Abstract
Let N be a sufficiently large integer. In this paper, it is proved that, with at most \(O(N^{4/27+\varepsilon })\) exceptions, all even positive integers up to N can be represented in the form \(p_1^2+p_2^2+p_3^3+p_4^3+p_5^6+p_6^6\), where \(p_1,p_2,p_3,p_4,p_5,p_6\) are prime numbers. This gives a large improvement of a recent result \(O(N^{127/288+\varepsilon })\) due to Liu (Proc. Indian Acad. Sci. (Math. Sci.) 130(1) (2020) Article ID. 8).
Similar content being viewed by others
Avoid common mistakes on your manuscript.
1 Introduction and main result
Let \(N,\,k_1,\,k_2,\ldots ,\,k_s\) be natural numbers such that \(2\leqslant k_1\leqslant k_2\leqslant \cdots \leqslant k_s,\,N>s\). Waring’s problem of mixed powers concerns the representation of N as the form
Almost since the invention of the circle method by Hardy and Littlewood nearly a century ago, it has been folklore that the method fails to establish the solubility of problems of Waring type when the sum of the reciprocals of the exponents does not exceed 2. As articulated in [3], this convexity barrier ‘arises from the relative sizes of the product of local densities associated with the system, and the square–root of the available reservoir of variables that is a limiting feature of associated exponential sum estimates’, and has been circumvented in very few cases by other devices. A variant of Waring’s problem in which one considers mixed sums of squares, cubes and higher powers have provided a rich environment for testing methods designed to approach this theoretical limit of the circle method. A problem of this type that fails to be accessible to the circle method by the narrowest of margins is the notorious one of representing integers as sums of two squares, two positive integral cubes and two sixth powers. In 2013, Wooley [14] applied the method of Golubeva [5] to show, subject to the truth of the generalized Riemann hypothesis (GRH), that all large integers are thus represented. However, Wooley’s work [14] employs representations of special type and fails to deliver the anticipated asymptotic formula for their total number. Also, Wooley [16] showed that, although the expected asymptotic formula may occasionally fail to hold, the set of such exceptional instances is extraordinarily sparse. Afterwards, Lü and Mu [8] refined the result of Wooley [16].
In view of the result of Wooley [14], it is reasonable to conjecture that, for every sufficiently large even integer N, the following Diophantine equation
is solvable. Here and below, the letter p, with or without subscript, always stands for a prime number. But this conjecture is perhaps out of reach at present. In 2020, Liu [7] considered the exceptional set of the representation (1.1). In [7], Liu mainly use the arguments of Wooley [15] and showed that \(E(N)\ll N^{\frac{127}{288}+\varepsilon }\), where E(N) denotes the number of positive even integers n up to N, which can not be represented as \(p_1^2+p_2^2+p_3^3+p_4^3+p_5^6+p_6^6\).
In this paper, we shall continue to improve the estimate of the exceptional set for the problem (1.1) and establish the following result.
Theorem 1.1
Let E(N) denote the number of positive even integers n up to N, which can not be represented as
Then, for any \(\varepsilon >0\), we have
Remark. In order to compare the result of Theorem 1.1 with that of Liu [7], we list the numerical result as follows:
We will establish Theorem 1.1 by using the method, which is created and developed by Kawada and Wooley [9], to bound the set of exceptional integers not represented by a given additive form in terms of the exceptional set corresponding to a subform. For the exceptional set corresponding to a subform, we shall employ pruning process into the Hardy–Littlewood circle method. In the treatment of the integrals over minor arcs, we will employ the methods, which is developed by Brüdern [4], combining with the new estimates for exponential sum over primes developed by Zhao [18] and Kumchev [10]. The full details will be explained in the following revelant sections.
Notation. Throughout this paper, let p, with or without subscripts, always denote a prime number; \(\varepsilon \) always denotes a sufficiently small positive constant, which may not be the same at different occurrences. As usual, we use \(\varphi (n)\) and d(n) to denote the Euler’s function and Dirichlet’s divisor function, respectively. Also, we use \(\chi \bmod q\) to denote a Dirichlet character modulo q, and \(\chi ^0\bmod q\) the principal character. \(e(x)=e^{2\pi i x}\); \(f(x)\ll g(x)\) means that \(f(x)=O(g(x))\); \(f(x)\asymp g(x)\) means that \(f(x)\ll g(x)\ll f(x)\). N is a sufficiently large integer and \(n\in [N/2,N]\), and thus \(\log N\asymp \log n\). The letter c, with or without subscripts or superscripts, always denote a positive constant.
