On the addition of residue classes mod p
On the addition of residue classes mod p
复制标题
DOI:
10.4064/aa-9-2-149-159
复制
发表时间:
1964
期刊:
影响因子:
0.7
通讯作者:
P. Erdös;H. Heilbronn
中科院分区:
文献类型:
--
作者:
P. Erdös;H. Heilbronn
In this paper we investigate the following question. Let p be a prime, a,, " ', cck distinct non-zero residue classes modp, N a residue class modp. denote Dhe number of solutions of the congruence ela,+... + ekak = N(modp) where the e,,. .. , ek are restricted to the values 0 and 1, What can be said about the function J?(N)? We prove two theorems. THEOREM I. 14 " (X) > 0 if k > 3 (6;~) " ~. THEOREM II. P(N) = 2kp-'(1+o(1)) $ k3pS2-+ 00 as p + co. Theorem I is almost best possible. Put al = ', '2 =-1, a3 = 2, a4 =-2,. .. . ckk = (Lx)k-'[&(k+l)]. Then it follows from an easy calc.ulation that F(S(p-1)) = 0 if k 1. p-too In the method of proof the two theorems differ considerably. The proof of Theorem I is elementary, depending entirely on the manipulation of residue classes mBdp, whereas the proof of Theorem II is based on the application of finite Fourier series and simple considerations on diophantine approximations. In an appendix we state various further conjectures which we are not able to prove.