Chapter 07
Medium

Direct Products & Direct Sums

00 · Symbol Glossary

$G \times H$External direct product

The set of ordered pairs {(g,h):g∈G,h∈H}\{(g,h) : g\in G, h\in H\}, with operation (g1,h1)(g2,h2):=(g1g2,h1h2)(g_1,h_1)(g_2,h_2) := (g_1g_2, h_1h_2) — combine in each coordinate independently. The most direct way to build a new, larger group from two known ones.

$(g,h)$An element of a direct product

An ordered pair, one coordinate from each factor. Two pairs are equal exactly when both coordinates match — no interaction between the coordinates is assumed.

$\mathrm{lcm}(a,b)$Least common multiple

The smallest positive integer divisible by both aa and bb. Governs exactly the order of an element in a direct product, as Theorem 7.1 shows.

$G \oplus H$Direct sum

Alternate notation for G×HG\times H, used almost exclusively when both groups are abelian (and written additively). For two factors, ⊕\oplus and ×\times denote the identical construction — the different symbol is a stylistic signal, not a different operation.

$HK$Product set of two subgroups (recalled)

{hk:h∈H,k∈K}\{hk : h\in H, k\in K\}, from Chapter 06. Central to recognizing when a group already splits as a direct product internally, without building one from scratch.


01 · The External Direct Product

Definition — Direct Product

For groups G,HG,H, the direct product G×HG\times H is the set of ordered pairs {(g,h):g∈G,h∈H}\{(g,h):g\in G,h\in H\} with operation:

(g1,h1)∗(g2,h2):=(g1g2, h1h2)(g_1,h_1)\ast(g_2,h_2) := (g_1g_2,\, h_1h_2)

(each coordinate combined using its own group's operation).

Theorem 7.1 — G × H Is a Group

G×HG\times H is a group, with identity (eG,eH)(e_G,e_H), inverse (g,h)−1=(g−1,h−1)(g,h)^{-1}=(g^{-1},h^{-1}), and ∣G×H∣=∣G∣⋅∣H∣|G\times H| = |G|\cdot|H|.

Proof

Closure: coordinatewise, g1g2∈Gg_1g_2\in G and h1h2∈Hh_1h_2\in H, so (g1g2,h1h2)∈G×H(g_1g_2,h_1h_2)\in G\times H. Associativity: each coordinate's computation is associative independently, so the pair-wise operation is too. Identity: (g,h)∗(eG,eH)=(geG,heH)=(g,h)(g,h)\ast(e_G,e_H) = (ge_G,he_H) = (g,h), and likewise on the left. Inverses: (g,h)∗(g−1,h−1)=(gg−1,hh−1)=(eG,eH)(g,h)\ast(g^{-1},h^{-1}) = (gg^{-1},hh^{-1}) = (e_G,e_H), and likewise on the left. Order: distinct pairs correspond exactly to independent choices of a GG-coordinate (∣G∣|G| options) and an HH-coordinate (∣H∣|H| options), giving ∣G∣⋅∣H∣|G|\cdot|H| total pairs.

Example — $\mathbb{Z}_2 \times \mathbb{Z}_3$

Elements: (0,0),(0,1),(0,2),(1,0),(1,1),(1,2)(0,0),(0,1),(0,2),(1,0),(1,1),(1,2) — six pairs, matching ∣Z2×Z3∣=2×3=6|\mathbb{Z}_2\times\mathbb{Z}_3| = 2\times3=6. Operation is coordinatewise addition, each mod its own modulus: (1,1)+(1,2)=(1+1 mod 2, 1+2 mod 3)=(0,0)(1,1)+(1,2) = (1+1 \bmod 2,\ 1+2\bmod 3) = (0,0).


02 · Order of an Element in a Direct Product

Theorem 7.2 — Order Formula for Direct Products

For (g,h)∈G×H(g,h) \in G\times H with ∣g∣,∣h∣|g|,|h| both finite: ∣(g,h)∣=lcm(∣g∣,∣h∣)|(g,h)| = \mathrm{lcm}(|g|,|h|).