2 Preliminary and outline of the proof of Theorem 1.1
In order to better illustrate Lemma 2.1 and Lemma 2.2 below, we first introduce some notations and definitions. When \(\mathcal {C}\subseteq {\mathbb {N}}\), we write \(\overline{\mathcal {C}}\) for the complement \({\mathbb {N}}\setminus \mathcal {C}\) of \(\mathcal {C}\) within \({\mathbb {N}}\). When a and b are non-negative integers, it is convenient to denote by \((\mathcal {C})_a^b\) the set \(\mathcal {C}\cap (a,b]\), and by \(|\mathcal {C}|_a^b\) the cardinality of \(\mathcal {C}\cap (a,b]\). Next, when \(\mathcal {C},\mathcal {D}\subseteq {\mathbb {N}}\), we define
It is convenient, when k is a natural number, to describe a subset \(\mathcal {Q}\) of \({\mathbb {N}}\) as being a high-density subset of the k-th powers when (i) one has \(\mathcal {Q}\in \{n^k:n\in {\mathbb {N}}\}\), and (ii) for each positive number \(\varepsilon \), whenever N is a natural number sufficiently large in terms of \(\varepsilon \), then \(|\mathcal {Q}|^N_0>N^{1/k-\varepsilon }\). Also, when \(\theta >0\), we shall refer to a set \(\mathcal {R}\subseteq {\mathbb {N}}\) as having complementary density growth exponent smaller than \(\theta \) when there exists a positive number \(\delta \) with the property that, for all sufficiently large natural numbers N, one has \(|\overline{\mathcal {R}}|_0^N<N^{\theta -\delta }\).
When q is a natural number and \(\mathfrak {a}\in \{0,1,\dots ,q-1\}\), we define \(\mathcal {P}_\mathfrak {a}=\mathcal {P}_{\mathfrak {a},q}\) by
Also, we describe a set \(\mathcal {L}\) as being a union of arithmetic progressions modulo q when, for some subset \(\mathfrak {L}\) of \(\{0,1,\dots ,q-1\}\), one has
In such circumstances, given a subset \(\mathcal {C}\) of \({\mathbb {N}}\) and integers a and b, it is convenient to write
Let \(\mathcal {L}\) be a union of arithmetic progressions modulo q, for some natural number q. When k is a natural number, we describe a subset \(\mathcal {Q}\) of \({\mathbb {N}}\) as being a high-density subset of the k-th powers relative to \(\mathcal {L}\) when (i) one has \(\mathcal {Q}\in \{n^k:n\in {\mathbb {N}}\}\), and (ii) for each positive number \(\varepsilon \), whenever N is a natural number sufficiently large in terms of \(\varepsilon \), then \(\langle \mathcal {Q}\wedge \mathcal {L}\rangle _0^N\gg _qN^{1/k-\varepsilon }\). In addition, when \(\theta >0\), we shall refer to a set \(\mathcal {R}\subseteq {\mathbb {N}}\) as having \(\mathcal {L}\)-complementary density growth exponent smaller than \(\theta \) when there exists a positive number \(\delta \) with the property that, for all sufficiently large natural numbers N, one has \(|\overline{\mathcal {R}}\cap \mathcal {L}|_0^N<N^{\theta -\delta }\).
