2

I am trying to prove the following problem ($p$ is a prime number, and $k \in \mathbb{P}$)

If $a, b$ are coprime, then if $p^k \, | \, ab$, then $p^k \,|\, a$ or $p^k \,|\, b$ without Fundamental Arithmetic Theorem.

I can derive that $p \, | \, ab$, $a \, | \, ab$, and $b \, | \, ab$. But I am lost as to where to go from there. If anybody could hint me in the right direction, that would be greatly appreciated.

Bill Dubuque
  • 272,048

5 Answers5

1

As $p^k|ab$, assume $p^\alpha|a$ and $p^\beta|b$, where $\alpha+\beta=k$. Unless either $\alpha=0$ or $\beta=0$, then $a$ and $b$ aren't coprime as stipulated.

JMP
  • 21,771
  • $\color{#c00}{{-}{\bf 1}},$ Not even a correct (complete) proof, let alone one that avoids FTA. You need to justify these claims, and the most common way uses FTA (or equivalents) - which OP seeks to avoid. – Bill Dubuque Oct 29 '21 at 17:44
0

Assume that $p \mid ab$ and that $p \not\mid a$. Then there are $x,y$ such that $$p^kx+ay=1$$ this is the Euclidean algorithm.

Multiply this equation by $b$, $$p^kxb+aby=b$$ then $p^k$ divides the left hand, so it divides $b$.

  • To get $px + ay = 1$, that means gcd(p, a) = 1 so you could write a linear combination of 1 from p and a. However, how did you get gcd(p, a) = 1? – Thomas Liu Oct 29 '21 at 01:49
  • I assume $p\not |a$ – Rene Schipperus Oct 29 '21 at 01:51
  • However, although from your answer you can conclude that $p , | , b$, the goal of the problem is to prove for this case that $p^k , | , b$. How do you get from $p , | , b$ to $p^k , | , b$? – Thomas Liu Oct 29 '21 at 02:22
  • 1
    @ThomasLiu Or there is a better way, I edited. – Rene Schipperus Oct 29 '21 at 02:34
  • If p does not divide a and p is prime, that means gcd(p, a) = 1. I still am confused as to how you got that $gcd(p^k, a) = 1$ to write that linear combination. – Thomas Liu Oct 29 '21 at 02:42
  • ERROR The argument assumes (without justification) that $p\nmid a\Rightarrow (p^k,a) = 1.,$ But common ways of justifying that do use FTA (or equivalent). So we can't judge if this is a correct proof avoiding FTA till you supply such justification. – Bill Dubuque Oct 29 '21 at 16:44
0

Proposition If $c | (ab)$ and $(c, a) = 1,$ then $c | b.$

Lemma Given $p$ a prime, if $c | p^k,$ then $c = p^i,\,0\le i\le k$


  • If $(p^k, a) = 1,$ then $p^k | b.$
  • If $(p^k, a) = p^k,$ then $p^k | a.$
  • If $(p^k, a) = p^i,\,0 < i < k.$ Suppose that $a = p^ic,\,(p^{k - i}, c) = 1,$ then, $p^k | ab\Leftrightarrow p^{k - i} | cb\Rightarrow p^{k - i} | b.$ Then $(a, b)\neq 1.$ Contradiction as $a$ and $b$ are coprime by assumption.

Lemma proof

Suppose that $c = p^ib,\, (b, p) = 1,$ as the only divisors of $p$ are $1$ and $p$ itself, then $c | p^k\Leftrightarrow b | p^{k - i}.$ Using the above Proposition, by induction, this eventually leads to $b | p.$ Then, $b = 1.$


Proposition proof As $(c, a) = 1,$ then $1$ is a linear combination of $c$ and $a,$ i.e. $1 = mc + na\Rightarrow b = mbc + n(ab).$ As $ab = ck,$ then $b = mbc + nck = c(mb + nk),$ so $c | b.$

0

A simple induction shows this is equivalent to EL = Euclid's Lemma (i.e. case $k\! =\! 1).\,$ Indeed, case $\,k\!=\!0\,$ is trivial & $\,k\!=\!1\,$ is EL, so assume $k> 1$. By EL: $\,p\mid p^k\mid ab\Rightarrow p\mid a\,$ or $\,b.\,$ Wlog $\,p\mid b.\,$ Cancel $p\Rightarrow p^{k-1}\mid \color{#c00}a(\color{#0a0}{b/p}).\,$ By induction $\,p^{k-1}\mid \color{#c00}a\,$ or $\,\color{#0a0}{b/p}$, but $\, p\mid p^{k-1}\mid\color{#c00} a\Rightarrow p\mid a,b\,$ contra $(a,b)\!=\!1.\,$ So it must be that $\,p^{k-1}\mid \color{#0a0}{b/p}\:\!\underset{\times\ p}\Longrightarrow p^k\mid b.\ $ QED

Due to this equivalence your question boils down to proving EL (case $k\!=\!1\,$) without FTA. But this is the most common way to prove EL and you can find most of these well-known proofs in prior answers here. The most common is to employ Bezout's GCD identity - see here. Alternatively we can give direct inductive proofs that essentially unwind the Euclidean descent inline, e.g. see here and here, which are employed in direct proofs of FTA (by Zermelo and others).

Or we can use $(p,a) = 1\Rightarrow (p^k,a)=1$ by here, so $\,p^k\mid ab\Rightarrow p^k\mid b\,$ by general EL here or here.

Remark $ $ See here for much more on the logical relationships between FTA and EL and closely related factorization properties.

Bill Dubuque
  • 272,048
  • @Downvoter Above is correct (and key to the matter). If something is not clear then please ask questions before downvoting. – Bill Dubuque Nov 05 '21 at 20:41
-1

Let $a=a_1^{k_1}×a_2^{k_2}×a_3^{k_3}×\dots a_m^{k_m}$(prime factorization of $a$) and $b=b_1^{p_1}×b_2^{p_2}×b_3^{p_3}×\dots b_n^{p_n}$(prime factorization of $b$) where $a_1,a_2,\dots a_m, b_1,b_2,\dots b_n$ are all distinct primes as $a$ and $b$ have no common factors.

If $p^k\,| \,ab$, then $p^k\,| \,(a_1^{k_1}×a_2^{k_2}×a_3^{k_3}×\dots a_m^{k_m}×b_1^{p_1}×b_2^{p_2}×b_3^{p_3}×\dots b_n^{p_n})$.

$p^k$ has no prime factor other than $p$. So $p^k$ must divide either $a_1^{k_1}$ or $a_2^{k_2}$ or $\dots $ or $a_m^{k_m}$ or $b_1^{p_1}$ or $b_2^{p_2}$ or $\dots$ or $b_n^{p_n}$.

If $p^k$ divides $a_x^{k_x}$, then $p^k\,| \,a$.

If $p^k$ divides $b_x^{p_x}$, then $p^k\,| \,b$

I have tried going as basic as I can. I hope now its clear to you.

An Alien
  • 516
  • 1
    This uses FTA (existence & uniqueness of prime factorizations) - which OP seeks to avoid, so this does not answer the question. – Bill Dubuque Oct 29 '21 at 17:37