Proof

(g,h)k=(gk,hk)(g,h)^k = (g^k,h^k) by definition of the coordinatewise operation, repeated kk times. So (g,h)k=(eG,eH)(g,h)^k = (e_G,e_H) exactly when gk=eGg^k=e_G and hk=eHh^k=e_H simultaneously.

gk=eGg^k=e_G holds precisely when ∣g∣|g| divides kk (a standard fact: powers of gg that return to the identity are exactly the multiples of ∣g∣|g|, else the division algorithm applied to kk against ∣g∣|g| would produce a smaller positive power equal to ee, contradicting minimality of ∣g∣|g|). Similarly hk=eHh^k=e_H iff ∣h∣|h| divides kk.

So (g,h)k=(eG,eH)(g,h)^k=(e_G,e_H) iff kk is a common multiple of ∣g∣|g| and ∣h∣|h|. The smallest positive such kk is, by definition, lcm(∣g∣,∣h∣)\mathrm{lcm}(|g|,|h|).

Step-by-step — Order of $(2,1)$ in $\mathbb{Z}_4 \times \mathbb{Z}_6$
1
Find ∣2∣|2| in Z4\mathbb{Z}_4: 1⋅2=21\cdot2=2, 2⋅2=4≡02\cdot2=4\equiv0. So ∣2∣=2|2|=2.
2
Find ∣1∣|1| in Z6\mathbb{Z}_6: 11 generates all of Z6\mathbb{Z}_6 (Chapter 03), so ∣1∣=6|1|=6.
3
Apply Theorem 7.2: ∣(2,1)∣=lcm(2,6)|(2,1)| = \mathrm{lcm}(2,6).
4
Compute the lcm: multiples of 2: 2,4,6,8,…2,4,6,8,\ldots; multiples of 6: 6,12,…6,12,\ldots; smallest common one is 66.
5
Conclusion: ∣(2,1)∣=6|(2,1)| = 6 in Z4×Z6\mathbb{Z}_4\times\mathbb{Z}_6.
Common mistake — Guessing the order is the product $|g|\cdot|h|$

For (2,1)(2,1) above, ∣g∣⋅∣h∣=2×6=12≠6=lcm(2,6)|g|\cdot|h| = 2\times6=12 \neq 6=\mathrm{lcm}(2,6). The product overcounts whenever ∣g∣|g| and ∣h∣|h| share a common factor. Only when gcd⁡(∣g∣,∣h∣)=1\gcd(|g|,|h|)=1 does lcm(∣g∣,∣h∣)=∣g∣⋅∣h∣\mathrm{lcm}(|g|,|h|)=|g|\cdot|h| — exactly the coprimality condition that governs the next theorem.


03 · When Is Zm×Zn\mathbb{Z}_m \times \mathbb{Z}_n Cyclic?

Theorem 7.3 — $\mathbb{Z}_m \times \mathbb{Z}_n \cong \mathbb{Z}_{mn}$ iff $\gcd(m,n)=1$

Zm×Zn\mathbb{Z}_m \times \mathbb{Z}_n is cyclic (hence isomorphic to Zmn\mathbb{Z}_{mn}) if and only if gcd⁡(m,n)=1\gcd(m,n)=1.

Proof

(⇐\Leftarrow) Suppose gcd⁡(m,n)=1\gcd(m,n)=1. Consider (1,1)∈Zm×Zn(1,1)\in\mathbb{Z}_m\times\mathbb{Z}_n. By Theorem 7.2, ∣(1,1)∣=lcm(∣1∣,∣1∣)=lcm(m,n)|(1,1)| = \mathrm{lcm}(|1|,|1|) = \mathrm{lcm}(m,n). A standard identity relates lcm and gcd: lcm(m,n)⋅gcd⁡(m,n)=mn\mathrm{lcm}(m,n)\cdot\gcd(m,n) = mn, so lcm(m,n)=mn/gcd⁡(m,n)=mn/1=mn\mathrm{lcm}(m,n) = mn/\gcd(m,n) = mn/1 = mn. So ∣(1,1)∣=mn=∣Zm×Zn∣|(1,1)| = mn = |\mathbb{Z}_m\times\mathbb{Z}_n| (by Theorem 7.1's order formula). By Theorem 3.4, ∣⟨(1,1)⟩∣=mn|\langle(1,1)\rangle| = mn, matching the size of the whole group — so (1,1)(1,1) generates everything, and Zm×Zn\mathbb{Z}_m\times\mathbb{Z}_n is cyclic. Any two cyclic groups of the same finite order are isomorphic (both are, in structure, "the integers mod that order," matching generators to generators), so Zm×Zn≅Zmn\mathbb{Z}_m\times\mathbb{Z}_n \cong \mathbb{Z}_{mn}.

