7/18/22, 9:20 AM 3111 Group Theory

Group Theory (3111)

Christian Hokaj

Friday
Jun 17, 2022 - Sep 16, 2022
7:30 - 9:30 PM ET (4:30 - 6:30 PM PT)

Overview
Week 4 (Jul 8) Class Transcript - Abelian Groups
< Go back to the class overview page
Copyright © AoPS Incorporated. This page is copyrighted material. You can view and print this page for your own use, but you
cannot share the contents of this file with others.
Display all student messages • Show few student messages • Hide student messages
CH328 2022-07-08 19:31:24
We've reached Week 4! How do you feel about the course so far?

cuber 2022-07-08 19:31:50

hard but lots of fun

isont 2022-07-08 19:31:50

pretty good, but it's a bit tough

Hulk23 2022-07-08 19:31:50

Difficult but fun

randomname333 2022-07-08 19:31:50

Awesome! but hard

unicone 2022-07-08 19:31:50

very very exciting!

ad1406 2022-07-08 19:31:50

It's awesome! Now I think of numbers entirely differently

just4fun 2022-07-08 19:31:50

good!

Mathgirl03 2022-07-08 19:31:50

Great!

CH328 2022-07-08 19:32:05

Great!

CH328 2022-07-08 19:32:12

Keep in mind it's completely okay if you find this hard.

CH328 2022-07-08 19:32:38

As a reminder, most people who learn group theory are math or science majors in their 2nd, 3rd or 4th year of college.

CH328 2022-07-08 19:33:03

It's a fantastic subject, but it is very difficult, even for people several years older.

CH328 2022-07-08 19:33:29

I'm really impressed with a lot of the work I've seen in the course. You're all doing great working hard to get as much out of the
course as you can.

CH328 2022-07-08 19:34:03

Keep up your effort and you'll unlock a lot of beautiful mathematics!

CH328 2022-07-08
Typesetting math: 100% 19:34:10
Let's start by introducing two more important bits of abstraction.
https://artofproblemsolving.com/class/3111-grouptheory/transcript/48543 1/39
7/18/22, 9:20 AM 3111 Group Theory

CH328 2022-07-08 19:34:16

Product Groups

CH328 2022-07-08 19:34:35

Recall the additive group of the reals, R = (R, + ). We are comfortable with the Cartesian plane, R × R. This is just the set of
ordered pairs (a, b) where each of a and b is real.

CH328 2022-07-08 19:34:38

Does R × R form a group under some natural operation?

Math5K 2022-07-08 19:35:32

yes

isont 2022-07-08 19:35:32

yes

ad1406 2022-07-08 19:35:32

We could think of something like pairwise addition

Hulk23 2022-07-08 19:35:32

Addition

Leo.Euler 2022-07-08 19:35:32

yes, it does when we define (a, b) + (c, d) = (a + c, b + d)

ajiten 2022-07-08 19:35:32

+

cuber 2022-07-08 19:35:32

it should under addition

YaoAOPS 2022-07-08 19:35:32

addition

HotSoup 2022-07-08 19:35:32

2 2
+ : R → R defined componentwise

just4fun 2022-07-08 19:35:32

yes

CH328 2022-07-08 19:35:42

Right! We just add pairs exactly how you would think to add pairs.

CH328 2022-07-08 19:35:52

Specifically, this set has a binary operation of addition: (a, b) + (a ′ , b ′ ) = (a + a ′ , b + b ′ ).

CH328 2022-07-08 19:35:59

You add component-wise.

CH328 2022-07-08 19:36:01

What are the identity, and what's the inverse of (a, b)?

The_goat 2022-07-08 19:36:50

(-a, -b)

Hulk23 2022-07-08 19:36:50

Identity is (0,0), inverse is (-a,-b)

randomname333 2022-07-08 19:36:50

(0, 0) and (-a, -b)

mbbchess 2022-07-08 19:36:50

(0,0) and (-a,-b)
Typesetting math: 100%
ad1406 2022-07-08 19:36:50

https://artofproblemsolving.com/class/3111-grouptheory/transcript/48543 2/39
7/18/22, 9:20 AM 3111 Group Theory
The identity is (0, 0) and the inverse of (a, b) is ( − a, − b)

Math5K 2022-07-08 19:36:50

the identity is (0,0) and the inverse of (a,b) is (-a,-b)

ScaryDragon 2022-07-08 19:36:50

Identity is (0,0), inverse is (-a,-b)?

ajiten 2022-07-08 19:36:50

(0,0)

(-a,-b)

romans58 2022-07-08 19:36:50

identity is (0, 0), and inverse of (a, b) is ( − a, − b)

HotSoup 2022-07-08 19:36:50

e = (0, 0), (a, b) − 1 = ( − a, − b)

CovalentBondIBDefinition 2022-07-08 19:36:50

identity is the origin and the inverse is -a, -b

just4fun 2022-07-08 19:36:50

(0,0) and (-a. -b)

cuber 2022-07-08 19:36:50

identity is (0,0), inverse is (-a, -b)

ChristiTheMagnificent 2022-07-08 19:36:50

identity is 0, inverse is (-a,-b)

CH328 2022-07-08 19:36:56

The identity is (0, 0), and we have inverses (a, b) − 1 = ( − a, − b).

CH328 2022-07-08 19:37:00

Is our binary operation associative?

Leo.Euler 2022-07-08 19:37:27

yes

Math5K 2022-07-08 19:37:27

yes

Hulk23 2022-07-08 19:37:27

Yes

mbbchess 2022-07-08 19:37:27

Yes

randomname333 2022-07-08 19:37:27

Yes

just4fun 2022-07-08 19:37:27

yes

ajiten 2022-07-08 19:37:27

Yes

romans58 2022-07-08 19:37:27

yup!

HotSoup 2022-07-08 19:37:27

Typesetting math: 100%
yes, because addition is associative

https://artofproblemsolving.com/class/3111-grouptheory/transcript/48543 3/39
7/18/22, 9:20 AM 3111 Group Theory

rav3nclaw 2022-07-08 19:37:27

yes

ad1406 2022-07-08 19:37:27

Yup!

ScaryDragon 2022-07-08 19:37:27

It's addition, so yes

Bean_Burrito 2022-07-08 19:37:27

yes

CH328 2022-07-08 19:37:29

Yep. Associativity follows from associativity in R.

CH328 2022-07-08 19:37:41

This group may be familiar to you, if you've studied some vector algebra: we add pairs like we add vectors.

CH328 2022-07-08 19:37:58

In an analogous way we can define the product of a pair of groups, G × H as the set of pairs (g, h). How do we define the group
operation?

Math5K 2022-07-08 19:39:18

(g,h) x (g',h') = (gxg', hxh')

Hulk23 2022-07-08 19:39:18

(g,h)(g',h') = (g+g',h+h')

HotSoup 2022-07-08 19:39:18

composition componentwise

cuber 2022-07-08 19:39:18

componentwise? If "x" is the operation, then (a,b)x(c,d) = (a x c, b x d)

YaoAOPS 2022-07-08 19:39:18

(g 1, h 1) ⋆ (g 2, h 2) = (g 1 + g 2, h 1 + h 2)

just4fun 2022-07-08 19:39:18

(gg',hh')

ad1406 2022-07-08 19:39:18

If G is the group (G, ⋅ ) and H is the group (H, ⋆ ), then (g, h) ∗ (g ′ , h ′ ) = (g ⋅ g ′ , h ⋆ h ′ )

Leo.Euler 2022-07-08 19:39:18

for the g element, the operation is the same as with G and for the h element, the operation is the same as with H

ChristiTheMagnificent 2022-07-08 19:39:20

(g,h)*(g',h') = (g*g', h*h')

CH328 2022-07-08 19:39:22

The group operation is "componentwise": (g 1, h 1)(g 2, h 2) = (g 1g 2, h 1h 2). Just like with R × R above, all the group axioms follow
directly from those for G and H.

CH328 2022-07-08 19:40:00

When people write this, they sort of just expect you to figure out what g 1g 2 means: it must be the product of g 1 g 2 using the group
operation in G.

CH328 2022-07-08 19:40:16

Similarly, h 1h 2 is computed using the group operation of H.

CH328 2022-07-08 19:40:22

What's the identity?

Math5K math:
Typesetting 2022-07-08
100% 19:41:22
(identity in G, identity in H)
https://artofproblemsolving.com/class/3111-grouptheory/transcript/48543 4/39
7/18/22, 9:20 AM 3111 Group Theory

ad1406 2022-07-08 19:41:22

(e G, e H)

Leo.Euler 2022-07-08 19:41:22

(e G, e H)

Hulk23 2022-07-08 19:41:22

(e g, e h)

just4fun 2022-07-08 19:41:22

(e_G, e_H)

mbbchess 2022-07-08 19:41:22

(e G, e H)

ScaryDragon 2022-07-08 19:41:22

(e, e)

The_goat 2022-07-08 19:41:22

(e g, e h)

HotSoup 2022-07-08 19:41:22

(e, e)

cuber 2022-07-08 19:41:22

(e g, e h)

HotSoup 2022-07-08 19:41:22

(e G, e H)

CH328 2022-07-08 19:41:24

The identity is (e G, e H), since (g, h)(e G, e H) = (ge G, he H) = (g, h), and similarly, (e G, e H)(g, h) = (e Gg, e Hh) = (g, h).

CH328 2022-07-08 19:41:41

The subscripts on e are just to clarify which group they're the identity of.

CH328 2022-07-08 19:41:49

You won't always see them written out like this, but they're helpful.

CH328 2022-07-08 19:41:57

−1
Inverses? What's (g, h) ?

ScaryDragon 2022-07-08 19:43:01

(g^-1, h^-1)

Hulk23 2022-07-08 19:43:01

(inverse of g, inverse of h)

HotSoup 2022-07-08 19:43:01

−1 −1
(g ,h )

CovalentBondIBDefinition 2022-07-08 19:43:01

(g^-1, h^-1)

Math5K 2022-07-08 19:43:01

−1 −1
(g ,h )

YaoAOPS 2022-07-08 19:43:01

(g − 1, h − 1)

cuber 2022-07-08 19:43:01

−1 −1
(g ,h )

Leo.Euler math:
Typesetting 100%
2022-07-08 19:43:01

https://artofproblemsolving.com/class/3111-grouptheory/transcript/48543 5/39
7/18/22, 9:20 AM 3111 Group Theory
−1 −1
(g in G, h is H)

just4fun 2022-07-08 19:43:01

(g^-1,h^-1)

ad1406 2022-07-08 19:43:01

(g − 1, h − 1) (I'm not sure how to specify which group we're talking about when writing the notation for inverses)

The_goat 2022-07-08 19:43:01

−1 −1
(g ,h )

romans58 2022-07-08 19:43:01

(g − 1, h − 1)

mbbchess 2022-07-08 19:43:01

(g − 1, h − 1)

romans58 2022-07-08 19:43:01

(g − 1, h − 1)

CH328 2022-07-08 19:43:03

Right, (g, h) − 1 is (g − 1, h − 1), since (g, h)(g − 1, h − 1) = (gg − 1, hh − 1) = (e G, e H), and similarly on the other side.