Lemma 2.1
Let \(\mathcal {L}\), \(\mathcal {M}\) and \(\mathcal {N}\) be unions of arithmetic progressions modulo q, for some natural number q, and suppose that \(\mathcal {N}\subseteq \mathcal {L}+\mathcal {M}\). Suppose also that \(\mathcal {S}\) is a high-density subset of the squares relative to \(\mathcal {L}\), and that \(\mathcal {A}\subseteq {\mathbb {N}}\) has \(\mathcal {M}\)-complementary density growth exponent smaller than 1. Then, whenever \(\varepsilon >0\) and N is a natural number sufficiently large in terms of \(\varepsilon \), one has
Proof
See Theorem 2.2 of Kawada and Wooley [9]. \(\square \)
Lemma 2.2
Let \(\mathcal {L}\), \(\mathcal {M}\) and \(\mathcal {N}\) be unions of arithmetic progressions modulo q, for some natural number q, and suppose that \(\mathcal {N}\subseteq \mathcal {L}+\mathcal {M}\). Suppose also that \(\mathcal {C}\) is a high-density subset of the cubes relative to \(\mathcal {L}\), and that \(\mathcal {A}\subseteq {\mathbb {N}}\) has \(\mathcal {M}\)-complementary density growth exponent smaller than \(\theta \), for some positive number \(\theta \). Then, whenever \(\varepsilon >0\) and N is a natural number sufficiently large in terms of \(\varepsilon \), without any condition on \(\theta \), one has
Proof
See Theorem 4.1(a) of Kawada and Wooley [9]. \(\square \)
In order to prove Theorem 1.1, we need the following proposition, whose proof will be given in Section 3.
PROPOSITION 2.3
Let \(E_1(N)\) denote the number of positive integers n up to N, which satisfies \(n\equiv 0\pmod 2,n\equiv \pm 1\pmod 3\), and can not be represented as \(p_1^2+p_2^3+p_3^6+p_4^6\). Then, for any \(\varepsilon >0\), we have
Proof of Theorem 1.1
Let
Thus, we have \(E_1(N)=|\mathscr {E}_1|_0^N\) and \(E(N)=|\mathscr {E}|_0^N\). Also, we write \(E_2(N)=|\mathscr {E}_2|_0^N\). Then \(\mathcal {L}_1\) is a union of arithmetic progression modulo 24, \(\mathcal {N}_2\) is a union of arithmetic progression modulo 2, and \(\mathcal {M}_1,\mathcal {N}_1\) and \(\mathcal {L}_2\) are unions of arithmetic progressions modulo 6, satisfying the condition that \(\mathcal {N}_1\subseteq \mathcal {L}_1+\mathcal {M}_1\) and \(\mathcal {N}_2\subseteq \mathcal {L}_2+\mathcal {N}_1\). Moreover, it follows from the Prime Number Theorem in arithmetic progression that
Therefore, \(\mathcal {S}_1\) is a high-density subset of the squares relative to \(\mathcal {L}_1\), and \(\mathcal {S}_2\) is a high-density subset of the cubes relative to \(\mathcal {L}_2\). By Proposition 2.3, it is easy to see that
and thus \(\overline{\mathcal {A}_1}\) has \(\mathcal {M}_1\) complementary density growth exponent smaller than 1. From Lemma 2.1, we know that
Let the integers \(N_j\) for \(j\geqslant 0\) by means of the iterative formula
where \(\lceil N\rceil \) denotes the least integer not smaller than N. Moreover, we define J to be the least positive integer with the property that \(N_j\leqslant 10\), then \(J\ll \log N\). Therefore, there holds
By (2.2), we know that
and thus \(\mathcal {A}_2\) has \(\mathcal {N}_1\)-complementary density growth exponent smaller than \(\frac{1}{2}\). From Lemma 2.2, we obtain
Therefore, with the same notation as in (2.1), we deduce that
which completes the proof of Theorem 1.1.
3 Outline of the proof of Proposition 2.3
In this section, we shall give the outline of the proof of Proposition 2.3. Let N be a sufficiently large positive integer. For \(k=2,3,6\), we define
where \(X_k=(N/16)^{\frac{1}{k}}\). Let
Then for any \(Q>0\), it follows from orthogonality that
In order to apply the circle method, we set
By Dirichlet’s lemma on rational approximation (for instance, see Lemma 2.1 of Vaughan [12]), each \(\alpha \in [1/Q,1+1/Q]\) can be written in the form
for some integers \(a,\,q\) with \(1\leqslant a\leqslant q\leqslant Q\) and \((a,q)=1\). Then we define the major arcs \(\mathfrak {M}\) and minor arcs \({\mathfrak {m}}\) as follows:
where
Then one has
In order to prove Proposition 2.3, we need the two following propositions, whose proofs will be given in Sects. 4 and 6, respectively.