(⇒\Rightarrow) Suppose d=gcd⁡(m,n)>1d=\gcd(m,n)>1. For any (a,b)∈Zm×Zn(a,b)\in\mathbb{Z}_m\times\mathbb{Z}_n, the order of aa divides mm and the order of bb divides nn, so by Theorem 7.2, ∣(a,b)∣=lcm(∣a∣,∣b∣)|(a,b)| = \mathrm{lcm}(|a|,|b|) divides lcm(m,n)=mn/d<mn\mathrm{lcm}(m,n) = mn/d < mn (strict, since d>1d>1). So no element of Zm×Zn\mathbb{Z}_m\times\mathbb{Z}_n has order mnmn, meaning no element generates the whole group (order mnmn) — the group is not cyclic. Since Zmn\mathbb{Z}_{mn} is cyclic, and isomorphic groups share the property "is cyclic," Zm×Zn≇Zmn\mathbb{Z}_m\times\mathbb{Z}_n \not\cong \mathbb{Z}_{mn}.

Example — Distinguishing $\mathbb{Z}_4$ from $\mathbb{Z}_2 \times \mathbb{Z}_2$

Both have order 4, resolving Exercise 5.2's open question. Since gcd⁡(2,2)=2≠1\gcd(2,2)=2\neq1, Theorem 7.3 says Z2×Z2\mathbb{Z}_2\times\mathbb{Z}_2 is not cyclic — confirmed directly: every nonidentity element (1,0),(0,1),(1,1)(1,0),(0,1),(1,1) has order 22 (e.g. (1,1)+(1,1)=(0,0)(1,1)+(1,1)=(0,0)), never order 44. Meanwhile Z4\mathbb{Z}_4 has the element 11 of order 44. Two genuinely different groups of order 4 — these are, in fact, the only two groups of order 4 up to isomorphism, a small case of the classification machinery developed further in Chapter 09.


04 · Recognizing an Internal Direct Product

Building G×HG\times H from scratch is one direction. The reverse question — does a given group GG secretly split apart as a direct product of two of its own subgroups? — is often more useful.

Theorem 7.4 — Internal Direct Product Criterion

Let H,K⊴GH,K\trianglelefteq G with H∩K={e}H\cap K=\{e\} and HK=GHK=G. Then:

G  ≅  H×KG \;\cong\; H\times K

Proof

Step 1 — elements of HH and KK commute. For h∈H,k∈Kh\in H,k\in K, consider the element x=hkh−1k−1x=hkh^{-1}k^{-1}. Since K⊴GK\trianglelefteq G, hkh−1∈Khkh^{-1}\in K, so x=(hkh−1)k−1∈Kx=(hkh^{-1})k^{-1} \in K. Since H⊴GH\trianglelefteq G, kh−1k−1∈Hkh^{-1}k^{-1}\in H, so x=h(kh−1k−1)∈Hx = h(kh^{-1}k^{-1}) \in H. So x∈H∩K={e}x\in H\cap K=\{e\}, giving hkh−1k−1=ehkh^{-1}k^{-1}=e, i.e. hk=khhk=kh.

Step 2 — define φ:H×K→G\varphi: H\times K \to G by φ(h,k)=hk\varphi(h,k)=hk.

Homomorphism: φ(h1,k1)φ(h2,k2)=h1k1h2k2=h1h2k1k2\varphi(h_1,k_1)\varphi(h_2,k_2) = h_1k_1h_2k_2 = h_1h_2k_1k_2 (swapping k1,h2k_1,h_2 using Step 1) =φ(h1h2,k1k2)=φ((h1,k1)(h2,k2))=\varphi(h_1h_2,k_1k_2) = \varphi\big((h_1,k_1)(h_2,k_2)\big).

Injective: if φ(h,k)=e\varphi(h,k)=e, then hk=ehk=e, so h=k−1h=k^{-1}. But h∈Hh\in H and k−1∈Kk^{-1}\in K, so h∈H∩K={e}h\in H\cap K=\{e\}, giving h=eh=e and then k=h−1=ek=h^{-1}=e. So ker⁡φ={(e,e)}\ker\varphi=\{(e,e)\}; by Theorem 6.4, injective.

