27.12.2016 Views

nL;

QuantumV4N6

QuantumV4N6

SHOW MORE
SHOW LESS

Create successful ePaper yourself

Turn your PDF publications into a flip-book with our unique Google optimized e-Paper software.

Throuuh thg decimal poiltt<br />

Where the quadratic x2 - x has four roots<br />

by A. B. Zhiglevich and N. N. Petrovl<br />

a<br />

:l-'<br />

---<br />

:.:==<br />

:=:t:<br />

=.-=:<br />

=<br />

=o<br />

-u<br />

g)<br />

o<br />

a'\<br />

f<br />

o<br />

f<br />

C<br />

(n<br />

x<br />

E WILL TRY TO PERSUADE YOU THAT THE<br />

number of roots of the ecluationx2 = x is not two/<br />

as everybody thinks, but four, as stated in the<br />

title of this article. But first we offer the following<br />

undisputed fact. If a natural number (in decimal notation)<br />

ends in any of the digits 0, l, 5, or 5, its square<br />

ends in the same digit. This also applies to the two-digit<br />

endings 00, 01, 25, and75-thatis, if a number ends in<br />

one of these pairs of digits, its square does too (for instance,<br />

17 62 = 30,97 6j 2252 = 50,625 ; and so on). A similar<br />

property holds for the three-digit endings 000, 001,<br />

625, and376. At the same time, no other one-, two-, and<br />

three- digit endings possess this property. What's so<br />

special about these combinations of numbers? The<br />

answer is given by the following theorem.<br />

THroRru. For any natural k there are exactly four<br />

sets of k digits (00. . .00, 00 . . .01, and two more, ending<br />

in 5 and 6, respectively) such that if a natural number<br />

ends in one of these sets of digits, the square of this<br />

number ends in the same set of digits.<br />

To eliminate any doubts, we'llgive two proofs of this<br />

theorem. But first let's try to understand better what<br />

we have to prove.<br />

We're looking for an integer & 0 < x < 10&, such that<br />

for any integer a > O the square of the number 10&a + x<br />

(which is the general form of a number whose k-digtt<br />

ending coincides with x) has the form 10kb + x, where b<br />

is an integer. But (108a + xlz = tCf(tOka2 + 2axl + x2, so<br />

our condition simply amounts to the fact that x2 - x is<br />

divisible by 1S. Thus we've arrived at a very simple problem:<br />

to find all integers x, 0 < x < 1#, such that x2 - x is<br />

divisible by 1S. We must make sure that this probiem<br />

has four solutions, and that these solutions have the properties<br />

specified in the theorem.<br />

First proof . The difference * - x = x(x - 1) must be<br />

divisible by 10k : 2k5k. But the numbers x and x - 1<br />

lSome material by V. Denisenko was incorporated in<br />

this article with his permission.-Ed.<br />

can't both be divisible by 2 arrd can't both be divisible<br />

by 5. So we're left with four possibilities:<br />

(1) x is divisible by 10ft;<br />

(21 x - 1 is divisible by 10ft;<br />

(3) x is divisible by 2k andx - 1 is divisible by 5k;<br />

(4) x is divisible by 5k and x - I is divisible by 2k.<br />

Since x < 10&, the {irst possibility can occur only i{ x = 0;<br />

similarly, the second possibility corresponds to x = 1.<br />

Lr case (3 ) the numberx (and in case (4) the numberx - 1 )<br />

isoneof thenumbers a.2k, andsince x>x- 1>5k,<br />

we know that 0 < a < 5k - 1. All these numbers yield<br />

different remainders when dividedby 5k, because if the<br />

numbers Zkb and2kc give the same remainders upon<br />

division by 5k, then their difference 2k(c - b) is divisible<br />

by 5k, and so by 10k as well, which is impossible<br />

< 5k. But there are only 5k possible remain-<br />

for 0 < c -b<br />

ders, so each of them is represented once among our<br />

numbers 2ka. This means that exactly one of these<br />

numbers-say,Zkar-'has the remainder l, so that for<br />

x = Zkav the numberx - I is divisible by 5ft. Similarly,<br />

exactly one of these numbers-say, Zkar-has the remainder<br />

5k - 1 when divided by 5k, and-so, for x - I =<br />

2ka, the number x is divisible by 5k.<br />

This conclusively shows that in each of our four<br />

cases there is one and only one x satisfying our conditions,<br />

so the problem has exactly four solutions. It's<br />

also clear that in case (3) the number x ends in 6, and<br />

in case (4) it ends in 5. The proof is complete.<br />

Second proof . We'll take it for ganted that the problem<br />

has no more than four solutions. (We know this from<br />

the first proof, but it can be proved directly as we11.)Also,<br />

we'll use induction over k: we'll assume that we've already<br />

found the numbers xu_, and )27._ 1<br />

ending in 5 and<br />

6, respectively, such that 0 (xo_, . lG- I, 0 ( /r_ r<br />

. 1S-',<br />

and the differences x? _ t - xx _ , and yoz_ f yx _, are divisible<br />

by lgr- 1 (we create a basis for the inductionthat<br />

is, the existence of x, and yr- by setting xr = 5,<br />

0U[]tlIUil/rtlIltRt 17

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!