CH328 2022-07-08 19:43:20

And associativity follows from associativity on G and H.

CH328 2022-07-08 19:43:25

[(g 1, h 1)(g 2, h 2)](g 3, h 3) = ((g 1g 2)g 3, (h 1h 2)h 3)) = (g 1(g 2g 3), h 1(h 2h 3)) = (g 1, h 1)[(g 2, h 2)(g 3, h 3)].

CH328 2022-07-08 19:43:28

No problem. How could you explain, intuitively, how G × H relates to G and H?

YaoAOPS 2022-07-08 19:44:38

its a merge of the two

CovalentBondIBDefinition 2022-07-08 19:44:38

GXH contains components from both G and H

just4fun 2022-07-08 19:44:38

Like a vector made out of two coordinates.

The_goat 2022-07-08 19:44:38

It constructs every pair of elements between G and H

isont 2022-07-08 19:44:38

the combining of the two groups to form a plane?

cuber 2022-07-08 19:44:38

GxH is like a coordinate plane, with one axis corresponding to actions in G, and the other corresponding to actions in H

Hulk23 2022-07-08 19:44:38

It's like combining the groups?

Math5K 2022-07-08 19:44:38

G×H = {(g, h) : g ∈ G, h inH}

CH328 2022-07-08 19:44:43

Those are some good thoughts!

CH328 2022-07-08 19:44:51

On one hand, it "merges" the two groups in some way.

CH328 2022-07-08
Typesetting math: 100% 19:45:14

https://artofproblemsolving.com/class/3111-grouptheory/transcript/48543 6/39
7/18/22, 9:20 AM 3111 Group Theory
2
But the way in which it merges the groups is like how R merges two copies of R: it keeps them quite independent of each other.

CH328 2022-07-08 19:45:19

We could say something like: "G × H is what you get when you put totally independent copies of G and H side-by-side." The group
operations don't interact at all!

CH328 2022-07-08 19:45:27

That makes it exactly as easy to do anything in G × H as it is to do it in G and then in H.

CH328 2022-07-08 19:45:52

We'll see later in the course ways to "merge" groups where things aren't so independent (called "semidirect products" if you're
reading ahead).

CH328 2022-07-08 19:46:03

For now let's study some product groups!

CH328 2022-07-08 19:46:04

Name a group we've seen in earlier weeks that can be written as a product group.

YaoAOPS 2022-07-08 19:46:53

klein

CH328 2022-07-08 19:46:58

That's a good one!

CH328 2022-07-08 19:47:06

The simplest is the Klein 4-group: we defined K as being the Abelian group having two generators a, b of order two. So
K = {e, a, b, ab}.

CH328 2022-07-08 19:47:16

This can be written as a product C 2 × C 2! Let's see how.

CH328 2022-07-08 19:47:39

Oh, but first...

cuber 2022-07-08 19:47:42

C6 = C3 × C2

just4fun 2022-07-08 19:47:42

Cyclic group of 6 for example (C_6)

Math5K 2022-07-08 19:47:42

also C 2 × C 3 ≅ C 6

CH328 2022-07-08 19:47:58

That's another good example! Many cyclic groups are just products of other cyclic groups!

CH328 2022-07-08 19:48:09

We'll state this more precisely later when we study the Chinese Remainder Theorem.

CH328 2022-07-08 19:48:15

Anyway, back to K!

CH328 2022-07-08 19:48:22

By definition, if C 2 = {e, r}, then we have C 2 × C 2 = {(e, e), (r, e), (e, r), (r, r)}. How can we relate C 2 × C 2 to K?

Math5K 2022-07-08 19:49:19

find a bijection between K = {e, a, b, ab} and C 2 × C 2 = {(e, e), (r, e), (e, r), (r, r)}.

CH328 2022-07-08 19:49:32

Yeah! Let's find a bijection between K and C 2 × C 2.

CH328 2022-07-08 19:49:52

They're two sets of size 4, so there are a lot of bijections... but let's find a bijection that plays nicely with the group structure.
Typesetting math: 100%

https://artofproblemsolving.com/class/3111-grouptheory/transcript/48543 7/39
7/18/22, 9:20 AM 3111 Group Theory

CH328 2022-07-08 19:49:59

What do you think e ∈ K should map to?

Hulk23 2022-07-08 19:50:43

(e,e)

smileapple 2022-07-08 19:50:43

(e, e)

Math5K 2022-07-08 19:50:43

(e, e)

The_goat 2022-07-08 19:50:43

(e, e)

cuber 2022-07-08 19:50:43

(e, e)

randomname333 2022-07-08 19:50:43

(e, e)

HotSoup 2022-07-08 19:50:43

(e, e)

lim15787 2022-07-08 19:50:43

probably (e,e)

bigwolfy 2022-07-08 19:50:43

(e,e)

just4fun 2022-07-08 19:50:43

(e,e)

TPSGriff 2022-07-08 19:50:43

(e, e)

ChristiTheMagnificent 2022-07-08 19:50:43

(e,e)

SoumyadeepBhattacharjee 2022-07-08 19:50:43

(e,e)

CH328 2022-07-08 19:50:49

Yeah, let's map it to (e, e). We probably want the identity to go to the identity since that's sort of a special element.

CH328 2022-07-08 19:50:56

What about a?

The_goat 2022-07-08 19:51:25

(r, e)

just4fun 2022-07-08 19:51:25

(r,e)

bigwolfy 2022-07-08 19:51:25

(r,e)

SoumyadeepBhattacharjee 2022-07-08 19:51:25

(r,e)

romans58 2022-07-08 19:51:25

(r, e)

randomname333 2022-07-08 19:51:25

(r, e)
Typesetting math: 100%

https://artofproblemsolving.com/class/3111-grouptheory/transcript/48543 8/39
7/18/22, 9:20 AM 3111 Group Theory

cuber 2022-07-08 19:51:25

(r, e)

Math5K 2022-07-08 19:51:25

(r, e)

Hulk23 2022-07-08 19:51:25

(r,e)

smileapple 2022-07-08 19:51:25

(r, e) (WLOG)

Bean_Burrito 2022-07-08 19:51:25

(r, e)

CH328 2022-07-08 19:51:28

Sure, let's send it to (r, e). That seems like a natural thing to do.

CH328 2022-07-08 19:51:33

Then b can go to (e, r).

CH328 2022-07-08 19:51:55

And that's nice because ab has to go to (r, r) and ab = (r, e)(e, r) = (r, r).

CH328 2022-07-08 19:52:07

So our map behaves well with the group structure.

CH328 2022-07-08 19:52:16

Specifically, there's a bijection φ : K → C 2 × C 2 sending e ↦ (e, e), a ↦ (r, e), b ↦ (e, r), and ab ↦ (r, r).

CH328 2022-07-08 19:52:26

The key point is that φ respects the group structure on both its domain and codomain. For instance, it sends the identity element to
an identity element.

CH328 2022-07-08 19:52:37

This is getting at the notion of "two groups being the same" I've owed you for weeks!

CH328 2022-07-08 19:52:44

We'll get back to thinking about product groups soon, but first let's firm up the idea we just had.

HotSoup 2022-07-08 19:52:46

isomorphism

CH328 2022-07-08 19:52:48

Isomorphism

CH328 2022-07-08 19:52:52

We say that two groups are isomorphic if they are "the same" group, just possibly described differently, as above.

CH328 2022-07-08 19:53:01

In general, we say that any two mathematical structures of the same kind are isomorphic if they are "the same" structure, just
described differently.

CH328 2022-07-08 19:53:15

Before we talk about isomorphisms of groups, let's talk about isomorphisms of other... stuff.

CH328 2022-07-08 19:53:21

When do you think two sets are isomorphic?

smileapple 2022-07-08 19:54:36

bijection mapping one to the othert

Math5K 2022-07-08 19:54:36

they have the same number of elements
Typesetting math: 100%

https://artofproblemsolving.com/class/3111-grouptheory/transcript/48543 9/39
7/18/22, 9:20 AM 3111 Group Theory

smileapple 2022-07-08 19:54:36

cardanalitys are same

The_goat 2022-07-08 19:54:36

when there's a bijection between them

ad1406 2022-07-08 19:54:36

Perhaps when you can find a bijection between their elements without repetition?

AaronSunFTW 2022-07-08 19:54:36

When there is a bijection between them

ChristiTheMagnificent 2022-07-08 19:54:36

well... sets don't have operations of really anything going on, so when they have the same amount of elements?

lim15787 2022-07-08 19:54:36

there's a bijection between them

CH328 2022-07-08 19:54:38

A set is the least complex of all mathematical structures: you just have a bag of elements with no relationships, operations, or
anything.

CH328 2022-07-08 19:55:01

So if we think of when two sets should be "the same," we don't really care which element gets mapped to which other element.

CH328 2022-07-08 19:55:15

All we need is for them to be "the same size."

CH328 2022-07-08 19:55:28

Otherwise, the sets are basically the same: the only thing that differs is the particular names we give to the elements in the sets.

CH328 2022-07-08 19:55:40

Formally, we say S is isomorphic to T if there is some bijection f : S → T. In this context, it's equivalent to call the bijection an
isomorphism (of sets.)

CH328 2022-07-08 19:55:55

Two sets that are in bijection with each other have all the same set-theoretic properties.

CH328 2022-07-08 19:56:01

We'll see repeatedly that an isomorphism between any two kinds of mathematical structure is some kind of bijection between the
structures, but we'll have to add more properties to the bijections once we move beyond sets.

CH328 2022-07-08 19:56:55

An isomorphism is basically saying "these two structures are the same in every way we care about... we just call the elements
different names."

CH328 2022-07-08 19:57:08

Before we get to groups, let's look at a simpler mathematical structure: graphs. You may have encountered these if you took some
C&P with us.

CH328 2022-07-08 19:57:18

A graph has vertices and some edges connecting some of the vertices, like this:

CH328 2022-07-08 19:57:20

Typesetting math: 100%

https://artofproblemsolving.com/class/3111-grouptheory/transcript/48543 10/39
7/18/22, 9:20 AM 3111 Group Theory

CH328 2022-07-08 19:57:24

Let's try to make the definition of "graph" more formal. What do we need?