PROPOSITION 3.1
Let the major arcs \(\mathfrak {M}\) be defined as in (3.2) with P and Q defined in (3.1). Then, for \(n\in (N/2,N]\) and any \(A>0\), there holds
where \(\mathfrak {S}(n)\) is the singular series defined in (4.1) below, which is absolutely convergent and satisfies
for any integer n satisfying \(n\equiv 0\,(\bmod \,2)\) and \(n\equiv \pm 1\,(\bmod \,3)\), and some fixed constant \(c^*>0\); while \(\mathfrak {J}(n)\) is defined by (4.9) and satisfies
For the properties (3.3) of singular series, we shall give the proof in Section 5.
PROPOSITION 3.2
Let the minor arcs \({\mathfrak {m}}\) be defined as in (3.2) with P and Q defined in (3.1). Then we have
The remaining part of this section is devoted to establishing Proposition 2.3 by using Propositions 3.1 and 3.2.
Proof of Proposition 2.3
Let \(\mathscr {U}(N)\) denote the set of integers \(n\in (N/2,N]\) such that
Then we have
By Bessel’s inequality, we have
Combining (3.4), (3.5) and Proposition 3.2, we have
Therefore, with at most \(O(N^{1-1/54+\varepsilon })\) exceptions, all the integers \(n\in (N/2,N]\) satisfy
Using the above equation and Proposition 3.1, we deduce that, with at most \(O(N^{1-1/54+\varepsilon })\) exceptions, all the positive integers \(n\in (N/2,N]\) satisfying \(n\equiv 0\,(\bmod 2)\) and \(n\equiv \pm 1\,(\bmod 3)\) can be represented in the form \(p_1^2+p_2^3+p_3^6+p_4^6\), where \(p_1,p_2,p_3,p_4\) are prime numbers. By a splitting argument, we obtain
This completes the proof of Proposition 2.3.
4 Proof of Proposition 3.1
In this section, we shall concentrate on proving Proposition 3.1. We first introduce some notations. For a Dirichlet character \(\chi \bmod q\) and \(k\in \{2,3,6\}\), we define
where \(\chi ^0\) is the principal character modulo q. Let \(\chi _2,\chi _3,\chi _6^{(1)},\chi _6^{(2)}\) be Dirichlet characters modulo q. Define
and write
Lemma 4.1
For \((a,q)=1\) and any Dirichlet character \(\chi \bmod q\), there holds
with \(\beta _k=(\log k)/\log 2\).
Proof
See Problem 14 of Chapter VI of [13]. \(\square \)
Lemma 4.2
The singular series \(\mathfrak {S}(n)\) satisfies (3.3).
The proof of Lemma 4.2 is given in Section 5.
Lemma 4.3
Let f(x) be a real differentiable function in the interval [a, b]. If \(f'(x)\) is monotonic and satisfies \(|f'(x)|\leqslant \theta <1\), then we have
Proof
See Lemma 4.8 of [11]. \(\square \)
Lemma 4.4
Let \(\chi _2 \bmod r_2\), \(\chi _3 \bmod r_3\) and \(\chi _6^{(i)} \bmod r_6^{(i)}\) with \(i=1,2\) be primitive characters, \(r_0=\big [r_2,r_3,r_6^{(1)},r_6^{(2)}\big ]\), and \(\chi ^0\) the principal character modulo q. Then there holds
Proof
By Lemma 4.1, we have
Therefore, the left-hand side of (4.2) is
This completes the proof of Lemma 4.4. \(\square \)
Write
where \(\delta _\chi =1\) or 0 according to whether \(\chi \) is principal or not. Then by the orthogonality of Dirichlet characters, for \((a,q)=1\), we have
For \(j=1,2,\dots ,12\), we define the sets \(\mathscr {S}_j\) as follows:
Also, we write \(\overline{\mathscr {S}_j}= \{2,3,6,6\}\setminus \mathscr {S}_j\). Then we have
where
In the following content of this section, we shall prove that \(I_1\) produces the main term, while the others contribute to the error term.
For \(k=2,3,6\), applying Lemma 4.3 to \(V_k(\lambda )\), we have
Putting (4.5) into \(I_1\), we see that
By using the elementary estimate
and Lemma 4.4 with \(r_0=1\), the O-term in (4.6) can be estimated as
If we extend the interval of the integral in the main term of (4.6) to \([-1/2,1/2]\), then from (3.1) we can see that the resulting error is
for some \(\varpi >0\). Therefore, by Lemma 4.2, (4.6) becomes
where
In order to estimate the contribution of \(I_j\) for \(j=2,3,\ldots ,12\), we shall need the following three preliminary lemmas, i.e. Lemmas 4.5–4.7, whose proofs are exactly the same as Lemmas 3.5–3.7 in Zhang and Li [17], so we omit the details herein. In view of this, for \(k=2,3,6\), we recall the definition of \(W_k(\chi ,\lambda )\) in (4.3) and write
and
Here and below, \(\Sigma ^*\) indicates that the summation is taken over all primitive characters.
