Kvant

Rough automatic draft, low quality: expect mistakes and awkward wording. It will be replaced by a better translation. Original →

26

Recall that a number is called rational if it is equal to mn\frac{m}{n}, where nn and mm are integers, and rational otherwise. Everyone knows that the number 2\sqrt{2} is irrational. It is easy to prove a more general statement: if the natural number kk is not a perfect square, then the number k\sqrt{k} is irrational. Note that kk can be represented as b2ab^2a, where aa and bb are natural numbers, and aa is not divisible by the square of an integer (as they say, “free from squares”), that is, aa is equal to a prime number or the product of different prime numbers: a=p1p2...pra=p_1p_2...p_r; r≥1r \ge 1. The statement that the number k=ba\sqrt{k}=b\sqrt{a} is irrational properties the following: there are no non-zero integers b1b_1 and b2b_2 such that b1a+b2=0b_1\sqrt{a}+b_2=0. Indeed, if such b1b_1 and b2b_2 exist, then b12a=b22b_1^2a=b_2^2. But this equality contradicts the theorem on the uniqueness of the decomposition of a natural number into prime factors: p1p_1 enters the right side to an even degree, and the left side to an odd degree.

In this note we will prove that if b1,b2,...,bnb_1, b_2,..., b_n are non-zero integers, a1,a2,...,ana_1, a_2,..., a_n are square-free natural numbers, then b1a1+b2a2+⋯+bnan≠0b_1\sqrt{a_1}+b_2\sqrt{a_2}+\dots+b_n\sqrt{a_n} \neq 0.

Up to n=4n=4 this is easy to prove by contradiction by raising both parts (grouped approaches in a manner) into a square several times. In the general case, we will carry out a proof of induction on the number of distinct prime numbers p1,...,pNp_1,..., p_N that make at least one of the numbers aia_i. As is often the case, it is more convenient to prove by induction a stronger statement (interesting in itself):

there are integers c1,d1,c2,d2,...,ce,dec_1, d_1, c_2, d_2,..., c_e, d_e such that di≠0d_i \neq 0, ci≥1c_i \ge 1, all prime divisors of the numbers cic_i are contained among the numbers p1,...,pNp_1,..., p_N and the product

(d1c1+⋯+dece)×(b1a1+⋯+bnan)\left(d_1\sqrt{c_1}+\dots+d_e\sqrt{c_e}\right)\times\left(b_1\sqrt{a_1}+\dots+b_n\sqrt{a_n}\right)

is a non-zero integer.

Let us denote b1a1+⋯+bnanb_1\sqrt{a_1}+\dots+b_n\sqrt{a_n} by SS, and eliminate the sum d1c1+⋯+deced_1\sqrt{c_1}+\dots+d_e\sqrt{c_e} by S′S'.

If N=1N=1, then s has the form b1p1+b2p2−b2b_1\sqrt{p_1}+b_2\sqrt{p_2}-b_2: we proved above that SS′=b12p1−b22≠0SS'=b_1^2p_1-b_2^2 \neq 0. Let us now assume that N≥2N \ge 2 and that our statement is valid for all smaller values ​​of NN.

We will denote by letters S1,S2,...,S8S_1, S_2,..., S_8 sums of the form β1α1+⋯+βmαm\beta_1\sqrt{\alpha_1}+\dots+\beta_m\sqrt{\alpha_m}, where βi\beta_i are integers, αi\alpha_i are square-free natural numbers, all of whose prime divisors are contained among the numbers p1,...,pN−1;S1,S2,...,S8p_1,..., p_{N-1}; S_1, S_2,..., S_8, unless otherwise stated, can be equal to zero.

The sum SS can be written as S=S1+S2pNS=S_1+S_2\sqrt{p_N}, where S2≠0S_2 \neq 0. By the induction hypothesis, there is a sum S3S_3 such that f=S3S2f=S_3S_2 is a nonzero integer. Produced