ScaryDragon 2022-07-08 19:58:42

We have a set of elements and a set of connections between elements

CovalentBondIBDefinition 2022-07-08 19:58:42

to define nodes and edges

ad1406 2022-07-08 19:58:42

Vertices and connections between vertices. We could express each vertex as a letter and express a connection using an ordered
pair, maybe

cuber 2022-07-08 19:58:42

a set of vertices, and a set of edges (i.e. pairs of vertices)

smileapple 2022-07-08 19:58:42

1. vertices

2. pairs of vertices representing and edge

smileapple 2022-07-08 19:58:42

1. vertices

2. pairs of vertices representing an edge connection between the two given points

CH328 2022-07-08 19:58:44

We need a set V of vertices and a set E of edges. And, as many of you noted, an "edge" can be specified as an ordered pair of two
vertices.

CH328 2022-07-08 19:58:57

So, a graph G is formally a pair (V, E), where E is a set of two-element subsets of V.

CH328 2022-07-08 19:59:37

When contrasting with other concepts of graph, this is called a simple graph. (Other people consider graphs where a direction is
assigned to an edge or you're allowed to repeated edges or all sorts of stuff, but let's just stick with simple graphs here.)

CH328 2022-07-08 19:59:45

Now, we're interested in isomorphism of graphs.

CH328 2022-07-08 19:59:50

Which of the graphs below do you think are isomorphic to this graph?

CH328 2022-07-08
Typesetting math: 100% 20:01:40

https://artofproblemsolving.com/class/3111-grouptheory/transcript/48543 11/39
7/18/22, 9:20 AM 3111 Group Theory
We all seem to think the 4th and 5th pictures should be isomorphic!

CH328 2022-07-08 20:02:02

That makes sense: the 4th is basically the same thing, just "upside-down."

CH328 2022-07-08 20:02:15

And "upside-down-ness" probably isn't something we care about when distinguishing graphs.

CH328 2022-07-08 20:02:28

Why should the 5th one be isomorphic?

just4fun 2022-07-08 20:03:46

You can move nodes around to make it look the same.

unicone 2022-07-08 20:03:46

you can make it the same without changing any connections or vertices

ScaryDragon 2022-07-08 20:03:46

It has the same structure of connections, we can draw it out into a straight line

ChristiTheMagnificent 2022-07-08 20:03:46

it preserves not only the number of lines and points, but also the number of connections each point has

AaronSunFTW 2022-07-08 20:03:46

They both have two vertices with one edge and two with two edges

SoumyadeepBhattacharjee 2022-07-08 20:03:46

esentially same graph, just points arranged differently

ad1406 2022-07-08 20:03:46

We have 2 vertices connected to one other vertex, and 2 vertices connected to two other vertices

mbbchess 2022-07-08 20:03:46

We can move to points to match the original graph

CH328 2022-07-08 20:04:06

Yeah, the arrangements all see the same, they're just laid out differently in the plane.

CH328 2022-07-08 20:04:29

(By the way, many of you saw your answers as "3, 4." This seems to use the programming convention of starting with the first
picture at 0 instead of 1. )

CH328 2022-07-08 20:04:47

Anyway, let's formalize these properties! Formally, what properties should isomorphic graphs have?

DottedCaculator 2022-07-08 20:06:06

there is a bijection from one graph to the other

isont 2022-07-08 20:06:06

they should preserve the number of vertices/points and edges, not necessarily in the same arrangement?

ad1406 2022-07-08 20:06:06

| V1 | = | V2 |

CH328 2022-07-08 20:06:10

Good thoughts!

CH328 2022-07-08 20:06:15

Isomorphic graphs should probably have the same number of vertices and the same number of edges. This is building off our
notion of isomorphism, i.e., bijection, for sets.

CH328 2022-07-08 20:06:33

Anything else?

The_goat 2022-07-08 20:07:24

Typesetting math: 100%
Order of the vertices should be the same?

https://artofproblemsolving.com/class/3111-grouptheory/transcript/48543 12/39
7/18/22, 9:20 AM 3111 Group Theory

cuber 2022-07-08 20:07:24

If a and b are in V 1 and a ′ and b ′ are in V 2, our bijection connects a to a ′ and b to b ′ , and ab was an edge in V 1, a ′ b ′ should be an
edge in V 2

just4fun 2022-07-08 20:07:24

Same number of connections at each point

CovalentBondIBDefinition 2022-07-08 20:07:24

the same number of edges attached to each node?

smileapple 2022-07-08 20:07:24

they presertive edges (ie, degree of vertices)

AaronSunFTW 2022-07-08 20:07:24

The corresponding vertices should be connected by corresponding edges

CH328 2022-07-08 20:07:26

Ah, yes, it's not really enough to just have the same number of vertices and edges. Not yet! For instance, the second graph in the
multiple-choice above does not look isomorphic to the given graph. The edges don't line up in the same shape.

CH328 2022-07-08 20:07:37

On the other hand, we see that it's possible to turn the given graph into the fourth or the fifth option, like this:

CH328 2022-07-08 20:07:39

CH328 2022-07-08 20:08:01

Let's formalize a definition of isomorphism between graphs G 1 = (V 1, E 1) and G 2 = (V 2, E 2) ?

CH328 2022-07-08 20:08:10

We said there should be the same number of vertices.

CH328 2022-07-08 20:08:17

So we definitely need an isomorphism of sets here, that is, a bijection f : V 1 → V 2. If two graphs have different numbers of
vertices, then they aren't anywhere close to interchangeable.

CH328 2022-07-08 20:08:24

But as we just saw in our example, any old bijection between vertex sets won't do the job. This bijection has to play nice with the
edges, somehow.

CH328 2022-07-08 20:08:29

Suppose we have an edge in the domain graph G 1. What should happen in G 2 ?

HotSoup 2022-07-08 20:09:25

the edge should be in G 2

isont 2022-07-08 20:09:25

the edge should also be mapped in G 2, and the number of "attachments" to it should remain constant

smileapple 2022-07-08 20:09:25

also be mapped in G2
Typesetting math: 100%
Bean_Burrito 2022-07-08 20:09:25

https://artofproblemsolving.com/class/3111-grouptheory/transcript/48543 13/39
7/18/22, 9:20 AM 3111 Group Theory
If an edge E_1 touches vertices V_1 and V_2 in G_1, then it should touch V_1 and V_2 in G_2 as well

Hulk23 2022-07-08 20:09:25

The edge is still there

ScaryDragon 2022-07-08 20:09:25

It should also be in G2?

CH328 2022-07-08 20:09:27

The image of that edge should still be an edge in G 2 ! That is, if {v 1, v 2} ∈ E 1, then we should have {f(v 1), f(v 2)} ∈ E 2.

CH328 2022-07-08 20:09:33

What about the converse? Do we need to separately assume that if there's an edge between f(v 1) and f(v 2), then there's an edge
between v 1 and v 2 ?

The_goat 2022-07-08 20:10:29

Yes

isont 2022-07-08 20:10:29

yes, since the number of edges has to be preserved

YaoAOPS 2022-07-08 20:10:29

yes

just4fun 2022-07-08 20:10:29

yes?

HotSoup 2022-07-08 20:10:29

yep

ScaryDragon 2022-07-08 20:10:29

Yes

CH328 2022-07-08 20:10:31

Yes! Think about these two graphs:

CH328 2022-07-08 20:10:32

CH328 2022-07-08 20:10:34

The identity function on their vertex sets sends edges to edges, but not conversely. So we need the converse condition to get an
isomorphism.

CH328 2022-07-08 20:10:51

In summary, an isomorphism of graphs is given by a bijection f : V 1 → V 2 such that {v 1, v 2} ∈ E 1 if and only if {f(v 1), f(v 2)} ∈ E 2.

CH328 2022-07-08 20:10:55

Let's label a couple of graphs from above and write down an isomorphism explicitly.

CH328 2022-07-08 20:10:57

Typesetting math: 100%

https://artofproblemsolving.com/class/3111-grouptheory/transcript/48543 14/39
7/18/22, 9:20 AM 3111 Group Theory

CH328 2022-07-08 20:10:59

What bijection between these vertex sets will give an isomorphism between the graphs?
\[a\mapsto\dragdrop{\frac{1}{1}??},b\mapsto\dragdrop{\frac{1}
{1}??},c\mapsto\dragdrop{\frac{1}{1}??},d\mapsto\dragdrop{\frac{1}
{1}??}\]

$x$ $y$ $z$ $w$

Math5K 2022-07-08 20:12:40

x, z, y, w

ScaryDragon 2022-07-08 20:12:40

x, z, y, w

TPSGriff 2022-07-08 20:12:40

x, z, y, w

randomname333 2022-07-08 20:12:40

x, z, y, w

Bean_Burrito 2022-07-08 20:12:40

x, z, y, w

HotSoup 2022-07-08 20:12:40

x, z, y, w

smileapple 2022-07-08 20:12:40

x, z, y, w

just4fun 2022-07-08 20:12:40

x, z, y, w

rav3nclaw 2022-07-08 20:12:40

x, z, y, w

unicone 2022-07-08 20:12:40

x, z, y, w

Hulk23 2022-07-08 20:12:40

x, w, y, z

isont 2022-07-08 20:12:40

x, z, y, w

bigwolfy 2022-07-08 20:12:40

x, z, y, w

mbbchess 2022-07-08 20:12:40

x, z, y, w

mathgeek23
Typesetting 2022-07-08 20:12:40
math: 100%

https://artofproblemsolving.com/class/3111-grouptheory/transcript/48543 15/39
7/18/22, 9:20 AM 3111 Group Theory
x, z, y, w

romans58 2022-07-08 20:12:40

x, z, y, w

The_goat 2022-07-08 20:12:40

x, z, y, w

CH328 2022-07-08 20:12:43

We can send a ↦ x, d ↦ w, b ↦ z, c ↦ y.

CH328 2022-07-08 20:12:57

Indeed, the edge {a, d} matches with {x, w}, as does {b, c} with {y, z}.

CH328 2022-07-08 20:13:14

Furthermore, this bijection sends the edge {a, b} in the domain to the edge {x, z} in the codomain, while sending the non-edges
{a, c}, {b, d}, and {c, d} to the non-edges {x, y}, {w, y}, and {w, z} respectively.

CH328 2022-07-08 20:13:27

So this is an isomorphism!

CH328 2022-07-08 20:13:36

Alright, now let's think about groups!

CH328 2022-07-08 20:13:38

Let's again get things rolling by considering an example of two groups that should end up isomorphic.

CH328 2022-07-08 20:13:51

Recall we've met the symmetry group of an unoriented triangle, D 3, and the group of permutations of a 3-element set, S 3.

