0% found this document useful (0 votes)
58 views6 pages

Math 8201 Homework 1 Fall 2014: Tianyu Tao September 9, 2014

This document provides the solution to the equation 4711x + 3926y = 1 using the Euclidean algorithm. It first uses long division to write 4711 = 3926 * 1 + 785 and 3926 = 785 * 5 + 1, showing that 1 = 3926 * 6 - 4711 * 5. This gives a particular solution of x = 5, y = 6. It then presents the general solution as x = 5 + 3926k, y = 6 + 4711k, where k is an integer. Finally, it reformulates the solution process using row operations on the coefficient matrix to systematically arrive at the general solution.

Uploaded by

Tianyu Tao
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
0% found this document useful (0 votes)
58 views6 pages

Math 8201 Homework 1 Fall 2014: Tianyu Tao September 9, 2014

This document provides the solution to the equation 4711x + 3926y = 1 using the Euclidean algorithm. It first uses long division to write 4711 = 3926 * 1 + 785 and 3926 = 785 * 5 + 1, showing that 1 = 3926 * 6 - 4711 * 5. This gives a particular solution of x = 5, y = 6. It then presents the general solution as x = 5 + 3926k, y = 6 + 4711k, where k is an integer. Finally, it reformulates the solution process using row operations on the coefficient matrix to systematically arrive at the general solution.

Uploaded by

Tianyu Tao
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 6

Math 8201 Homework 1 Fall 2014

Tianyu Tao
September 9, 2014
Problem 1:

0.2 Exx 1.

Proof: (a) We want to show that given x1 , x2 2 X such that gf (x1 ) = g(f (x1 )) =
g(f (x2 )) = gf (x2 ), it will follow that x1 = x2 . By assumtion, since g is injective,
we have f (x1 ) = f (x2 ), now again, by the assumption that f is injective, it follows
that x1 = x2 .
(b) Let x1 , x2 2 X be such that f (x1 ) = f (x2 ), we want to show that x1 = x2 .
Now by our choice of x1 , x2 , we have gf (x1 ) = g(f (x1 )) = g(f (x2 )) = gf (x2 ), since
gf is injective, we must have x1 = x2 .
(c) Let z be an arbitrary member of Z, we want to show that we can find an
element x in X such that gf (x) = z. Now since g is surjective, there is some y 2 Y
such that g(y) = z. Similarly, by surjectivity of f , we can find some x 2 X with
f (x) = y, then gf (x) = g(f (x)) = g(y) = z.
(d) Let z be an arbitrary member of Z, we want to show that we can find an
element y in Y such that g(y) = z. Since gf is surjective, we can find x 2 X with
gf (x) = z, let y = f (x) 2 Y , then g(y) = g(f (x)) = gf (x) = z.
(e) For the first example, let f : R ! R be the exponential function: f (x) = ex
for all x 2 R. Then f is injective since if ex1 = ex2 , take natural logarithm of both
sides give x1 = x2 . But f is not surjective since ex > 0 for any x 2 R, the image
of f is all the positive real numbers.
For the second example, let f : R ! R be the following polynomial: f (x) =
x(x 1)(x + 1). Clearly f is not injective since f (1) = f ( 1) = 0. But f is
surjective, as f (x) ! 1 as x ! 1 and f (x) ! 1 as x ! 1 and we can
apply intermediate value theorem for f on intervals of the form [ M, M ] where
M is arbitrary positive number.
(f) If gf is bijective, then it is both injective and surjective, therefore by part
(b) and (d) we immediately conclude that f is injective and g is surjective. But
f is not necessarily surjective and neither g has to be injective. For example, let
X = Z = R 0 and Y = R. Let f : X ! Y be the function f (x) = x3/2 and
let g : Y ! Z be the function g(y) = y 2 , then gf : X ! Z is the function
1

Left Heading

Right Heading

gf : R 0 ! R 0 with gf (x) = x3 which is clearly bijiective. However f is not


surjective, and g is not injective either.

Problem 2:

0.2 Exx 4.

Proof: (i) We have to show both (A [ B) (A) [ (B) and (A [ B)


(A) [ (B).
For the first direction, take t 2 (A [ B), then t = (s) for some s 2 A [ B,
so s 2 A or s 2 B, and hence t = (s) 2 (A) or t 2 (B), which means exactly
t 2 (A) [ (B).
For the reverse direction, take t 2 (A) [ (B). hence t 2 (A) or t 2 (B),
without loss of generality assume t 2 (A), then t = (s) for some s 2 A, certainly
for s we have s 2 A A [ B, therefore t = (s) 2 (A [ B).
(ii) We only need to show (A \ B) (A) \ (B), take t 2 (A \ B),
so t = (s) for some s 2 A \ B. Hence s 2 A and s 2 B, which implies
t = (s) 2 (A) and t 2 (B), this means precisely that t 2 (A) \ (B).
(iii) To see it could happen that there might be some t 2 (A) \ (B) with
t2
/ (A\B), we devise the followin example: let : R ! R be the square function:
(x) = x2 , set A = ( 2, 1) and B = (1, 2), then A \ B = ;, so (A \ B) = ;.
However, (A) = (B) = (1, 4), so (A) \ (B) = (1, 4) 6= ;.