Lemma 4.5
Let \(P,\,Q\) be defined as in (3.1). For \(k=6\), we have
Lemma 4.6
Let \(P,\,Q\) be defined as in (3.1). Then we have
Lemma 4.7
Let \(P,\,Q\) be defined as in (3.1). Then, for any \(A>0\), we have
Now, we concentrate on estimating the terms \(I_j\) for \(j=2,3,\ldots ,12\). We begin with the term \(I_{12}\), which is the most complicated one. Reducing the Dirichlet characters in \(I_{12}\) into primitive characters, we have
where \(\chi ^0\) is the principal character modulo q and \(r_0=[r_2,r_3,r_6^{(1)},r_6^{(2)}]\). For \(q\leqslant P\) and \(X_k<p\leqslant 2X_k\) with \(k=2,3,6\), we have \((q,p)=1\). From this and the definition of \(W_k(\chi ,\lambda )\), we obtain \(W_2\big (\chi _2\chi ^0,\lambda \big )=W_2\big (\chi _2,\lambda \big )\), \(W_3\big (\chi _3\chi ^0,\lambda \big )=W_3\big (\chi _3,\lambda \big )\) and \(W_6\big (\chi _6^{(i)}\chi ^0,\lambda \big )=W_6\big (\chi _6^{(i)},\lambda \big )\) for primitive characters \(\chi _2,\chi _3,\chi _6^{(i)}\) with \(i=1,2\). Therefore, by Lemma 4.4, we obtain
In the last integral, we pick out \(\big |W_2(\chi _2,\lambda )\big |\) and \(\big |W_3(\chi _3,\lambda )\big |\), and then use Cauchy’s inequality to derive that
Now we introduce the iterative procedure to bound the sums over \(r_6^{(2)},r_6^{(1)},r_3,r_2\), consecutively. We first estimate the above sum over \(r_6^{(2)}\) in (4.10) via Lemma 4.5. Since
the sum over \(r_6^{(2)}\) is
Again, by Lemma 4.5, the contribution of the quantity on the right-hand side of (4.11) to the sum over \(r_6^{(1)}\) in (4.10) is
By Lemma 4.6, the contribution of the quantity on the right-hand side of (4.12) to the sum over \(r_3\) in (4.10) is
Finally, from Lemma 4.7, inserting the bound on the right-hand side of (4.13) to the sum over \(r_2\) in (4.10), we get
For the estimation of the terms \(I_2,I_3,\ldots ,I_{11}\), by noting (4.5) and (4.7), we obtain
Using this estimate and the upper bound of \(V_k(\lambda )\), which derives from (4.5) and (4.7), that \(V_k(\lambda )\ll N^{\frac{1}{k}}\), we can argue similarly to the treatment of \(I_{12}\) and obtain
Combining (4.4), (4.8), (4.14) and (4.15), we can derive the conclusion of Proposition 3.1.
5 The singular series
In this section, we shall investigate the properties of the singular series which appear in Proposition 3.1.
Lemma 5.1
Let p be a prime and \(p^\alpha \Vert k\). For \((a,p)=1\), if \(\ell \geqslant \gamma (p)\), we have \(C_k(p^\ell ,a)=0\), where
Proof
See Lemma 8.3 of [6]. \(\square \)
For \(k\geqslant 1\), we define
Lemma 5.2
Suppose that \((p,a)=1\). Then
where \(\mathscr {A}_k\) denotes the set of non-principal characters \(\chi \) modulo p for which \(\chi ^k\) is principal, and \(\tau (\chi )\) denotes the Gauss sum
Also, there hold \(|\tau (\chi )|=p^{1/2}\) and \(|\mathscr {A}_k|=(k,p-1)-1\).
