TAOCP 1.2.4 Exercise 21

The existence of a factorization into primes follows by induction on $n$, using exercise 1.

Section 1.2.4: Integer Functions and Elementary Number Theory

Exercise 21. [M22] (Fundamental theorem of arithmetic.) Use Law B and exercise 1.2.1–5 to prove that every integer $n>1$ has a unique representation as a product of primes (except for the order of the factors). In other words, show that there is exactly one way to write

$$n=p_1p_2\cdots p_k,$$

where each $p_j$ is prime and $p_1 \le p_2 \le \cdots \le p_k$.

Verified: no
Solve time: 7m21s


Solution

The existence of a factorization into primes follows by induction on $n$, using exercise 1.2.1, 5.

If $n$ is prime, then $n$ itself is a product of primes.

If $n$ is composite, exercise 1.2.1, 5 implies that $n$ has a prime divisor $p$. Hence

$$ n=pm, $$

where $1<m<n$. By the induction hypothesis, $m$ can be written as a product of primes. Therefore $n$ can also be written as a product of primes. Rearranging the factors into nondecreasing order gives a representation

$$ n=p_1p_2\cdots p_k, \qquad p_1\le p_2\le\cdots\le p_k. $$

Thus every integer $n>1$ has at least one such representation.

It remains to prove uniqueness.

Suppose that

$$ n=p_1p_2\cdots p_k =q_1q_2\cdots q_\ell, $$

where

$$ p_1\le p_2\le\cdots\le p_k, \qquad q_1\le q_2\le\cdots\le q_\ell, $$

and all $p_i$ and $q_j$ are prime.

We shall show that

$$ k=\ell \quad\text{and}\quad p_i=q_i \ \ (1\le i\le k). $$

Since

$$ p_1 \mid q_1q_2\cdots q_\ell, $$

Euclid's lemma implies that $p_1$ divides one of the factors $q_j$. Here Euclid's lemma is obtained from Law B: if a prime $p$ divides $ab$ and does not divide $a$, then $\gcd(p,a)=1$, so Law B gives $p\mid b$. Repeating this argument yields the corresponding statement for any finite product.

Therefore $p_1\mid q_j$ for some $j$. Since $q_j$ is prime, its only positive divisors are $1$ and $q_j$. Because $p_1$ is also prime and $p_1\mid q_j$, we must have

$$ p_1=q_j. $$

We now show that in fact $j=1$.

Since the $q$'s are arranged in nondecreasing order,

$$ q_1\le q_j=p_1. $$

Similarly, because

$$ q_1\mid p_1p_2\cdots p_k, $$

the same argument shows that $q_1=p_i$ for some $i$. Since the $p$'s are arranged in nondecreasing order,

$$ p_1\le p_i=q_1. $$

Combining the inequalities gives

$$ p_1\le q_1\le p_1, $$

hence

$$ p_1=q_1. $$

Substituting this into

$$ p_1p_2\cdots p_k=q_1q_2\cdots q_\ell $$

and dividing both sides by the common nonzero factor $p_1=q_1$, we obtain

$$ p_2\cdots p_k=q_2\cdots q_\ell. $$

The same reasoning applied to this new equality yields

$$ p_2=q_2. $$

Dividing again gives

$$ p_3\cdots p_k=q_3\cdots q_\ell. $$

Continuing in this way, we obtain successively

$$ p_r=q_r $$

for every index $r$ for which both sides still contain factors.

Suppose, for example, that $k<\ell$. After cancelling the first $k$ common prime factors, we would obtain

$$ 1=q_{k+1}\cdots q_\ell, $$

which is impossible because each $q_j$ is a prime and therefore greater than $1$. Similarly, $k>\ell$ is impossible. Hence

$$ k=\ell. $$

Since the cancellation process yields $p_r=q_r$ for every $r=1,\dots,k$, we conclude that

$$ p_1=q_1,\quad p_2=q_2,\quad \ldots,\quad p_k=q_k. $$

Therefore the representation

$$ n=p_1p_2\cdots p_k, \qquad p_1\le p_2\le\cdots\le p_k, $$

is unique.

$\square$