Left Heading
Problem 3:

Right Heading
0.3 Exx 4.

Proof: (i) Suppose we have C is a correspondence from S to T , D from T to U and


E from U to V . Then both (ED)C and E(DC) are correspondence from S to V .
Take (s, u) 2 (ED)C, we want to show that (s, u) 2 E(DC). But (s, u) 2 (ED)C
means that (s, u) 2 S V and there is t 2 T with (s, t) 2 C and (t, u) 2 ED, and
(t, u) 2 ED means there (t, u) 2 T V and there is k 2 U with (t, k) 2 D and
(k, u) 2 E. Thus we have the following:
(s, u) 2 (ED)C ! 9t 2 T 9k 2 U (((s, t) 2 C) ^ ((t, k) 2 D) ^ ((k, u) 2 E))
But, if there is t 2 T, k 2 U with (s, t) 2 C and (t, k) 2 D, this would imply
that (s, k) 2 DC by definition, and the together with the condition (k, u) 2 E, we
have (s, u) 2 (DC)E, therefore we are done in proving (ED)C E(DC).
However, the above unwrapping of definitions is completely reverseble: (s, u) 2
E(DC) would give the same formula and we immediately get E(DC) = (ED)C.
(ii) Suppose C is a correspondence from S to T . Take (s, t) 2 C1S , this means
the existence of s0 2 S with (s, s0 ) 2 1S and (s0 , t) 2 C, however, if (s, s0 ) 2 1S , we
must have s = s0 , thus (s, t) = (s0 , t) 2 C, hence C1S C. For the other direction,
if (s, t) 2 C, take s0 = s and it is clear to see that this s0 makes (s, t) 2 C1S .
Similarly, take (s, t) 2 1T C, then there is some t0 2 T with (s, t0 ) 2 C and
0
(t , t) 2 1T , it then follows t0 = t so (s, t) 2 C and 1T C C. For the other
direction just take t0 = t and we see C 1T C.
Problem 4:

0.3 Exx 6.

Proof: (a) We have to show that the three condition for a equivalence relation for
E holds: (i) aEa (ii) aEb implies bEa and (iii) aEb and bEc implies aEc. In the
following I prefer the notation (a, b) 2 E instead aEb.
For (i), we note 1S E, hence for any a 2 S we have (a, a) 2 1S E.
For (ii), we suppose (a, b) 2 E, we want to show (b, a) 2 E, by the definition of
E, either (a, b) 2 1S or (a, b) 2 (C [C 1 )n for some n. The former case implies a =
b and hence (b, a) = (a, b) 2 E. In the later case, suppose (a, b) 2 (C [ C 1 )n , this
would imply the existence of s1 , . . . , sn 1 such that (a, s1 ), (s1 , s2 ), . . . , (sn 1 , b) 2
C [ C 1 . Now (a, s1 ) 2 C [ C 1 means either (a, s1 ) 2 C or (a, s1 ) 2 C 1 , this is
the same thing as saying either (s1 , a) 2 C 1 or (s1 , a) 2 C or (s1 , a) 2 C [ C 1 ;
similarly, we have (a, s1 ), (s2 , s1 ), . . . , (b, sn 1 ) 2 C [ C 1 , set s0i = sn i for i =
1, 2, . . . , n 1 and we see this means exactly that (b, a) 2 (C [ C 1 )n E.
For (iii), suppose (s, t), (t, u) 2 E, it is clear that (s, u) 2 E if one of (s, t), (t, u) 2
1S for then we have (s, u) = (t, u) or (s, t). So suppose (s, t) 2 (C [ C 1 )n1 and
(t, u) 2 (C [ C 1 )n2 for some n1 , n2 > 1. This implies the existence of s1 , . . . , sn1 1
with (s, s1 ), . . . , (sn1 1 , t) 2 C [ C 1 and t1 , . . . , tn2 1 with (t, t1 ), (tn2 1 , u) 2
3

Left Heading

Right Heading

C [C 1 , set s0i = si for i = 1, . . . , n1 1, s0n1 = t and s0ni +n1 = ti for i = 1, . . . , n2 1.