CH328 2022-07-08 20:13:54

If we let R be a nonzero rotation and T some reflection in D 3, then we know D 3 = {e, R, R 2, T, RT, R 2T}, and we know how to
multiply those things together.

CH328 2022-07-08 20:13:59

We also know that S 3 = {e, (12), (13), (23), (123), (321)}.

CH328 2022-07-08 20:14:02

Why might we suspect D 3 and S 3 are isomorphic?

The_goat 2022-07-08 20:15:37

Same number of elements?

DottedCaculator 2022-07-08 20:15:37

they have the same number of elements

YaoAOPS 2022-07-08 20:15:37

both have six elements

ScaryDragon 2022-07-08 20:15:37

Same number of elements?

romans58 2022-07-08 20:15:37

they both have 6 elements for a start, and both have 2 elements of order 3

CH328 2022-07-08 20:15:39

They have the same number of elements. That's certainly necessary for an isomorphism.

CH328 2022-07-08 20:15:48

We might dig deeper to see they have three elements of order 2 and one of order 3.

CH328 2022-07-08 20:16:06

And we could dive deeper, too, as many of you have.

CH328 2022-07-08
Typesetting math: 100% 20:16:09

https://artofproblemsolving.com/class/3111-grouptheory/transcript/48543 16/39
7/18/22, 9:20 AM 3111 Group Theory
An isomorphism of graphs was a bijection on vertices respecting the extra structure determined by the edges. We want an
analogous definition for an isomorphism of groups.

CH328 2022-07-08 20:16:34

Fill out the above analogy between graphs and groups.
$$\text{Vertices of a graph are to }\dragdrop{\frac{1}
{1}?????????????????????}\text{ as }$$ $$\text{Edges of a graph
are to }\dragdrop{\frac{1}{1}?????????????????????}$$

$\text{Elements of a group}$

$\text{Multiplication operation of a group}$

$\text{Identity element of a group}$

ScaryDragon 2022-07-08 20:17:13

Elements of a group, Multiplication operation of a group

Bean_Burrito 2022-07-08 20:17:13

Elements of a group, Multiplication operation of a group

CovalentBondIBDefinition 2022-07-08 20:17:13

Elements of a group, Multiplication operation of a group

HotSoup 2022-07-08 20:17:13

Elements of a group, Multiplication operation of a group

isont 2022-07-08 20:17:13

Elements of a group, Multiplication operation of a group

smileapple 2022-07-08 20:17:13

Elements of a group, Multiplication operation of a group

AaronSunFTW 2022-07-08 20:17:13

Elements of a group, Multiplication operation of a group

The_goat 2022-07-08 20:17:13

Elements of a group, Multiplication operation of a group

YaoAOPS 2022-07-08 20:17:13

Elements of a group, Multiplication operation of a group

ajiten 2022-07-08 20:17:13

Elements of a group, Multiplication operation of a group

Hulk23 2022-07-08 20:17:13

Elements of a group, Multiplication operation of a group

Math5K 2022-07-08 20:17:13

Elements of a group, Multiplication operation of a group

Leo.Euler 2022-07-08 20:17:13

Elements of a group, Multiplication operation of a group

randomname333 2022-07-08 20:17:13

Elements of a group, Multiplication operation of a group

just4fun 2022-07-08 20:17:13

Elements of a group, Multiplication operation of a group

romans58math:
Typesetting 2022-07-08
100% 20:17:13
Elements of a group, Multiplication operation of a group
https://artofproblemsolving.com/class/3111-grouptheory/transcript/48543 17/39
7/18/22, 9:20 AM 3111 Group Theory

mbbchess 2022-07-08 20:17:13

Elements of a group, Multiplication operation of a group

TPSGriff 2022-07-08 20:17:13

Elements of a group, Multiplication operation of a group

CH328 2022-07-08 20:17:18

We can think of both a graph and a group as a set (the vertices, respectively the elements), plus some extra decorations (the
edges, respectively the multiplication operation!)

CH328 2022-07-08 20:17:22

(In fact, if you think of the edges of a graph as a set of pairs (v, w) of vertices, you can think of the multiplication of the group as a
set of triples (g, h, gh) of elements, and these things look very much the same indeed.)

CH328 2022-07-08 20:17:44

So seems like a good idea to define an isomorphism of groups to be a bijection which "plays nicely with" the multiplication
operation in some way. After all, an isomorphism of graphs was a bijection of vertices that "plays nicely with" the edges.

CH328 2022-07-08 20:17:56

Here are our Cayley Tables:

CH328 2022-07-08 20:17:59

D3 e T RT R 2T R R2
S3 e (12) (13) (23) (123) (321)
e e T RT R 2T R R2 e e (12) (13) (23) (123) (321)
T T e R2 R R 2T RT (12) (12) e (132) (123) (23) (13)
RT RT R e R2 T R 2T (13) (13) (123) e (321) (12) (23)
2
R T R T 2
R 2
R e RT T (23) (23) (132) (123) e (13) (12)
2 2 (123) (123) (13) (23) (12) (321) e
R R RT R T T R e
(321) (321) (23) (12) (13) e (123)
R2 R2 R 2T R RT e R

CH328 2022-07-08 20:18:09

If φ : G → H for some groups G and H, what would it mean for φ to "play nicely with" the multiplication operation?

just4fun 2022-07-08 20:19:43

phi(gh) = phi(g)phi(h)

HotSoup 2022-07-08 20:19:43

ϕ(a) × ϕ(b) = ϕ(a × b)

romans58 2022-07-08 20:19:43

ϕ(x)ϕ(y) = ϕ(xy)

cuber 2022-07-08 20:19:43

ϕ(g)ϕ(h) = ϕ(gh)

ad1406 2022-07-08 20:19:48

ϕ(a ⋅ b) = ϕ(a) ⋆ ϕ(b)

CH328 2022-07-08 20:20:26

Right! We want "multiplication before φ", which is a ⋅ b to be sent to the same thing as "multiplication after applying φ, which is
φ(a) ⋅ φ(b).

CH328 2022-07-08 20:20:37

If g = g 1 ⋆ Gg 2, we want φ(g) = φ(g 1) ⋆ Hφ(g 2). In other words, we want

φ(g 1 ⋆ Gg 2) = φ(g 1) ⋆ Hφ(g 2), for every g 1, g 2 ∈ G.

Note that the group operations are different on the two sides here.
Typesetting math: 100%

https://artofproblemsolving.com/class/3111-grouptheory/transcript/48543 18/39
7/18/22, 9:20 AM 3111 Group Theory

CH328 2022-07-08 20:21:09

Normally people won't adorn the products with the G and H symbols, but it's helpful while we're getting used to things.

CH328 2022-07-08 20:21:13

This is an equation we'll be seeing a ton of for the rest of the course! We call it the homomorphism property.

CH328 2022-07-08 20:21:34

Thus, our proposed definition of isomorphism of groups is:

CH328 2022-07-08 20:21:35

An isomorphism between groups (G, ⋆ G) and (H, ⋆ H) is a bijection φ : G → H satisfying the homomorphism property:
φ(g 1g 2) = φ(g 1)φ(g 2).

CH328 2022-07-08 20:21:49

Let's construct an isomorphism φ : S 3 → D 3.

CH328 2022-07-08 20:21:51

Again, write S 3 = {e, (12), (13), (23), (123), (321)} and D 3 = {e, R, R 2, T, RT, R 2T}.

CH328 2022-07-08 20:21:57

Let's start easy: Where should φ send e ∈ S 3?

Math5K 2022-07-08 20:22:34

e

Hulk23 2022-07-08 20:22:34

e

romans58 2022-07-08 20:22:34

e ∈ D3

just4fun 2022-07-08 20:22:34

e

bigwolfy 2022-07-08 20:22:34

e ∈ D3

ScaryDragon 2022-07-08 20:22:34

e

isont 2022-07-08 20:22:34

to e in D 3

HotSoup 2022-07-08 20:22:34

e ∈ D3

ad1406 2022-07-08 20:22:34

eD
3

just4fun 2022-07-08 20:22:34

e_D_3

The_goat 2022-07-08 20:22:34

e ∈ D3

CH328 2022-07-08 20:22:36

Probably to e ∈ D 3, since the identities play the same role in their respective groups.

CH328 2022-07-08 20:22:39

How about (12)?

smileapple 2022-07-08 20:23:31

T
Typesetting math: 100%

https://artofproblemsolving.com/class/3111-grouptheory/transcript/48543 19/39
7/18/22, 9:20 AM 3111 Group Theory

The_goat 2022-07-08 20:23:31

T

smileapple 2022-07-08 20:23:31

φ((12)) = T

ScaryDragon 2022-07-08 20:23:31

T

Bean_Burrito 2022-07-08 20:23:31

T?

AaronSunFTW 2022-07-08 20:23:31

T

Leo.Euler 2022-07-08 20:23:31

T

just4fun 2022-07-08 20:23:31

T

CH328 2022-07-08 20:23:34

Since (12) 2 = e, it makes sense to send (12) to some element of order 2 in D 3. Let's try φ((12)) = T.

CH328 2022-07-08 20:23:43

(There are more possible correct answers. )

CH328 2022-07-08 20:23:44

Now let's think about (123). Where should it go?

Hulk23 2022-07-08 20:24:19

R

bigwolfy 2022-07-08 20:24:19

R

romans58 2022-07-08 20:24:19

R

just4fun 2022-07-08 20:24:19

R

AaronSunFTW 2022-07-08 20:24:19

R

ScaryDragon 2022-07-08 20:24:19

R

HotSoup 2022-07-08 20:24:19

R

The_goat 2022-07-08 20:24:19

R

smileapple 2022-07-08 20:24:19

R

isont 2022-07-08 20:24:19

to R in D 3

CovalentBondIBDefinition 2022-07-08 20:24:19

R of D3

CH328 2022-07-08 20:24:21

Typesetting math: 100%
Similarly, if we want φ((123)) to have order 3, then we need it to be R or R 2. Let's send (123) ↦ R.

https://artofproblemsolving.com/class/3111-grouptheory/transcript/48543 20/39
7/18/22, 9:20 AM 3111 Group Theory

CH328 2022-07-08 20:24:24

OK, I claim that completely determines φ! How come?

Hulk23 2022-07-08 20:25:30

Everything else is generated by those two

just4fun 2022-07-08 20:25:30

The other ones are combination of the T and R.

