What is a good way to understand the intuition behind the third isomorphism theorem? Is it something looking like zooming out of the group structure?(i.e. discard the detailed info by modding out a normal subgroup).
Asked
Active
Viewed 1,803 times
5
-
3I always thought of it as a cancellation property: (G / M) / (N / M) is isomorphic to G / N. – Loreno Heer Jul 05 '14 at 23:23
-
Here's a quote from Dummit & Foote: "[The third isomorphism theorem] shows that we gain no new structural information from taking quotients of a quotient group." – Henry May 18 '21 at 23:57
4 Answers
8
The following diagram is one classical way that the theorem is frequently "illustrated" (excerpted from Burris and Sankappanavar's A Course in Universal Algebra). Essentially the diagrams illustrate the set-theoretic view of the matter, in terms of equivalence classes or, equivalently, partitions. Before studying the algebraic versions of these theorems, for motivation, it helps to have a good grasp on their much simpler set-theoretic versions, i.e. "forget" the algebraic structure (operations), so reducing from algebras and their maps (homs), to simply sets and functions. Then a congruence reduces to an equivalence relation (no operations to preserve), i.e. a partition, which leads to the vivid depiction below, in terms of finer and coarser partitions.
Bill Dubuque
- 282,220
-
It seems this does not require that $\theta$ and $\phi$ being normal. How can one show that this equivalence reduces to normality. I knew normality is an equivalence. However, is this one to one correspondence between normality with this equivalence(i.e. normality iff equivalence)? – user45765 Jul 06 '14 at 18:21
-
@user45765 Above $\theta$ and $\phi$ are congruences. As in the classical case for $,\Bbb Z,,$ in universal algebra one defines a congruence as an equivalence relation that is compatible with all of the operations of the algebra. For algebras like groups and rings whose congruences are determined by single congruence class, this reduces to the classical definitions using normal subgroups and ideals, e.g. see here. – Bill Dubuque Jul 06 '14 at 18:28
2
A very silly (and not "mathematically" rigorous) way I remember the theorem is by examining the analogue in rational numbers. Everyone who knows anything gets that $\dfrac{a/b}{c/b} = \dfrac{a}{b}\cdot\dfrac{b}{c} =\dfrac{a}{c}$.
We can intuitively think of the Third Isomorphism theorem in the same fashion. $$\dfrac{A/B}{C/B} ``=" \dfrac{A}{B}\cdot\dfrac{B}{C} \cong A/C$$
Again, not mathematically valid, but a good mnemonic.
Vladhagen
- 5,048
2
The best motivation I have comes from the classical groups: $\mathbb{Z}$ and $\mathbb{Z}/n\mathbb{Z}$. In this way when you start with $m|n$ so that we have a subgroup relation to talk about
$$\mathbb{Z}/n\mathbb{Z}\bigg/n\mathbb{Z}/m\mathbb{Z}\cong \mathbb{Z}/(n/m)\mathbb{Z}$$
so that things really work like fractions, which is how the notation is setup, and which motivates the idea for more general groups.
This also works if you start with $A=\mathbb{Z}/n\mathbb{Z}$. Then taking $k|m|n$ we have
$$A/kA\bigg/mA/kA\cong A/(m/k)A$$
This is comparable to actual fractions since it looks like
$${n\over k}\bigg/{n\over m}={m\over k}$$
Adam Hughes
- 37,795
- 10
- 61
- 85
2
Your suggestion is fine.
Another intuition I think is important is that it's something to understand in the context of higher algebra: you're doing arithmetic with groups. Not with elements of groups, but with the groups themselves. The third isomorphism theorem is a basic algebraic identity used either for simplifying quotients of quotients, or conversely to write a quotient as a quotient of quotients as needed.
I often use it in the context of expressing a quotient $G/(HK)$ as first modding out by $H$ and then modding out by $K$, which can be justified as
$$ G / (HK) \cong (G / H) / (HK/H) \cong (G/H) / \pi(K) $$
where $\pi$ is the projection map $G \to G/H$.
I find the third isomorphism particularly appealing if you think in terms of setoids -- that is, you think in terms of sets equipped with equivalence relations, and modding out by a normal subgroup means equipping your group with the corresponding equivalence relation rather than constructing the set of equivalence classes. The third isomorphism theorem then says if you equip things with an equivalence relation, and then a coarser equivalence relation, you get the same result as if you just skipped the intermediate step.