We see with s0i this means exactly that (t, u) 2 (C [ C 1 )n1 +n2 E.
(b) I will follow the notation in part (a). Now let E 0 be another equivalence
relation on S which contains C, we have to show that E E 0 . That is, we need
to show 1S E 0 and (C [ C 1 )n E 0 for n = 1, 2, . . ..
Since E 0 is an equivalence relationship on S, we must have 1S E 0 by the first
requirement of being an equivalence relationship.
Next we show C [ C 1 E 0 . Since C E 0 , let (a, b) 2 C, by reflexity we have
(b, a) 2 E 0 , hence C 1 E 0 and so C [ C 1 E 0 .
To show for any n we have (C [ C 1 )n E 0 , we proceed by induction: assume (C [ C 1 )m E 0 , consider (s, t) (C [ C 1 )m+1 , then we know there exist
s1 , . . . , sm such that (s, s1 ), . . . , (sm , t) 2 C[C 1 . The fact that (s, s1 ), . . . , (sm 1 , sm )
belong to C [C 1 says exactly that (s, sm ) (C [C 1 )m . By induction hypothesis,
(s, sm ) E 0 , we also know (sm , t) C [ C 1 E 0 by above, thus by transitivity
we have (s, t) E 0 . This shows (C [ C 1 )m E 0 for any m 1.
Therefore we conclude that E = 1S [ (C [ C 1 ) [ (C [ C 1 )2 [ E 0 if E 0
is any equivalence relationship on S that contains C.

Problem 5 : 0.3 Exx 11.


Proof. For any z 2 C , we may write z = rei with r 2 (0, 1) and 2 [0, 2).
We know r = |z| and so g : C ! C is the map g(rei ) = ei . By definition the
equivalence relation Eg on C defined by g is the subset of C C specified by
(z1 , z2 ) 2 Eg if and only if g(z1 ) = g(z2 ). That is Eg = {(r1 ei1 , r2 ei2 ) | 1 = 2 },
in other words z1 Eg z2 if they lie on the same ray emnating from the origin.

Left Heading

Right Heading

Problem 6: Wrriten problem:


Proof. We are finding all integral solutions of the equation 4711x + 3926y = 1.
The solution of this type of problem is from the famaliar Euclidean Algorithm:
First we use division algorithm to write
4711 = 3926 1 + 785
3926 = 785 5 + 1
So 1 = 3926 785 5 = 3926 (4711 3926) 5 = 3926 6 4711 5, we have
thus find a patricular solution x = 5, y = 6, then the general solution is given by
x = 5 + 3926k, y = 6 + 4711k where k 2 Z.
Of course, we want a more neat way to present this solution as suggested in
the problem. Below is the algorithm:

4711 1 0
, R1 = R1 1 R2
3926 0 1

785 1
1
, R2 = R2 5 R1
3926 0 1

785 1
1
, R1 = R1 785 R2
1
5 6

0 3926
4711
1
5
6
The last two entries in the second row give us one particular solution: x = 5, y =
6.
In general, if we have an equation of the form ax + by = 1 (where a, b are
integer constants with gcd(a, b) = 1, suppose a > b). We form the matrix

0
a 1 0
R1
A0 =
=
b 0 1
R20
where R10 = (a, 1, 0) = (10 , 10 , 10 ) and R20 = (b, 0, 1) = (20 , 20 , 20 ), notice that
we have 10 = a 10 + b 10 and 20 = a 20 + b 20 .
Now by Euclidean Algorithm, we may find q0 , q1 , . . . , r0 , r1 , . . . such that:
a = q0 b + r 0
b = q1 r 0 + r 1
r 0 = q2 r 1 + r 2
5

Left Heading

Right Heading

ri = qi+2 ri+1 + ri+2

R1i
with
R2i
i
i
i
i
R1(2)
= (1(2)
, 1(2)
, 1(2)
) is formed, if i is even, we set R1i+1 = R1i and R2i+1 =
R2i
qi R2i ; if i is odd, we set R2i+1 = R2i and R1i+1 = R1i
qi R1i . It is easy to
i
i
i
see that the entries satisfy the relation mentioned above: 1(2)
= a 1(2)
+ b 1(2)
)
for each i. Since by induction if this holds for i, in the case when i is even, by
set up we have 1i+1 = a 1i+1 + b 1i+1 , and 2i+1 = 2i
qi 2i = (a 2i + b 2i )
qi (a 2i + b 2i ) = a( 2i qi 2i ) + b( 2i qi 2i ) = a 2i+1 + b 2i+1 , the case when i is
odd is compeltely similar. With this relation established, we simply recall that by
Euclidean Algorithm, we will get rn = 1 for some n since it has to stop at some
point, and rn = 1n+1 or 2n+1 , then we have 1 = kn+1 = a kn+1 + b kn+1 (with k = 1
or 2), and x = kn+1 , b = kn+1 is the desired particular solution.
We then recursively form Ai for i > 0 as follows: suppose Ai =

You might also like