Leo.Euler 2022-07-08 20:25:30

We can use multiplication to determine the remaining elements.

ad1406 2022-07-08 20:25:30

We know that the multiplication structure is preserved, so we can figure out any element in the group besides those two since they
generate S 3

CovalentBondIBDefinition 2022-07-08 20:25:30

because T and R generate the rest of the elements

HotSoup 2022-07-08 20:25:30

because of the homomorphism property

CH328 2022-07-08 20:25:32

We use the homomorphism property. Everything else in S 3 can be written as a product of copies of (12) and (123). Once we specify
what φ does to a set of generators, everything else is determined by the homomorphism property.

CH328 2022-07-08 20:25:38

For instance, what happens to (13)?

CH328 2022-07-08 20:25:47

(Recall (13) = (123)(12).)

Math5K 2022-07-08 20:26:32

RT

cuber 2022-07-08 20:26:32

(13) goes to RT

ScaryDragon 2022-07-08 20:26:32

It goes to RT

The_goat 2022-07-08 20:26:32

RT

randomname333 2022-07-08 20:26:32

RT

HotSoup 2022-07-08 20:26:32

ϕ((13)) = ϕ((123))((12)) = RT =

just4fun 2022-07-08 20:26:32

RT

mbbchess 2022-07-08 20:26:32

RT

smileapple 2022-07-08 20:26:32

(13) = (123)(12) so its just mapped to RT

ChristiTheMagnificent 2022-07-08 20:26:32

RT

CH328 2022-07-08 20:26:34

We have (13) = (123)(12), so we must have
Typesetting math: 100%

https://artofproblemsolving.com/class/3111-grouptheory/transcript/48543 21/39
7/18/22, 9:20 AM 3111 Group Theory
φ((13)) = φ((123)(12)) = φ((123))φ((12)) = RT.

CH328 2022-07-08 20:26:35

The other values can also be determined in this way!

CH328 2022-07-08 20:26:37

What are the remaining two values of ?
$$\varphi((23))=\dragdrop{\frac{1}{1}???}\quad
\varphi((321))=\dragdrop{\frac{1}{1}???}$$

$e$ $R$ $R^2$ $T$ $RT$ $R^2T$

smileapple 2022-07-08 20:27:15

R 2T, R 2

AaronSunFTW 2022-07-08 20:27:15

2 2
R T, R

isont 2022-07-08 20:27:15

2 2
R T, R

randomname333 2022-07-08 20:27:15

R 2T, R 2

ScaryDragon 2022-07-08 20:27:15

2 2
R T, R

HotSoup 2022-07-08 20:27:15

R 2T, R 2

The_goat 2022-07-08 20:27:15

2 2
R T, R

just4fun 2022-07-08 20:27:15

R 2T, R 2

mbbchess 2022-07-08 20:27:15

2 2
R T, R

romans58 2022-07-08 20:27:15

R 2T, R 2

CH328 2022-07-08 20:27:17

We must have φ((23)) = φ((12)(123)) = TR = R 2T and φ((321)) = φ((123) 2) = R 2.

CH328 2022-07-08 20:27:19

So, we have our bijection S 3 → D 3 :

e ↦ e, (12) ↦ T, (13) ↦ RT, (23) ↦ R 2T, (123) ↦ R, (321) ↦ R 2.

CH328 2022-07-08 20:27:31

In principle, to check that this is an isomorphism, we would have to check the homomorphism property φ(g 1g 2) = φ(g 1)φ(g 2) for all
36 pairs (g 1, g 2) ∈ S 3 × S 3.

CH328 2022-07-08 20:27:47

You could write out the Cayley Table to see this, but it's still pretty tedious.

CH328 2022-07-08 20:27:57

Later on, we'll introduce the concept of presentations, which allow us to show that some bijection is an isomorphism using many
fewer checks.
Typesetting This is critical when you have bigger groups, without a computer. With infinite groups, even a computer won't save
math: 100%
you!
https://artofproblemsolving.com/class/3111-grouptheory/transcript/48543 22/39
7/18/22, 9:20 AM 3111 Group Theory

CH328 2022-07-08 20:28:08

So, that's the idea of isomorphism. Are we doing OK up to there?

CH328 2022-07-08 20:28:42

As always, you can ask questions or scream for help any time.

CH328 2022-07-08 20:28:50

Another popular way to visualize an isomorphism is as follows. With φ : G → H in hand, we have two main ways to "get from" G × G
to H: either multiply in G first, or do φ to both coordinates of G × G first.

CH328 2022-07-08 20:28:55

The property says that both these ways are the same! We sometimes visualize this using the diagram below, which is said to be
"commutative" since both routes through it are equal:

CH328 2022-07-08 20:28:59

CH328 2022-07-08 20:29:35

Show that any cyclic group of order 32 is isomorphism to Z / 32Z, but that this isomorphism is not generally unique.

CH328 2022-07-08 20:29:53

This is kind of paying off a debt from earlier weeks: we keep vacillating between proving results for arbitrary cyclic groups of order
n and for specific ones, like Z / nZ or the group of rotations in D n.

CH328 2022-07-08 20:30:03

This result justifies our vacillation: anything you can prove for one cyclic group (well, anything that's group-theoretically
meaningful), you can prove for another one.

CH328 2022-07-08 20:30:14

OK, so suppose we have some cyclic group C 32. How can we try to define an isomorphism φ : C 32 → Z / 32Z ?

smileapple 2022-07-08 20:31:17

let C 32 = ⟨r⟩. then r → 1 (mod 32)

HotSoup 2022-07-08 20:31:17

ϕ(e) = 0, ϕ(x) = 1, ϕ(x n) = n for n < 32

cuber 2022-07-08 20:31:17

Assuming C 32 is generated by g, we can have a bijection such that ϕ(g a) maps to a

YaoAOPSmath:
Typesetting 100%
2022-07-08 20:31:17

https://artofproblemsolving.com/class/3111-grouptheory/transcript/48543 23/39
7/18/22, 9:20 AM 3111 Group Theory
n
ϕ(g ) = n

Math5K 2022-07-08 20:31:17

i
Send g to i

CH328 2022-07-08 20:31:23

Wow some of you see the whole thing already!

CH328 2022-07-08 20:31:38

If you didn't, the key is that we can start by picking a generator g ∈ C 32.

CH328 2022-07-08 20:31:47

It seems natural to set φ(g) = 1 (mod 32).

CH328 2022-07-08 20:31:59

Any other element of the domain is g k for some k, so we must have φ(g k) = φ(g) k = k (mod 32).

CH328 2022-07-08 20:32:19

There is something to show when defining it this way.

CH328 2022-07-08 20:32:22

We used k in the formula for φ, but the representation of an element of C 32 as g k is not unique! For instance, g k = g k + 32.

CH328 2022-07-08 20:32:32

So we have to show that this proposed function is well-defined. That's when you check that a formula you've written down actually
defines a function, not some more general relation where a single input has many outputs.

CH328 2022-07-08 20:32:37

That is, we have to check that we've only assigned one value to every element of the domain. How do we do it?

DottedCaculator 2022-07-08 20:33:40

k ≡ k + 32 (mod 32)

CH328 2022-07-08 20:33:46

Indeed!

CH328 2022-07-08 20:33:53

We have g k = g ℓ if and only if g k − ℓ = e is the identity. Since g has order 32, we know this happens if and only if k − ℓ is a multiple of
32, so k ≡ l (mod 32) as desired.

CH328 2022-07-08 20:34:01

Good! So, φ is a well-defined function C 32 → Z / 32Z.

CH328 2022-07-08 20:34:15

Then we have to show it has the homomorphism property, which we can do with some modular arithmetic, too.

CH328 2022-07-08 20:34:17

We defined φ(g kg ℓ) as k + ℓ (mod 32), which is φ(g k + ℓ), too, again using that g is of order 32.

CH328 2022-07-08 20:34:22

OK, what's left?

just4fun 2022-07-08 20:35:10

invertable

CH328 2022-07-08 20:35:25

Yeah! It has to be bijective (equivalently, invertible).

CH328 2022-07-08 20:35:52

Noting that the domain and codomain have the same cardinality, we can do it by just checking injectivity or surjectivity.

CH328 2022-07-08 20:36:02

We've essentially already proven injectivity.
Typesetting math: 100%
CH328 2022-07-08 20:36:03

https://artofproblemsolving.com/class/3111-grouptheory/transcript/48543 24/39
7/18/22, 9:20 AM 3111 Group Theory
k ℓ k ℓ
If φ(g ) = φ(g ), then k ≡ ℓ (mod 32), so that g = g . This is just observing that we can reverse our logic in the argument that φ is
well-defined.

CH328 2022-07-08 20:36:08

So we're done with that! We have an isomorphism C 32 → Z / 32Z sending g to 1.

CH328 2022-07-08 20:36:12

OK, why isn't φ usually unique?

CH328 2022-07-08 20:36:55

(Think about that first step we did; sending g to 1.)

romans58 2022-07-08 20:37:22

usually there's more than 1 generator

HotSoup 2022-07-08 20:37:22

because any number n ∈ Z / mZ coprime to m could also be chosen as a generator

smileapple 2022-07-08 20:37:22

we can also map g to 3, 5, 7, ⋯, any odd number (ie relatively prime to 32)

CH328 2022-07-08 20:37:25

Right!

CH328 2022-07-08 20:37:27

We could have gone through the same process for any choice of a generator of Z / 32Z for φ(g).

CH328 2022-07-08 20:37:31

There are 16 : any odd residue will do.

CH328 2022-07-08 20:37:41

It might be hard to understand why we would care about that. What this means is that, while C 32 is basically the same as Z / 32Z for
group-theoretic purposes, there's a key difference: Z / 32Z comes with a distinguished generator, namely 1, which other cyclic
groups do not.

CH328 2022-07-08 20:37:48

So there are 16 different, equally good ways to see a general C 32 as Z / 32Z. And we might like to study the structure of this set of
16 isomorphisms!

CH328 2022-07-08 20:38:12

By the way, what about the converse of the fact that φ can send g to any generator of Z / nZ ?

CH328 2022-07-08 20:38:17

If g is a generator of G and φ : G → H is an isomorphism, can we prove h = φ(g) must be a generator of H ?

CH328 2022-07-08 20:38:42

Well, what has to be proved?

smileapple 2022-07-08 20:39:13

that ⟨h⟩ = H

CH328 2022-07-08 20:39:15

We have to prove that every element of H is a power of h.

CH328 2022-07-08 20:39:24

Let's take t ∈ H.

CH328 2022-07-08 20:39:28

We can consider φ − 1(t).

CH328 2022-07-08 20:39:35

We know we have φ − 1(t) = g k for some k.

CH328 2022-07-08 20:39:39

Typesetting math: 100%
k k
We have t = φ(g ) = h , as desired!

https://artofproblemsolving.com/class/3111-grouptheory/transcript/48543 25/39
7/18/22, 9:20 AM 3111 Group Theory

CH328 2022-07-08 20:39:44

So, isomorphisms send generators to generators. (This is true for generating sets too, just messier to prove.)

