Leonhard Euler · 1770 (1828 English ed.) · Internet Archive
Elements of Algebra
All 627 leaves · leaf 596
Transcription
CHAP. VII.
ADDITIONS.
559
equation; for we shall only have to take, for $v$ and $\xi$, the corresponding terms of the series $p'$, $p''$, $p'''$, &c. $q'$, $q''$, $q'''$, &c. calculated according to the formulæ of Art. 25; and we may even find an infinite number of satisfactory values for $v$ and $\xi$, by continuing the same series to infinity.
Now, as soon as we know two values of $v$ and $\xi$, we shall have, from the equation, $v = M\psi - N\xi$, that of $\psi$, which will also be a whole number; then we may go back from these values of $\xi$ and $\psi$, that is to say, of $\theta_{p+1}$, and $\theta_p$, to those of $\theta$ and $\theta'$, or of $y$ and $z$ (Art. 70).
But if in the series $P'$, $P''$, $P'''$, &c. there is no term $= M$, we are sure that the equation proposed admits of no solution in whole numbers.
It is proper to observe, that, as the series $p^0$, $p'$, $p''$, &c. as well as the two others, $q^0$, $q'$, $q''$, &c. and $\mu$, $\mu'$, $\mu''$, &c. depend only on the number A; the calculation, once made for a given value of A, will serve for all the equations in which A, or $n^2 - CB$, shall have the same value; and hence the foregoing method is preferable to that of Art. 68, which requires a new calculation for each equation.
Lastly, so long as A does not exceed 100, we may make use of the Table given, Art. 41, which contains for each radical $\sqrt{A}$, the values of the terms of the two series $p^0$, $p'$, $p''$, $p'''$, &c. and $\mu$, $\mu'$, $\mu''$, &c. continued, until one of the terms $P'$, $P''$, $P'''$, &c. becomes $= 1$; after which, all the succeeding terms of both series return in the same order. So that, by means of this Table, we may judge, immediately, whether the equation, $v^2 - A\xi^2 = M$, be resolvible, or not.
*Of the manner of finding all the possible solutions of the equation, $cy^2 - 2nyz + Bz^2 = 1$, when we know only one of them.*
72. Though, by the methods just given, we may successively find all the solutions of this equation, when it is resolvible in integer numbers; yet this may be done, in a manner still more simple, as follows:
Call $p$ and $q$ the values found for $y$ and $z$; so that we have
$cp^2 - 2npq + Bq^2 = 1$,
and take two other whole numbers, $r$ and $s$, such, that $ps - qr = 1$; which is always possible, because $p$ and $q$ are necessarily prime to each other; then suppose
$y = pt + ru$, and $z = qt + su$,
$t$ and $u$ being two new indeterminate numbers; substituting these expressions in the equation,
$cy^2 - 2nyz + Bz^2 = 1$,
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.