We extend two results of Ruzsa and Vu on the additive complements of primes.

Suppose that ${\lambda }_1,{\lambda }_2,{\lambda }_3,{\lambda }_4$ are nonzero real numbers, not all negative, $\delta >0$, $𝒱$ is a well-spaced set, and the ratio ${\lambda }_1/{\lambda }_2$ is algebraic and irrational. Denote by $E(𝒱,N,\delta )$ the number of $v\in 𝒱$ with $v\le N$ such that the inequality $$|{\lambda }_1{p}_1^2+{\lambda }_2{p}_2^3+{\lambda }_3{p}_3^4+{\lambda }_4{p}_4^5-v|<v^{-\delta }$$ has no solution in primes $p_1$, $p_2$, $p_3$, $p_4$. We show that $$E(𝒱,N,\delta )\ll N^{1+2\delta -1/72+\epsilon }$$ for any $\epsilon >0$.

1. Introduction. A positive number which is a sum of two odd primes is called a Goldbach number. Let E(x) denote the number of even numbers not exceeding x which cannot be written as a sum of two odd primes. Then the Goldbach conjecture is equivalent to proving that E(x) = 2 for every x ≥ 4. E(x) is usually called the exceptional set of Goldbach numbers. In [8] H. L. Montgomery and R. C. Vaughan proved that $E\left(x\right)=O\left(x^{1-\Delta }\right)$ for some positive constant Δ > 0$.In\left[3\right]ChenandPanprovedthatonecantake\Delta >0.01.In\left[6\right],weprovedthatE\left(x\right)=O\left(x^{0.921}\right)$. In this paper we prove the following result. Theorem....

We study the sum τ of divisors of the quadratic form m₁² + m₂² + m₃². Let $S₃\left(X\right)={\sum }_{1\le m₁,m₂,m₃\le X}\tau (m₁²+m₂²+m₃²)$. We obtain the asymptotic formula S₃(X) = C₁X³logX + C₂X³ + O(X²log⁷X), where C₁,C₂ are two constants. This improves upon the error term $O_{\epsilon }\left(X^{8/3+\epsilon }\right)$ obtained by Guo and Zhai (2012).

The object of this paper is to present new proofs of the classical ternary theorems of additive prime number theory. Of these the best known is Vinogradov's result on the representation of odd numbers as the sums of three primes; other results will be discussed later. Earlier treatments of these problems used the Hardy-Littlewood circle method, and are highly analytical. In contrast, the method we use here is a (technically) elementary deduction from the Siegel-Walfisz Prime Number Theory. It uses...

For a large odd integer N and a positive integer r, define b = (b₁,b₂,b₃) and $(N,r)=b\in \mathbb{N}³:1\le b_j\le r,(b_j,r)=1andb₁+b₂+b₃\equiv N\left(modr\right).$It is known that    $(N,r)=r²{\prod }_{p|r}p|N((p-1)(p-2)/p²){\prod }_{p|r}p\nmid N((p²-3p+3)/p²)$. Let ε > 0 be arbitrary and $R=N^{1/8-\epsilon }$. We prove that for all positive integers r ≤ R, with at most $O\left(Rlo{g}^{-A}N\right)$ exceptions, the Diophantine equation ⎧N = p₁+p₂+p₃, ⎨ $p_j\equiv b_j\left(modr\right),$ j = 1,2,3,$$⎩ with prime variables is solvable whenever b ∈ (N,r), where A > 0 is arbitrary.

Let $f\left(𝐩\right)$ be an additive form of degree $k$ with $s$ prime variables $p_1,p_2,\cdots ,p_s$. Suppose that $f$ has real coefficients ${\lambda }_i$ with at least one ratio ${\lambda }_i/{\lambda }_j$ algebraic and irrational. If s is large enough then $f$ takes values close to almost all members of any well-spaced sequence. This complements earlier work of Brüdern, Cook and Perelli (linear forms) and Cook and Fox (quadratic forms). The result is based on Hua’s Lemma and, for $k\ge 6$, Heath-Brown’s improvement on Hua’s Lemma.