CH328 2022-07-08 20:39:52

We might ask lots of similar questions:

CH328 2022-07-08 20:39:55

Suppose and are isomorphic groups. Which of the following results do you think should be true?

CH328 2022-07-08 20:41:15

Yeah, lots of you have it!

CH328 2022-07-08 20:41:18

The key idea here is: basically everything is the same for any two isomorphic groups. These kind of results are a bit boring to
prove formally.

CH328 2022-07-08 20:41:37

The only one that doesn't hold is the purple one!

CH328 2022-07-08 20:41:39

In fact, I claim that it's possible to find a group isomorphic to R itself, seen as a group under addition, that doesn't consist of real
numbers.

CH328 2022-07-08 20:41:57

Find a group H not contained in R that's isomorphic to R.

CH328 2022-07-08 20:41:59

Any suggestions?

smileapple 2022-07-08 20:42:48

iR is a good example

Bean_Burrito 2022-07-08 20:42:48

imaginary numbers?

Math5K 2022-07-08 20:42:48

take all complex numbers along a line

CH328 2022-07-08 20:42:55

We could try another subgroup of C that's geometrically a line, such as the imaginary axis Ri = {yi ∣ y ∈ R}.

CH328 2022-07-08 20:43:12

Then we can get an isomorphism Ri → R by dividing by i!

CH328 2022-07-08 20:43:32

It's pretty easy to prove, but I'll leave that to you on your own if you'd like.

CH328 2022-07-08 20:43:36

So, to refine our general lesson: isomorphic groups have all the same group-theoretic properties. But they don't have to match up
on properties like "what the elements 'really are.'"

CH328 2022-07-08 20:43:55

I think it's a good time for a break. We resume at 8:48 EST (5:48 PST).

CH328 2022-07-08 20:48:50

Here we go!

CH328 2022-07-08 20:48:51

Prove that a finitely generated subgroup of (Q, + ) is necessarily cyclic.

CH328 2022-07-08 20:49:01

Typesetting math: 100%
A group G is finitely generated if there is some finite subset of the group, S, such that G is generated by S, that is, G = ⟨S⟩.

https://artofproblemsolving.com/class/3111-grouptheory/transcript/48543 26/39
7/18/22, 9:20 AM 3111 Group Theory

CH328 2022-07-08 20:49:12

Before anything else, are the rationals themselves a cyclic group?

YaoAOPS 2022-07-08 20:49:52

no

HotSoup 2022-07-08 20:49:52

no

romans58 2022-07-08 20:49:52

no

Hulk23 2022-07-08 20:49:52

No

ad1406 2022-07-08 20:49:52

No

ScaryDragon 2022-07-08 20:49:52

No?

just4fun 2022-07-08 20:49:52

no?

CH328 2022-07-08 20:49:55

That would make this problem pretty easy! No, this would say you can write any rational as an integer multiple of some fixed
rational. That must be why "finitely generated" is in there.

CH328 2022-07-08 20:50:08

Consider, say, H =
⟨ 3 2
, , −
2 3
4
5 ⟩
. That's got three generators, which is about the biggest finite number I can think of, so hopefully it'll

get the idea across.

CH328 2022-07-08 20:50:31

We want to prove H is cyclic. We'd be done if we could find a generator, but it might be tough to spot the generator right away. Is
there a weaker problem we can solve?

CH328 2022-07-08 20:51:48

Here's an idea: is there any element of Q that we can write every element in H as an integer multiple of?

Hulk23 2022-07-08 20:52:44

1/30?

HotSoup 2022-07-08 20:52:44

1 1
=
lcm ( 3 , 4 , 5 ) 60

romans58 2022-07-08 20:52:44

1
60

CH328 2022-07-08 20:52:56

Yeah, both 1 / 30 and 1 / 60 work!

CH328 2022-07-08 20:53:03

Let's stick with 1 / 30 since it has a smaller denominator.

CH328 2022-07-08 20:53:07

1
Every element of H is a multiple of 30 .

CH328 2022-07-08 20:53:17

This is because every element in H can be written as a "word" in the three generators.

CH328 2022-07-08
Typesetting math: 100% 20:53:20

https://artofproblemsolving.com/class/3111-grouptheory/transcript/48543 27/39
7/18/22, 9:20 AM 3111 Group Theory
3 2 4
Since we're in an Abelian group, if h ∈ H we can write h = a + b − c for some a, b, c ∈ Z, and this can be written with 30 in the
2 3 5
denominator.

CH328 2022-07-08 20:53:35

So we've just shown that H < ⟨⟩ 1

30
.

CH328 2022-07-08 20:53:42

How does this help?

ad1406 2022-07-08 20:54:12

We know that a subset of a cyclic group is cyclic

cuber 2022-07-08 20:54:12

any subgroup of a cyclic group is necessarily cyclic by FTCG

just4fun 2022-07-08 20:54:12

If that's cyclic, the subgroup would also be cyclic?

CH328 2022-07-08 20:54:14

We saw last week that every subgroup of a cyclic group is cyclic! So, H is cyclic, bam.

CH328 2022-07-08 20:54:50

1
In fact, H turns out to be equal to ⟨ 30 ⟩.

CH328 2022-07-08 20:55:07

I''ll leave that as a challenge for you to show so that we can explore other things in class.

CH328 2022-07-08 20:55:12

Or maybe we'll circle back if we finish early.

CH328 2022-07-08 20:55:21

Some big theorems

CH328 2022-07-08 20:55:26

Now let's take a look at one of the most famous statements in number theory, expressed in group-theoretic language.

CH328 2022-07-08 20:55:36

First, something concrete.

CH328 2022-07-08 20:55:39

How many of the elements of Z / 8Z × Z / 9Z × Z / 5Z generate the group?

CH328 2022-07-08 20:55:47

Let's call this group G. First, when will some (a, b, c) generate this group?

CH328 2022-07-08 20:56:43

Remember, in a product group, each factor sort of acts "independently." What should then be true about a, b, and c?

ad1406 2022-07-08 20:57:50

a generates Z / 8Z, and so on

YaoAOPS 2022-07-08 20:57:50

When a generates Z / 8Z, etc.

smileapple 2022-07-08 20:57:50

a generates Z / 8Z, b generates Z / 9Z, and c generates Z / 5Z

The_goat 2022-07-08 20:57:50

relatively prime to the modulo

CH328 2022-07-08 20:57:52

Right! a, b,math:
Typesetting and100%
c should each generate their component.

https://artofproblemsolving.com/class/3111-grouptheory/transcript/48543 28/39
7/18/22, 9:20 AM 3111 Group Theory

CH328 2022-07-08 20:58:11

Otherwise, we couldn't possibly get the whole group, since the product is component-wise and each component acts
independently from the others.

CH328 2022-07-08 20:58:25

We might hope the converse is true, too: if a, b, and c each generate their respective components, then (a, b, c) generates G.

CH328 2022-07-08 20:58:38

Let's try to prove it!

CH328 2022-07-08 20:58:41

Let's assume a, b, and c are each generators of their components from here on out.

CH328 2022-07-08 20:58:44

How can we try to prove that (a, b, c) generates G ?

CH328 2022-07-08 20:59:29

There's a lot of elements in G, so I don't really want to go through them one at a time to see if we can get every element... how can
we reduce this amount of work?

isont 2022-07-08 21:00:29

maybe find simple elements to show similarity?

CH328 2022-07-08 21:00:52

We can show some simple elements are multiples of (a, b, c)! Then we can build the rest of G out of these simple elements.

CH328 2022-07-08 21:01:00

What simple elements do we know would generate all of G?

DottedCaculator 2022-07-08 21:01:50

(1, 0, 0) (0, 1, 0) (0, 0, 1)

YaoAOPS 2022-07-08 21:01:50

(1, 0, 0), (0, 1, 0), (0, 0, 1)

CH328 2022-07-08 21:01:57

Yeah! There are other choices, but those definitely generate all of G!

CH328 2022-07-08 21:01:59

We know that G is generated by elements (1, 0, 0), (0, 1, 0), and (0, 0, 1).

CH328 2022-07-08 21:02:04

So, if a, b, and c generate their respective factors, we just need to show that (1, 0, 0), (0, 1, 0), and (0, 0, 1) are multiples of (a, b, c).

CH328 2022-07-08 21:02:12

How can we get (1, 0, 0) as such a multiple?

CH328 2022-07-08 21:03:10

Well there are 0's in the 2nd and 3rd component. How can we get rid of those?

CH328 2022-07-08 21:03:28

I mean - 0's in the second and 3rd component of (1, 0, 0). How do we get rid of the b, c?

DottedCaculator 2022-07-08 21:03:59

multiply by 45

CH328 2022-07-08 21:04:03

Ah, clever!

CH328 2022-07-08 21:04:10

First, we might write 45(a, b, c) = (45a, 0, 0), since | b | = 9 and | c | = 5. That looks like progress.

CH328 2022-07-08 21:04:23

To get (1, 0, 0) as a multiple of (45a, 0, 0), we need to argue that 45a generates Z / 8Z if a does. Why is that true?
Typesetting math: 100%

https://artofproblemsolving.com/class/3111-grouptheory/transcript/48543 29/39
7/18/22, 9:20 AM 3111 Group Theory

smileapple 2022-07-08 21:05:18

gcd (45, 8) = 1

Hulk23 2022-07-08 21:05:18

45 is reltatively prime to 8

YaoAOPS 2022-07-08 21:05:18

gcd (45, 8) = 1

HotSoup 2022-07-08 21:05:18

45 ≡ 5 (mod 8) is coprime to 8

The_goat 2022-07-08 21:05:18

45 is relatively prime to 8

ad1406 2022-07-08 21:05:18

45 mod 8 ≅ 1 (mod 2)

CH328 2022-07-08 21:05:26

Well, 45a ≡ 5a (mod 8), and 5(5a) = a, so 45a generates Z / 8Z. If, say, k ⋅ 45a = 1, then we know that 45k ⋅ (a, b, c) = (1, 0, 0).

CH328 2022-07-08 21:05:38

Now what about (0, 1, 0) and (0, 0, 1)?

HotSoup 2022-07-08 21:06:39

similar argument

Hulk23 2022-07-08 21:06:39

The same thing

YaoAOPS 2022-07-08 21:06:39

Multiply by 40, 72

isont 2022-07-08 21:06:39

multiply by 40 and 72?