Surjective: HK=GHK=G means every g∈Gg\in G is hkhk for some h∈H,k∈Kh\in H,k\in K — exactly φ(h,k)\varphi(h,k).

φ\varphi is a bijective homomorphism: an isomorphism. G≅H×KG\cong H\times K.

Example — $\mathbb{Z}_6$ as an internal direct product

In Z6\mathbb{Z}_6, let H={0,3}H=\{0,3\} (order 2) and K={0,2,4}K=\{0,2,4\} (order 3), both normal (abelian group, Chapter 05). H∩K={0}H\cap K = \{0\} (their only shared element). HKHK: sums h+kh+k for h∈H,k∈Kh\in H,k\in K give 0,2,4,3,5,7≡10,2,4,3,5,7\equiv1 — all six elements of Z6\mathbb{Z}_6, so HK=Z6HK=\mathbb{Z}_6. By Theorem 7.4:

Z6  ≅  Z2×Z3\mathbb{Z}_6 \;\cong\; \mathbb{Z}_2\times\mathbb{Z}_3

consistent with Theorem 7.3, since gcd⁡(2,3)=1\gcd(2,3)=1.

Common mistake — Forgetting to check the trivial-intersection condition

Take G=Z4G=\mathbb{Z}_4, H={0,2}H=\{0,2\}, K={0,2}K=\{0,2\} (the same subgroup twice). HK={0,2}=H≠GHK = \{0,2\}=H\neq G already fails the second condition, but even setting that aside, H∩K={0,2}≠{0}H\cap K = \{0,2\} \neq \{0\} — the trivial-intersection condition fails badly (they're not even distinct subgroups). Theorem 7.4 requires both conditions; a group failing either does not split as the corresponding direct product, and indeed Z4≇Z2×Z2\mathbb{Z}_4 \not\cong \mathbb{Z}_2\times\mathbb{Z}_2 by Theorem 7.3's proof (Section 03).


05 · Direct Sums: A Note on Notation

Same construction, different symbol

For exactly two factors, G⊕HG\oplus H and G×HG\times H describe the identical group — the same set of pairs, the same coordinatewise operation. The ⊕\oplus symbol is conventionally reserved for abelian groups (almost always written additively), signaling to the reader "these pieces combine independently, coordinate by coordinate, with no cross-interaction," which is exactly the intuition behind ordinary addition. You will see Z2⊕Z2\mathbb{Z}_2\oplus\mathbb{Z}_2 and Z2×Z2\mathbb{Z}_2\times\mathbb{Z}_2 used interchangeably across different textbooks describing the exact same group from Section 03. The distinction matters more once infinitely many factors are combined — direct sums restrict to elements with only finitely many nonzero coordinates, while direct products allow all coordinates simultaneously — but for the finite constructions in this course, the two notations name the same object.


06 · Exercises

EXERCISE 7.1

Find the order of each coordinate separately first, then apply Theorem 7.2.

In Z3\mathbb{Z}_3: ∣2∣=3|2|=3 (since 2,4≡1,6≡02,4\equiv1,6\equiv0 — smallest positive multiple of 2 hitting 0 mod 3 is at k=3k=3). In Z4\mathbb{Z}_4: ∣3∣|3|: 3,6≡2,9≡1,12≡03,6\equiv2,9\equiv1,12\equiv0, so ∣3∣=4|3|=4.

By Theorem 7.2, ∣(2,3)∣=lcm(3,4)=12|(2,3)| = \mathrm{lcm}(3,4) = 12.

Find ∣(2,3)∣|(2,3)| in Z3×Z4\mathbb{Z}_3 \times \mathbb{Z}_4.

EXERCISE 7.2

Apply Theorem 7.3 directly — you only need to check gcd⁡(6,10)\gcd(6,10).

gcd⁡(6,10)=2≠1\gcd(6,10)=2\neq1. By Theorem 7.3, since the gcd is not 1, Z6×Z10\mathbb{Z}_6\times\mathbb{Z}_{10} is not cyclic, hence not isomorphic to Z60\mathbb{Z}_{60}.

Is Z6×Z10≅Z60\mathbb{Z}_6\times\mathbb{Z}_{10} \cong \mathbb{Z}_{60}? Justify using Theorem 7.3.