Proof
See Lemma 4.3 of [12]. \(\square \)
Lemma 5.3
For \((p,n)=1\), we have
Proof
We denote by \(\mathcal {S}\) the left-hand side of (5.1). By Lemma 5.2, we have
If \(|\mathscr {A}_k|=0\) for some \(k\in \{2,3,6\}\), then \(\mathcal {S}=0\). If this is not the case, then
From Lemma 5.2, the quadruple outer sums have not more than 50 terms. In each of these terms, we have
Since in any one of these terms \(\overline{\chi _2(a)\chi _3(a)\chi _6^{(1)}(a)\chi _6^{(2)}(a)}\) is a Dirichlet character \(\chi \,(\bmod p)\), the inner sum is
From the fact that \(\tau (\chi ^0)=-1\) for principal character \(\chi ^0\bmod p\), we have
By the above arguments, we obtain
This completes the proof of Lemma 5.3. \(\square \)
Lemma 5.4
Let \(\mathscr {L}(p,n)\) denote the number of solutions of the following congruence;
Then, for \(n\equiv 0\,(\bmod \,2)\) and \(n\equiv \pm 1\,(\bmod \,3)\), we have \(\mathscr {L}(p,n)>0\).
Proof
We have
where
By Lemma 5.2, we obtain
It is easy to check that \(|E_p|<(p-1)^4\) for \(p\geqslant 67\). Therefore, we obtain \(\mathscr {L}(p,n)>0\) for \(p\geqslant 67\). For \(p=2,3,5,7,11,13,17,19,23,29,31,37,41,43,47,53,59,61\), we can check \(\mathscr {L}(p,n)>0\) directly. This completes the proof of Lemma 5.4. \(\square \)
Lemma 5.5
A(n, q) is multiplicative in q.
Proof
By the definition of A(n, q) in (4.1), we only need to show that B(n, q) is multiplicative in q. Suppose \(q=q_1q_2\) with \((q_1,q_2)=1\). Then we have
For \((q_1,q_2)=1\) and \(k\in \{2,3,6\}\), there holds
Putting (5.3) into (5.2), we deduce that
This completes the proof of Lemma 5.5. \(\square \)
Lemma 5.6
Let \(A(n\mathrm{,}q)\) be as defined in (4.1).
(i) We have
and thus the singular series \(\mathfrak {S}(n)\) is absolutely convergent and satisfies \(\mathfrak {S}(n)\ll d(n)\);
(ii) There exists an absolute positive constant \(c^*>0\), such that, for \(n\equiv 0\,(\bmod \,2)\) and \(n\equiv \pm 1\,(\bmod \,3)\),
Proof
From Lemma 5.5, we know that B(n, q) is multiplicative in q. Therefore, there holds
From (5.4) and Lemma 5.1, we deduce that \(B(n,q)=\prod _{p\Vert q}B(n,p)\) or 0 according to q being square-free or not. Thus, one has
Write
Then
Applying Lemma 4.1 and noticing that \(S_k(p,a)=C_k(p,a)+1\), we get \(S_k(p,a)\ll p^{\frac{1}{2}}\), and thus \(\mathscr {V}(p,a)\ll p^{\frac{3}{2}}\). Therefore, the second term in (5.6) is \(\leqslant c_1p^{-\frac{3}{2}}\). On the other hand, from Lemma 5.3, we can see that the first term in (5.6) is \(\leqslant 2^4\cdot 50p^{-\frac{3}{2}}=800p^{-\frac{3}{2}}\). Let \(c_2=c_{1}+800\). Then we have proved that, for \(p\not \mid n\), there holds
Moreover, if we use Lemma 4.1 directly, it follows that
and therefore,
Let \(c_3=\max (c_2,27648)\). Then, for square-free q, we have
Hence, by (5.5), we obtain
This proves (i) of Lemma 5.6.
To prove (ii) of Lemma 5.6, by Lemma 5.5, we first note that
From (5.7), we have
By (5.8), we know that there exists \(c_5>0\) such that
On the other hand, it is easy to see that
By Lemma 5.4, we know that \(\mathcal {L}(p,n)>0\) for all p with \(n\equiv 0\,(\bmod \,2)\) and \(n\equiv \pm 1\,(\bmod \,3)\), and thus \(1+A(n,p)>0\). Therefore, there holds
Combining the estimates (5.9)–(5.11) and (5.13), and taking \(c^*=c_5>0\), we derive that
This completes the proof of Lemma 5.6. \(\square \)
6 Proof of Proposition 3.2
In this section, we first present some lemmas that will be used to prove Proposition 3.2.
