Steps / Historical AI

Leonhard Euler · 1770 (1828 English ed.) · Internet Archive

Elements of Algebra

All 627 leaves · leaf 534

Transcription

CHAP. II. ADDITIONS. 497 wherefore $\frac{p}{q} - \frac{R}{s} = \pm \frac{1}{qs}$; and, since we suppose the doubtful sign to be taken conformably to that of the quantity $p-aq$, or $r$, the quantity $\frac{p}{q} - \frac{R}{s}$ must be positive, if $r$ be positive; and negative, if $r$ be negative: now, as $s < q$, and $R > P$ (hyp.), it is evident that $\frac{R}{s} > \frac{P}{q}$, (abstracting from the sign); therefore, the quantity $\frac{p}{q} - \frac{R}{s}$ will always have its sign different from that of $\frac{R}{s}$; that is to say, from that of $R$, since $s$ is positive; and, consequently, $p$ and $R$ will necessarily have different signs. This being laid down, we shall have, by substituting the above values of $y$ and $z$, $y - az = (p-aq)t + (r-as)u = rt + Ru$. Now $t$ and $u$ having different signs, as well as $p$ and $R$, it is evident, that $pt$ and $Ru$ will be quantities of like signs; therefore, since $t$ and $u$ are integer numbers, it is clear that the value of $y - az$ will always be greater than $r$; that is to say, than the value of $p-aq$, abstracting from the signs. But it remains to know whether, when the numbers $p$ and $q$ are given, we can always find numbers $r$ and $s$ less than those, and such that $ps-qr = \pm 1$, the doubtful signs being arbitrary; now, this follows evidently from the theory of continued fractions; but it may be demonstrated directly, and independently of that theory. For the difficulty is reduced to proving, that there necessarily exists an integer and positive number less than $p$, which being assumed for $r$, will make $qr \pm 1$ divisible by $p$. Now, suppose we successively substitute for $r$ the natural numbers $1, 2, 3$, &c. as far as $p$, and that we divide the numbers $q \pm 1, 2q \pm 1, 3q \pm 1$, &c. $pq \pm 1$ by $p$, we shall then have $p$ remainders less than $p$, which will necessarily be all different from one another; since, for example, if $mq \pm 1$, and $nq \pm 1$ ($m$ and $n$ being distinct integer numbers not exceeding $p$), when divided by $p$, give the same remainder, it is evident that their difference $(m-n)q$, must be divisible by $p$; now, this is impossible, because $q$ is prime to $p$, and $m-n$ is a number less than $p$. Therefore, since all the remainders in question are integer, positive numbers less than $p$, and different from each other, KK
confidence 0.97 59 expressions source scan at the Internet Archive

This text was produced by machine and has not been corrected by hand. Where it matters, read the scan. Confidence is the model's own, not an external check.