EXERCISE 7.3

Check the two conditions of Theorem 7.4 directly: is H∩K={0}H\cap K=\{0\}, and does HKHK give every element of Z12\mathbb{Z}_{12}?

H=⟨4⟩={0,4,8}H=\langle4\rangle=\{0,4,8\} (order 3), K=⟨6⟩={0,6}K=\langle6\rangle=\{0,6\} (order 2). Both are subgroups of the abelian group Z12\mathbb{Z}_{12}, hence normal.

H∩KH\cap K: comparing {0,4,8}\{0,4,8\} and {0,6}\{0,6\}, the only shared element is 00. So H∩K={0}H\cap K=\{0\}. ✓

HKHK: all sums h+kh+k: 0,4,8,6,10,14≡20,4,8,6,10,14\equiv2 — giving {0,2,4,6,8,10}\{0,2,4,6,8,10\}, only six elements, not all twelve. HK≠Z12HK\neq\mathbb{Z}_{12}. ✗

Since HK≠GHK\neq G, Theorem 7.4 does not apply — Z12\mathbb{Z}_{12} does not split as H×KH\times K for this particular choice. (This makes sense: ∣H∣⋅∣K∣=3×2=6≠12=∣G∣|H|\cdot|K|=3\times2=6\neq12=|G|, so H×KH\times K couldn't have matched GG's order in the first place.)

In Z12\mathbb{Z}_{12}, let H=⟨4⟩H=\langle4\rangle and K=⟨6⟩K=\langle6\rangle. Check whether the hypotheses of Theorem 7.4 hold for H,KH,K.

EXERCISE 7.4

Try H=⟨3⟩H=\langle3\rangle (order 4) and K=⟨4⟩K=\langle4\rangle (order 3) instead — check gcd⁡(4,3)\gcd(4,3) first to predict the outcome via Theorem 7.3.

H=⟨3⟩={0,3,6,9}H=\langle3\rangle = \{0,3,6,9\} (order 4, since ∣3∣=4|3|=4 by Theorem 3.4 and Section 04 of Chapter 03's divisor logic). K=⟨4⟩={0,4,8}K=\langle4\rangle=\{0,4,8\} (order 3).

H∩KH\cap K: comparing the two lists, only 00 is shared. H∩K={0}H\cap K=\{0\}. ✓

HKHK: since ∣H∣⋅∣K∣=4×3=12=∣Z12∣|H|\cdot|K|=4\times3=12=|\mathbb{Z}_{12}| and H∩KH\cap K is trivial, HKHK must have exactly 1212 elements (no overlap-driven repeats), so HK=Z12=GHK=\mathbb{Z}_{12}=G. ✓

Both conditions hold: Z12≅Z4×Z3\mathbb{Z}_{12} \cong \mathbb{Z}_4\times\mathbb{Z}_3 — consistent with gcd⁡(4,3)=1\gcd(4,3)=1 and Theorem 7.3.

Find a pair of subgroups H,K≤Z12H,K\le\mathbb{Z}_{12} that do satisfy Theorem 7.4's hypotheses, and state the resulting isomorphism.


07 · Chapter Summary

ConceptStatement
Direct product G×HG\times HPairs (g,h)(g,h); coordinatewise operation; $
Order of (g,h)(g,h)lcm(∣g∣,∣h∣)\mathrm{lcm}(\lvert g\rvert,\lvert h\rvert) (Thm 7.2)
Zm×Zn\mathbb{Z}_m\times\mathbb{Z}_n cyclicIff gcd⁡(m,n)=1\gcd(m,n)=1; then ≅Zmn\cong\mathbb{Z}_{mn} (Thm 7.3)
Internal direct productH,K⊴GH,K\trianglelefteq G, H∩K={e}H\cap K=\{e\}, HK=GHK=G   ⟹  G≅H×K\implies G\cong H\times K (Thm 7.4)
Direct sum ⊕\oplusSame as ×\times for two factors; conventional notation for abelian groups
Z4\mathbb{Z}_4 vs Z2×Z2\mathbb{Z}_2\times\mathbb{Z}_2Same order, genuinely different groups (one cyclic, one not)

Next: Chapter 08 — Group Actions studies how a group can move the elements of an entirely different set, unifying symmetry, counting, and conjugation under a single framework that powers the Sylow theorems in Chapter 09.