Lemma 6.1
Suppose that \(\alpha \) is a real number, and that there exist integers \(a\in \mathbb {Z}\) and \(q\in {\mathbb {N}}\) satisfying
where
Then, for \(k\in \{2,3\}\), we have
where
Proof
For the proof of the upper bound of \(f_2(\alpha )\), one can see Theorem 3 of [10]; while for the proof of the upper bound of \(f_3(\alpha )\), one can see Lemma 2.3 of [18]. \(\square \)
Lemma 6.2
Let \(2\leqslant k_1\leqslant k_2\leqslant \cdots \leqslant k_s\) be natural numbers such that
Then we have
Proof
See Lemma 1 of [1]. \(\square \)
Lemma 6.3
Let \(f_k(\alpha )\) be defined as above. Then we have
Proof
The conclusion can be deduced by counting the number of solutions of the underlying Diophantine equation:
with \(X_2<x_1,x_2\leqslant 2X_2\) and \(X_6<y_i\leqslant 2X_6\) for \(i=1,2,\ldots ,8\). If \(x_1\not =x_2\), the contribution is bounded by \(X_6^{8+\varepsilon }\). If \(x_1=x_2\), the contribution is bounded by
By Lemma 2.5 of [12], we have
and thus the contribution with \(x_1=x_2\) is \(\ll X_2\cdot X_6^{5+\varepsilon }\). Combining the above two cases, we deduce that
This completes the proof of Lemma 6.3. \(\square \)
For the proof of Proposition 3.2, we define a general Hardy–Littlewood dissection employed in our application of the circle method. When X is a positive number with \(X\leqslant \sqrt{N}\), we take \({\mathfrak {N}}(X)\) to be the union of the intervals
with \(1\leqslant a\leqslant q\leqslant X\) and \((a,q)=1\). Also, when \(X\leqslant \sqrt{N}/2\), we put \({\mathfrak {R}}(X)={\mathfrak {N}}(2X)\setminus {\mathfrak {N}}(X)\). Finally, we take
For \(\alpha \in {\mathfrak {m}}_2\), by Dirichlet’s lemma on rational approximation (for instance, see Lemma 2.1 of Vaughan [12]), there exist \(a\in \mathbb {Z}\) and \(q\in {\mathbb {N}}\) satisfying
Since \(\alpha \in {\mathfrak {m}}_2\), we know that either \(q>N^{\frac{1}{8}}\) or \(N|q\alpha -a|>N^{\frac{1}{8}}\). Therefore, by Lemma 6.1, it is easy to see that
which combines Hölder’s inequality, Lemma 6.2 and Lemma 6.3 yields
We define the function \(\Upsilon :[0,1]\rightarrow [0,1]\) by putting \(\Upsilon (\alpha )=0\) for \(\alpha \in [0,1]\setminus {\mathfrak {N}}(N^{\frac{1}{8}})\), and when \(\alpha \in {\mathfrak {N}}(N^{\frac{1}{8}})\cap {\mathfrak {N}}(q,a,N^{\frac{1}{8}})\) we write
Define
By noting the fact that \({\mathfrak {m}}_4\subseteq {\mathfrak {N}}(N^{\frac{1}{8}})\setminus {\mathfrak {N}}(N^{\frac{1}{18}})\), hence for \(\alpha \in {\mathfrak {m}}_4\), it follows from Lemma 6.1 that
which combined with the trivial estimate \(f_6(\alpha )\ll N^{\frac{1}{6}+\varepsilon }\) yields
By Lemma 2 of [2], we obtain
Using the above estimate and (6.2), we conclude that
For \(\alpha \in {\mathfrak {m}}_3\), by Lemma 6.1, we get
Hence, for \(\alpha \in {\mathfrak {m}}_3\), there holds
which combined with the trivial estimate \(f_6(\alpha )\ll N^{\frac{1}{6}+\varepsilon }\) yields
This leaves the set \({\mathfrak {m}}\cap {\mathfrak {N}}(N^{\frac{1}{18}})\) for treatment, and this set is covered by the union of sets \({\mathfrak {R}}(Y)={\mathfrak {N}}(2Y)\setminus {\mathfrak {N}}(Y)\) as Y runs over the sequence \(2^{-j}N^{\frac{1}{18}}\) with \(P\ll Y\leqslant N^{\frac{1}{18}}/2\). Note that \(\Upsilon (\alpha )\ll Y^{-1}\) for \(\alpha \not \in {\mathfrak {N}}(Y)\). Moreover, Lemma 2 of [2] supplies the following upper bound
which implies that
By a splitting argument, from (6.4) and (6.5), we derive that
Combining (6.1), (6.3) and (6.6), we obtain the conclusion of Proposition 3.2.