ScaryDragon 2022-07-08 21:06:39

40 and 72

ad1406 2022-07-08 21:06:39

We can use 40 and 72 and the same logic

The_goat 2022-07-08 21:06:39

Same strategy?

cuber 2022-07-08 21:06:39

to get (0, 1, 0) we multiply (a, b, c) by a multiple of 40, and for (0, 0, 1) we multiply (a, b, c) by a multiple of 72

CovalentBondIBDefinition 2022-07-08 21:06:39

multiply by 40 and 72 respectively

CH328 2022-07-08 21:06:42

Same thing, we just multiply by something different to start with.

CH328 2022-07-08 21:06:45

In the same way, we see that (0, 1, 0) and (0, 0, 1) are multiples of (a, b, c) because 40b generates Z / 9Z and 72c generates Z / 5Z.

CH328 2022-07-08 21:06:52

So, how many generators are there in total?

CH328 2022-07-08 21:06:53

Typesetting math: 100%

smileapple 2022-07-08 21:08:00

https://artofproblemsolving.com/class/3111-grouptheory/transcript/48543 30/39
7/18/22, 9:20 AM 3111 Group Theory
4⋅6⋅4

ad1406 2022-07-08 21:08:00

4⋅6⋅4

YaoAOPS 2022-07-08 21:08:00

4⋅6⋅4

CH328 2022-07-08 21:08:04

We showed in the homework that any C p k has p k − p k − 1 = p k − 1(p − 1) generators, since that's the number of residues relatively
prime to p k. Thus Z / 2 3Z has 4 generators, Z / 3 2Z has 6, and Z / 5Z has 4.

CH328 2022-07-08 21:08:11

Som G has 4 ⋅ 6 ⋅ 4 = 96 generators.

CH328 2022-07-08 21:08:17

Do we see any other way to get the number 96 out of this group?

CH328 2022-07-08 21:08:43

(Hint: 8 ⋅ 9 ⋅ 5 = 360.)

smileapple 2022-07-08 21:09:00

ϕ(8 ⋅ 9 ⋅ 5) = ϕ(360) = 96

CH328 2022-07-08 21:09:06

We have 96 = ϕ(360) ! Gosh, that's surely not a coincidence.

CH328 2022-07-08 21:09:17

We know that Z / 360Z also has ϕ(360) = 96 generators. Why might Z / 360Z and Z / 8Z × Z / 9Z × Z / 5Z have the same number of
generators?

smileapple 2022-07-08 21:09:58

They are isomorphic

romans58 2022-07-08 21:09:58

the groups are isomorphic

isont 2022-07-08 21:09:58

are they isomorphic?

cuber 2022-07-08 21:09:58

maybe they're isomorphic

Hulk23 2022-07-08 21:09:58

They're isomorphic?

CH328 2022-07-08 21:10:01

One way would be if they're isomorphic! Indeed...

CH328 2022-07-08 21:10:02

The Chinese Remainder Theorem for groups states that if m and n are coprime integers, there is an isomorphism between

Z / mnZ

and

Z / mZ × Z / nZ

given by sending k in Z / mnZ to (k (mod m), k (mod n)).

CH328 2022-07-08 21:10:35

This is how we make precise what I said early in the class, that cyclic groups can often be written as products of smaller cyclic
groups.
Typesetting math: 100%
CH328 2022-07-08 21:10:56

https://artofproblemsolving.com/class/3111-grouptheory/transcript/48543 31/39
7/18/22, 9:20 AM 3111 Group Theory
Let's see how this connects to the Chinese Remainder Theorem you likely already know if you took Intro Number Theory. The
statement is something like this:

CH328 2022-07-08 21:10:58

For relatively prime integers m, n, the system of congruences k ≡ a (mod m), k ≡ b (mod n) has a solution k which is unique modulo
mn.

CH328 2022-07-08 21:11:04

The connection to the group-theoretic CRT is a bit tricky.

CH328 2022-07-08 21:11:09

Note that the fact that the mapping k (mod mn) ↦ (k (mod m), k (mod n)) is an isomorphism means that it has an inverse! How
does this connect to solving a system of congruences in the classical CRT?

cuber 2022-07-08 21:12:12

the "coordinate" (k mod m, k mod n) has a unique value k mod mn

smileapple 2022-07-08 21:12:12

basically every element in Z / mZ × Z / nZ has a 'partner' in Z / mnZ, (thinking of the congruence as representing the product and its
solution as representing Z / mnZ)

CH328 2022-07-08 21:12:14

This inverse maps (a (mod m), b (mod n)) to k (mod mn) such that a ≡ k (mod m) and b ≡ k (mod n). This is the existence part of
the classical theorem.

CH328 2022-07-08 21:12:23

What about uniqueness? Does that show up in the group-theoretic theorem?

smileapple 2022-07-08 21:13:07

yes, because isomorphism implies uniqueness in the bijeciton

romans58 2022-07-08 21:13:07

yes, because Group theory requires a bijection

The_goat 2022-07-08 21:13:07

yes, it can only correspond to one pair

CH328 2022-07-08 21:13:09

Uniqueness comes from the fact that k ↦ (k (mod m), k (mod n)) is injective. So we recover the whole classical theorem from the
group-theoretic statement.

CH328 2022-07-08 21:13:31

We learn more from the group-theoretic statement, though!

CH328 2022-07-08 21:13:34

We know this mapping is not just a bijection, but an isomorphism of groups.

CH328 2022-07-08 21:13:40

It means that the solution to the system k ≡ a + a ′ (mod m), k ≡ b + b ′ (mod n) is the sum of the solutions to the two systems
k ≡ a (mod m), k ≡ b (mod n) and k ≡ a ′ (mod m), k ≡ b ′ (mod n), modulo mn.

CH328 2022-07-08 21:13:52

This is easy enough to check directly, but the group-theoretic formulation bakes it into the statement more neatly.

CH328 2022-07-08 21:14:04

We know from number theory that the CRT doesn't work when m and n are not relatively prime. For instance, the system
k ≡ 1 (mod 2), k ≡ 0 (mod 2), certainly has no solution, modulo 4, 2, or anything else!

CH328 2022-07-08 21:14:09

What does this failure of the classical theorem correspond to, group-theoretically?

CH328 2022-07-08 21:15:03

Think about something we showed earlier in the class.
Typesetting math: 100%
smileapple 2022-07-08 21:15:28

https://artofproblemsolving.com/class/3111-grouptheory/transcript/48543 32/39
7/18/22, 9:20 AM 3111 Group Theory
K4

just4fun 2022-07-08 21:15:28

The example of C_2xC_2 not isomorphic to C_4?

smileapple 2022-07-08 21:15:28

K4 ≅ C2 × C2

CH328 2022-07-08 21:15:32

Ah!

CH328 2022-07-08 21:15:34

Here, we're seeing that C 2 × C 2 is not isomorphic to C 4.

CH328 2022-07-08 21:15:55

Instead, C 2 × C 4 ≅ K where K is the Klein-four group, that is different from C 4.

CH328 2022-07-08 21:16:10

How could we see easily that K is not isomorphic to C 4?

smileapple 2022-07-08 21:17:01

all elements have order one or two

cuber 2022-07-08 21:17:01

K is not generated by one element

smileapple 2022-07-08 21:17:01

unlike C_4 with order 4, 2, 4, 1

Ilikeapos 2022-07-08 21:17:01

Orders

ad1406 2022-07-08 21:17:01

It has no elements of order 4

just4fun 2022-07-08 21:17:01

different order of elements

CH328 2022-07-08 21:17:02

One way is to consider the orders of various elements. C 4 has two elements of order 4, for instance, but C 2 × C 2 has none. Or the
generators: C 4 is generated by 1 element, and K is not.

CH328 2022-07-08 21:17:07

Alright, let's get to the proof.

CH328 2022-07-08 21:17:10

Prove the CRT.

CH328 2022-07-08 21:17:25

That's Chinese Remainder Theorem, if you haven't seen that abbreviation before.

CH328 2022-07-08 21:17:34

We need to show that Z / mZ × Z / nZ is cyclic, of order mn. Well, I'm happy with the order being mn. . .

CH328 2022-07-08 21:17:39

What should this group be generated by, if the stated map is really an isomorphism?

CH328 2022-07-08 21:18:40

(Think about what easy element generates Z / mnZ. )

ad1406 2022-07-08 21:19:24

(1, 1)?

romans58 2022-07-08 21:19:24

Typesetting
(1, 1) math: 100%

https://artofproblemsolving.com/class/3111-grouptheory/transcript/48543 33/39
7/18/22, 9:20 AM 3111 Group Theory

Ilikeapos 2022-07-08 21:19:24

(1, 1)

cuber 2022-07-08 21:19:24

(1, 1)

CH328 2022-07-08 21:19:29

All we need to do is show that Z / mZ × Z / nZ is generated by (1 (mod m), 1 (mod n)). (we'll drop the mod's where it woun't be too
confusing.) How could we show this?

CH328 2022-07-08 21:20:05

(What could we show about the element (1, 1)? )

cuber 2022-07-08 21:20:40

Well the order of (1, 1) has to be mn, and two cyclic groups with the same order have to be isomorphic

CH328 2022-07-08 21:20:45

We could compute the order!

CH328 2022-07-08 21:20:49

Suppose we have (1, 1) a = (0 (mod m), 0 (mod n)).

CH328 2022-07-08 21:20:51

This tells us that a is a multiple of both m and n. What does that imply?

The_goat 2022-07-08 21:21:57

it mn since m and n are coprime

romans58 2022-07-08 21:21:57

a = kmn (for integer k) because m and n are coprime

HotSoup 2022-07-08 21:21:57

a = klcm(m, n) = kmn

cuber 2022-07-08 21:21:57

since m and n are coprime, the smallest positive value of a has to be $mn

Ilikeapos 2022-07-08 21:21:57

mn ∣ a

smileapple 2022-07-08 21:21:57

mn ∣ a

Hulk23 2022-07-08 21:21:57

a is 0 mod mn

CH328 2022-07-08 21:21:59

Since m and n are coprime, a must be a multiple of mn. Thus | (1, 1) | = mn, as desired!

CH328 2022-07-08 21:22:10

Neat! We've proven the Chinese Remainder Theorem.

CH328 2022-07-08 21:22:13

Now let's think about how to use it.

