod45) has solutions {4,9,14,19,24,29,34,39,44}.

How would we have known?

  1. find GCD(a,n) via Extended Euclidean Algorithm (GCD(27,45)=9; 27(2)+45(1)=9)
  2. divide out equation by g (3x=2(mod5))
  3. solve equation above (x=4)
  4. solutions are: {x,x+ng,,x+(g1)ng}


Proof

First note that gb implies no solutions.

Indeed ax=b(modn) implies ax+ny=b for some x,y.

g(ax+ny), so gb leads to a contradiction.

For example, 2x=1(mod8) cannot be solved since 2x+8y is even and 1 is odd.


Let's assume gb

if g{a,b,n}, and x is any solution to ax=b(modn), so is x+i(ng) for any i{0,,g1}


Indeed, a(x+i(ng))ax+i(ang)b+in(ag)b(modn)


Moreover, ng,2ng,,(g1)ng are clearly distinct mod n

It's enough to consider g=1

To conclude, axb(modn)ax+ny=b has only one solution.

Now suppose if there are two solutions (x1,y1) and (x2,y2), then we get

ax1+ny1=bax2+ny2=ba(x1x2)+n(y1y2)=0

GCD(a,n)=1, so since n divides the RHS, it also divides the LHS

Therefore, n(x1x2)x1x2(modn)

Q.E.D.


CAS Stuff

Maple

> (* Matrices *)
> M := matrix([[a,b,c],[d,e,f],[g,h,i]]);


Sage

sage: # Affine Ciphers == of the form: f(x) = a*x + b (mod 26)
sage: A = AffineCryptosystem(AlphabeticStrings());
sage: P = A.encoding("Hey! How's it going?");
  => HEYHOWSITGOING
sage: a,b = (9,13);
sage: cipher = A.enciphering(a,b,P);
  => YXVYJDTHCPJHAP


Solving Linear Systems

(mod an integer)


{2x+3y=1(mod26)7x+14y=2(mod26)

Put into an augmented matrix

[2317142]

First operation would be to subtract 7/2 times the first row from the second row... TROUBLE! What's 1/2 mod 26? (undefined)

2 options

  1. Do gaussian elimination without dividing by two (similar to E.E.A.)
  2. Use adjoints


First Option

[2317142][151073]

x+5y=1(mod26)7y=3(mod26)

We can invert -1/7 mod 26:

17=15=11(mod26)

y=311=7(mod26)x=36=16(mod26)

Second Option

Cramer's rule

(x,y)=(|13214||23714|,|2172||23714|)


We only have to invert 7, which is relatively prime to 26, so it is possible.

Theorem

You can solve Ax=b(modn) for any k×k matrix of integers A and any k×1 vector of integers b when GCD(|A|,n)=1