References
Brüdern J, Sums of squares and higher powers, J. London Math. Soc. 35(2) (1987) 233–243
Brüdern J, A problem in additive number theory, Math. Proc. Cambridge Philos. Soc. 103(1) (1988) 27–33
Brüdern J and Wooley T D, Subconvexity for additive equations: Pairs of undenary cubic forms, J. Reine Angew. Math. 696 (2014) 31–67
Brüdern J, A ternary problem in additive prime number theory, in: From arithmetic to zeta-functions (eds) J. Sander et al. (2016) (Springer) pp. 57–81
Golubeva E P, A bound for the representability of large numbers by ternary forms, and nonhomogeneous Waring equations, J. Math. Sci. 157(4) (2009) 543–552
Hua L K, Additive Theory of Prime Numbers, American Mathematical Society, Providence, Rhode Island (1965)
Liu Y, Exceptional set in Waring–Goldbach problem: Two squares, two cubes and two sixth powers, Proc. Indian Acad. Sci. Math. Sci. 130(1) (2020) Article ID 8
Lü X D and Mu Q W, Exceptional sets in Waring’s problem: two squares, two cubes and two sixth powers, Taiwanese J. Math. 19(5) (2015) 1359–1368
Kawada K and Wooley T D, Relations between exceptional sets for additive problems, J. London Math. Soc. 82(2) (2010) 437–458
Kumchev A V, On Weyl sums over primes and almost primes, Michigan Math. J. 54(2) (2006) 243–268
Titchmarsh E C, The Theory of the Riemann Zeta–Function, 2nd edn, revised by D. R. Heath-Brown (1986) (Oxford: Oxford University Press)
Vaughan R C, The Hardy–Littlewood Method, 2nd edn (1997) (Cambridge: Cambridge University Press)
Vinogradov I M, Elements of Number Theory (1954) (New York: Dover Publications)
Wooley T D, On Waring’s problem: Some consequences of Golubeva’s method, J. London Math. Soc. 88(3) (2013) 699–715
Wooley T D, Slim exceptional sets and the asymptotic formula in Waring’s problem, Math. Proc. Cambridge Philos. Soc. 134(2) (2003) 193–206
Wooley T D, On Waring’s problem: two squares, two cubes and two sixth powers, Quart. J. Math. 65(1) (2014) 305–317
Zhang M and Li J, Exceptional set for sums of unlike powers of primes, Taiwanese J. Math. 22(4) (2018) 779–811
Zhao L, The additive problem with one cube and three cubes of primes, Michigan Math. J. 63(4) (2014) 763–779
Acknowledgements
The authors would like to express their most sincere gratitude to Professor Wenguang Zhai for his valuable advice and constant encouragement. Also, the authors appreciate the referee for his/her patience in refereeing this paper. This work is supported by the National Natural Science Foundation of China (Grant Nos. 11901566, 12001047, 11971476, 12071238), the Fundamental Research Funds for the Central Universities (Grant No. 2021YQLX02), the National Training Program of Innovation and Entrepreneurship for Undergraduates (Grant No. 202107010), the Undergraduate Education and Teaching Reform and Research Project for China University of Mining and Technology (Beijing) (Grant No. J210703), and the Scientific Research Funds of Beijing Information Science and Technology University (Grant No. 2025035).
Author information
Authors and Affiliations
Corresponding author
Additional information
Communicated by Sanoli Gun.
Rights and permissions
About this article
Cite this article
Li, J., Zhang, M. & Zhao, Y. Slim exceptional set for sums of mixed powers of primes. Proc Math Sci 131, 29 (2021). https://doi.org/10.1007/s12044-021-00622-x
Received:
Revised:
Accepted:
Published:
DOI: https://doi.org/10.1007/s12044-021-00622-x