CH328 2022-07-08 21:22:15

In general, when does the CRT allow us (and not allow us) to decompose C n as a product?

romans58 2022-07-08 21:23:16

when n is the product of two or more distinct prime powers

smileapple 2022-07-08 21:23:16

fwhen the decomposed values are (pairwise) relativelu prime

just4fun math:
Typesetting 100% 21:23:16
2022-07-08

https://artofproblemsolving.com/class/3111-grouptheory/transcript/48543 34/39
7/18/22, 9:20 AM 3111 Group Theory
n is a product of two caprimes

Ilikeapos 2022-07-08 21:23:16

We can write C n as the product of the cyclic groups of orders that are the prime powers dividing n

ad1406 2022-07-08 21:23:16

If n is the product of relatively prime integers, we can decompose it into those

HotSoup 2022-07-08 21:23:16

when n = ∏ in i where n i ≠ n j for all i ≠ j and the n i's are all pairwise coprime

DottedCaculator 2022-07-08 21:23:16

if p ∣ n but p2 ∤ n then C n = C pC n
p

just4fun 2022-07-08 21:23:16

n is a product of two coprimes

CH328 2022-07-08 21:23:21

If n = km has two nontrivial relatively prime factors, then we can apply the CRT.

CH328 2022-07-08 21:23:30

We can't if and only if n = p t is a prime power. So, CRT lets us understand finite cyclic groups, in essence, in terms of cyclic groups
of prime power order.

CH328 2022-07-08 21:23:44

Let's revisit our example. Since 360 = 72 ⋅ 5, we have Z / 360Z ≅ Z / 72Z × Z / 5.

CH328 2022-07-08 21:23:49

Since 72 = 8 ⋅ 9, we furthermore have Z / 360Z ≅ Z / 8Z × Z / 9Z × Z / 5Z.

CH328 2022-07-08 21:23:53

That explains why both groups have ϕ(360) = 96 generators, as we had guessed!

CH328 2022-07-08 21:24:19

We just computed ϕ(360) as ϕ(8)ϕ(9)ϕ(5) = 2 2(2 − 1) ⋅ 3(3 − 1) ⋅ 1(5 − 1). It's nice to rewrite this like so:

ϕ(360) = 2 2(2 − 1) ⋅ 3(3 − 1) ⋅ 1(5 − 1)

2−1 3−1 5−1
= 23 ⋅ 32 ⋅ 5 ⋅ ⋅ ⋅
2 3 5

( )( )( )
= 360 1 −
1
2
1−
1
3
1−
1
5
.

CH328 2022-07-08 21:24:36

Does anybody know another way to compute ϕ(360)?

smileapple 2022-07-08 21:25:09

PIE ?

CH328 2022-07-08 21:25:12

It can be done using inclusion-exclusion. A natural number n ≤ 360 is relatively prime to 360 if and only if it's divisible by none of
2, 3, and 5.

CH328 2022-07-08 21:25:19

So we can subtract the number of options for n divisible by 2, 3, and 5, add the number of options divisible by two of these, and
subtract the number divisible by all three:

ϕ(360) = 360 − (180 + 120 + 72) + (60 + 36 + 24) − 12 = 96.

Typesetting math: 100%
CH328 2022-07-08 21:25:27

https://artofproblemsolving.com/class/3111-grouptheory/transcript/48543 35/39
7/18/22, 9:20 AM 3111 Group Theory

( )( )( )
It's not too obvious that the expressions 360 − (180 + 120 + 72) + (60 + 36 + 24) − 12 and 360 1 −
1
2
1−
1
3
1−
1
5
should be

equal. But now we know why they are!

CH328 2022-07-08 21:25:33

This general argument can readily be generalized to a nice explicit formula for ϕ(n).

CH328 2022-07-08 21:25:42

Intimately related to the Chinese Remainder Theorem is the Fundamental Theorem of Finitely Generated Abelian Groups, which
classifies all finitely generated Abelian groups.

CH328 2022-07-08 21:26:00

Let's work our way up to it with an example.

CH328 2022-07-08 21:26:01

What are all the Abelian groups of order 16 you can think of?

smileapple 2022-07-08 21:26:40

C 16

Leo.Euler 2022-07-08 21:26:40

C 16

mbbchess 2022-07-08 21:26:40

C 16

just4fun 2022-07-08 21:26:40

C_16

mbbchess 2022-07-08 21:26:40

C 16

Leo.Euler 2022-07-08 21:26:40

(mod 16)

CH328 2022-07-08 21:26:44

Well, the cyclic group C 16 comes to mind right away. Are there any others we can build out of cyclic groups?

DottedCaculator 2022-07-08 21:27:08

C 16, C 4 × C 4, C 8 × C 2

smileapple 2022-07-08 21:27:08

C 16, C 4 × C 4, C 4 × C 2 × C 2, C 2 × C 2 × C 2 × C 2

ad1406 2022-07-08 21:27:08

C 16, C 4 × C 4, C 2 × C 2 × C 2 × C 2

Ilikeapos 2022-07-08 21:27:08

C4 × C4

ScaryDragon 2022-07-08 21:27:08

C_8 * C_2?

romans58 2022-07-08 21:27:08

C 16, C 4 × C 4, C 2 × C 8, K × K

isont 2022-07-08 21:27:08

C 2 * C 8?

CH328 2022-07-08 21:27:13

We can take products! Any product of cyclic groups will be Abelian.

CH328 2022-07-08 21:27:20

Typesetting math: 100%
They correspond to all the factorizations of 16:

https://artofproblemsolving.com/class/3111-grouptheory/transcript/48543 36/39
7/18/22, 9:20 AM 3111 Group Theory
C 16, C 8 × C 2, C 4 × C 4, C 4 × C 2 × C 2, C 2 × C 2 × C 2 × C 2.

So, that's five Abelian groups of order 16. Can we think of any others?

smileapple 2022-07-08 21:27:44

hmm i gess not

Ilikeapos 2022-07-08 21:27:44

Nope

CH328 2022-07-08 21:27:55

I hope not! Or else the theorem I'm about to tell you wouldn't be true.

CH328 2022-07-08 21:28:08

The fundamental theorem of finite Abelian groups says, in short, that every finite Abelian group is a product of cyclic groups!

CH328 2022-07-08 21:28:20

Together with the Chinese remainder theorem, this lets us decompose every finite Abelian group uniquely as a product of cyclic
groups of prime-power order.

CH328 2022-07-08 21:28:30

This should remind you of the fundamental theorem of arithmetic: every integer can be factored uniquely into a product of prime
powers. In fact, in many ways, finite Abelian group theory is just number theory with a bow tied on.

CH328 2022-07-08 21:28:45

Formally, the fundamental theorem of finite Abelian groups says:
If G is any finite Abelian group, then there exists a unique (up to reordering) set of prime powers,

{p 1d 1 , …p dnn}

such that

G ≅ C p 1d 1 × … × C p dnn.

CH328 2022-07-08 21:29:03

Pretty neat!

The_goat 2022-07-08 21:29:18

Woah

Ilikeapos 2022-07-08 21:29:18

Yay!

ScaryDragon 2022-07-08 21:29:18

CH328 2022-07-08 21:29:20

It's a huge theorem in Group Theory! We can classify all finite Abelian groups!

CH328 2022-07-08 21:29:26

In fact, we can cheaply upgrade this to the Fundamental Theorem of Finitely Generated Abelian Groups, which we'll reliably
abbreviate to FTFGAG.

CH328 2022-07-08 21:29:44

I bet you haven't seen a math abbreviation that fun since CPCTC.

CH328 2022-07-08 21:29:51

Suppose that an Abelian group G is generated by g and h, with | g | = 3 and | h | = ∞. So G won't be finite! What might G be?

CH328 2022-07-08 21:30:34

Take a guess if you're not sure! Remember, there's only one infinite cyclic group.
Typesetting math: 100%
Ilikeapos 2022-07-08 21:31:03

https://artofproblemsolving.com/class/3111-grouptheory/transcript/48543 37/39
7/18/22, 9:20 AM 3111 Group Theory
C3 × Z

YaoAOPS 2022-07-08 21:31:03

C3 × Z

CH328 2022-07-08 21:31:06

Well, the only obvious possibility is that G = Z / 3Z × Z.

CH328 2022-07-08 21:31:20

The only infinite cyclic group is Z.

CH328 2022-07-08 21:31:30

So we might think G is a product of Z and C 3.

CH328 2022-07-08 21:31:34

The FTFGAG tells us that this is the only possibility (up to isomorphism!)

CH328 2022-07-08 21:31:42

Formally, the FTFGAG states: If G is any finitely generated Abelian group, then there exists a unique natural number r and a unique
(up to reordering) set of prime powers,

{p 1d 1 , …p dnn}

such that

G ≅ C p 1d 1 × … × C p dnn × Z r.

CH328 2022-07-08 21:31:53

We can't prove the FTFGAG right now. The key piece of machinery we're missing is the notion of quotient: once you've found a
subgroup isomorphic to C p d, you need some way to inductively get rid of it, resulting in a smaller group. But you can't simply take
the complement of a subgroup–that's never another subgroup.

CH328 2022-07-08 21:32:16

And if you study abstract algebra further you'll go onto see a very beautiful module-theoretic proof of it...

CH328 2022-07-08 21:32:20

But that's a story for another day.

CH328 2022-07-08 21:32:28

I'll stick around after class if anyone has any lingering questions!

CH328 2022-07-08 21:32:29

SUMMARY

CH328 2022-07-08 21:32:34

Today, we met product groups. A product group is just a group whose elements are ordered pairs of elements: G × H = {(g, h)},
with the binary operations happening independently in each coordinate.

CH328 2022-07-08 21:32:39

We learned the definition of isomorphism. This is the notion we intend whenever we say that two mathematical are the "same:" an
isomorphism is a bijection preserving all relevant structure.

CH328 2022-07-08 21:32:52

Philosophically, isomorphism is extremely natural. Again and again in mathematics, we have two objects that come from different
concrete contexts but are abstractly "effectively the same." Isomorphism is the concept that allows us to talk about these notions
less vaguely.

CH328 2022-07-08 21:33:00

We also discussed Abelian groups in some generality.

CH328 2022-07-08 21:33:04

Finally, we met a group-theoretic formulation of the Chinese Remainder Theorem, which led to the fundamental theorem of finitely
Typesetting math: 100%
generated Abelian groups (FTFGAG), which states that any finitely generated Abelian group is isomorphic to a product of cyclic

https://artofproblemsolving.com/class/3111-grouptheory/transcript/48543 38/39
7/18/22, 9:20 AM 3111 Group Theory
groups of prime power or infinite order, in a unique way.

CH328 2022-07-08 21:33:18

See you next time! Keep working hard everyone!

CH328 2022-07-08 21:33:33

And enjoy it as much as you can.

CH328 2022-07-08 21:33:50

Group Theory is a beautiful subject. Don't stress yourself out; it's natural to not understand some stuff. You're doing awesome by
being in the class and trying problems.

© 2022 Art of Problem Solving

About Us • Contact Us • Terms • Privacy

Copyright © 2022 Art of Problem Solving

Typesetting math: 100%

https://artofproblemsolving.com/class/3111-grouptheory/transcript/48543 39/39