1

As the title says I am trying to prove $a \mathbb Z + b \mathbb Z = gcd(a,b) \mathbb Z$.

My steps so far:

$$a \mathbb Z + b \mathbb Z = \{ a z_1 + bz_2 : z_1, z_2 \in \mathbb Z \} = \{ kp z_1 + lpz_2 : z_1, z_2 \in \mathbb Z \}$$

where $p=gcd(a,b)$, $kp=a$ and $lp=b$.

$$\ldots = \{ p \cdot (k z_1 + l z_2) : z_1, z_2 \in \mathbb Z \} = p \{(k z_1 + l z_2) : z_1, z_2 \in \mathbb Z \}$$

So far, so good. What remains to show is that

$$\{(k z_1 + l z_2) : z_1, z_2 \in \mathbb Z \} = \mathbb Z $$

since then it would follow that

$$p \{(k z_1 + l z_2) : z_1, z_2 \in \mathbb Z \} = p \mathbb Z $$

but here I am stuck. I suspect it has something to do with the fact that $k$ and $l$ are coprime? Maybe there is a theorem about linear combinations of coprime numbers that somehow generates all integers? In any case, help is much appreciated!

  • 3
    Do you know about Bezout's identity? – AnilCh Jan 14 '21 at 09:19
  • We covered the euclidean algorithm, but I don't see how I would apply this here... I assume you would use it to replace $kz_1 + lz_2$ with something? – Rafael Bankosegger Jan 14 '21 at 09:21
  • 2
    It says that there exists some $z_1, z_2$ such that $kz_1 + lz_2 = gcd(k,l)$. In this case, because they're coprime, $kz_1 + lz_2 = 1$. Now it's not hard to see that you can obtain every integer from there. – AnilCh Jan 14 '21 at 09:27
  • Ah, okay thanks! But we could also apply it directly to the first line right? – Rafael Bankosegger Jan 14 '21 at 09:30
  • 1
    I guess, yeah. In the end this property is equivalent to Bezout's identity so once you obtain it you're almost done. If you haven't seen it explicitly in class, try to obtain it through the euclidean algorithm. – AnilCh Jan 14 '21 at 09:36
  • 1
    Thanks for the help. Unfortunately I can't mark your answer as the correct one since you didn't really write an answer. Still gave your comments some upvotes though :) – Rafael Bankosegger Jan 14 '21 at 09:45

3 Answers3

2

I think the most direct way to show this is by dual inclusion, as this is set equality after all. Let $d$ be the gcd of $a,b$. We want to show that $$ d\mathbb{Z} = a\mathbb{Z} + b\mathbb{Z} $$ Fix some $ax+by \in a\mathbb{Z} + b\mathbb{Z}$. Since $d|a$ and $d|b$ we have that: $$ ax+by = da'x+db'y = d(a'x+b'y) \in d \mathbb{Z} $$ for some $a',b' \in \mathbb{Z}$. Thus we have that $a\mathbb{Z} + b\mathbb{Z} \subseteq d \mathbb{Z}$.

Now fix some $dx \in d\mathbb{Z}$. By Bezout's identity, we know that there exists some $u,v \in \mathbb{Z}$ such that $ua+vb = d$, so we have that: $$ dx = (ua+vb)x = (ux)a + (uv)b \in a\mathbb{Z}+b\mathbb{Z} $$ so we have that $d\mathbb{Z} \subseteq a\mathbb{Z}+b\mathbb{Z}$ and thus the equality of sets.

Noah Solomon
  • 1,837
2

Sometimes Bezout's identity is derived from the equality $a\Bbb Z+b\Bbb Z =\gcd(a,b)\Bbb Z$, so here here is a proof that avoids that. (Of course it is very similar arguing and if you already proved Bezout's identity, e.g. using the euclidean algorithm, then the other solution is probably better).
By the euclidean algorithm $\Bbb Z$ is a PID, hence there is some $c\in\Bbb Z$ such that $$a\Bbb Z+b\Bbb Z=c\Bbb Z$$ We want to show that $c=\varepsilon\gcd(a,b)$ for some unit $\varepsilon\in\{-1,1\}$. Since $a,b\in c\Bbb Z$ we see $c\mid a,b$ so $c$ is a common divisor of $a,b$. We need to verify the universal property in the sense that for any other common divisor $d\mid a,b$ we have $d\mid c$. So let $d\mid a,b$. Then $a\Bbb Z,b\Bbb Z\subseteq d\Bbb Z$. Therefore $$c\Bbb Z=a\Bbb Z+b\Bbb Z\subseteq d\Bbb Z$$ This implies $d\mid c$. Hence $c$ is a greatest common divisor of $a,b$ and the claim follows.

leoli1
  • 6,979
  • 1
  • 12
  • 36
  • Worth emphasis: this boils down to specializing the definition of the ideal sum into a PID, and it is more insightful to explicitly view it this way, as I explain in my answer. – Bill Dubuque Jan 14 '21 at 11:11
0

It follows immediately by the (universal) definition of ideal sum and the fact that $\,\Bbb Z\,$ is a PID, i.e.

$\begin{align} \forall\ C\!:\:\! \ \ \ C\, \supseteq\ A,\,\ B\ &\iff\, C\,\supseteq \ A\,+\,B,\, \text{the def'n of ideal sum, becomes in a PID}\\[.5em] \forall\,(c)\!:\ (c) \supseteq (a),(b)&\iff (c)\supseteq (a)+(b) =: (d),\ \text{becomes in divisibility language}\\[.5em] \forall\, \ c\!:\,\ \ \ \ c\ \ \mid\,\ \ a,\,\ \ b\ \,&\color{#c00}\iff c\,\mid\, d,\,\ \text{by contains = divides for principal ideals}\\[.5em] \!\!\!\iff \gcd(a,b) = |d|\ \ \ {\rm [\,or}\ &\,d\,\ {\rm is\ associate\ to}\ \,\gcd(a,b)\,\ \text{in a general domain } D\,] \end{align}$

since $\ d\mid a,b\ $ by $\,c\!=\!d\;$ in $\,(\!\color{#c00}{\Longleftarrow}\!),\,$ so $\,|d|\,$ is a common divisor of $\,a,b,\,$ necessarily the greatest such since it is divisible by every common divisor $\,c\,$ by $\,(\!\color{#c00}{\Longrightarrow}\!)$

Remark $ $ The $\color{#c00}{\rm red}$ equivalence above is in fact the definition of a gcd in general domains, i.e. a gcd is a common divisor that is divisibly greatest, i.e. an infimum (greatest lower bound) in the poset induced by the divisibility partial order.

Bill Dubuque
  • 272,048