Universal sums of generalized octagonal numbers
Jangwon Ju, Byeong-Kweon Oh

TL;DR
This paper proves the universality of certain sums of generalized octagonal numbers, confirming conjectures by Sun, and provides a criterion for the universality of arbitrary such sums, extending the 15-theorem.
Contribution
It establishes the universality of specific quaternary sums of generalized octagonal numbers and generalizes the 15-theorem to these sums.
Findings
Proved universality for five specific sums of generalized octagonal numbers.
Provided an effective criterion for the universality of arbitrary sums of generalized octagonal numbers.
Confirmed conjectures posed by Sun regarding these sums.
Abstract
An integer of the form for some integer is called a generalized octagonal number. A quaternary sum of generalized octagonal numbers is called {\it universal} if has an integer solution for any positive integer . In this article, we show that if and or , then is universal. These were conjectured by Sun in \cite {sun}. We also give an effective criterion on the universality of an arbitrary sum of generalized octagonal numbers, which is a generalization of "-theorem" of Conway and Schneeberger.
Peer Reviews
No public reviews on file for this paper yet. If you reviewed it on a platform where reviews are public (OpenReview, ICLR, NeurIPS, ICML), you can paste yours below so the community can read it here.
Videos
No videos yet. Explain this paper in a talk, walkthrough, or lecture? Add one.
Taxonomy
TopicsAdvanced Mathematical Identities Β· Mathematics and Applications Β· Analytic Number Theory Research
Universal sums of generalized octagonal numbers
Jangwon Ju and Byeong-Kweon Oh
Department of Mathematical Sciences, Seoul National University, Seoul 08826, Korea
Department of Mathematical Sciences and Research Institute of Mathematics, Seoul National University, Seoul 08826, Korea
Abstract.
An integer of the form for some integer is called a generalized octagonal number. A quaternary sum of generalized octagonal numbers is called universal if has an integer solution for any positive integer . In this article, we show that if and or , then is universal. These were conjectured by Sun in [11]. We also give an effective criterion on the universality of an arbitrary sum of generalized octagonal numbers , which is a generalization of β-theoremβ of Conway and Schneeberger.
Key words and phrases:
Lagrangeβs four square theorem, Generalized octagonal numbers
2000 Mathematics Subject Classification:
Primary 11E12, 11E20
This work of the first author was supported by BK21 PLUS SNU Mathematical Sciences Division.
This work of the second author was supported by the National Research Foundation of Korea (NRF-2017R1A2B4003758).
1. introduction
The famous Lagrangeβs four square theorem states that every positive integer can be written as a sum of at most four integral squares. Motivated by Lagrangeβs four square theorem, Ramanujan provided a list of 55 candidates of diagonal quaternary quadratic forms that represent all positive integers. In [4], Dickson pointed out that the quaternary form that is included in Ramanujanβs list represents all positive integers except and confirmed that Ramanujanβs assertion for all the other forms is true. A positive definite integral quadratic form is called universal if it represents all non-negative integers. The problem on determining all universal quaternary forms was completed by Conway and Schneeberger. They proved that there are exactly universal quaternary quadratic forms. Furthermore, they proved the so called β15-theoremβ, which states that every positive definite integral quadratic form that represents and is, in fact, universal, irrespective of its rank (see [1]). Recently, Bhargava and Hanke [2] proved, so called, β290-theoremβ, which states that every positive definite integer-valued quadratic form is universal if it represents
[TABLE]
In general, a polygonal number of order (or an -gonal number) for is defined by
[TABLE]
for some non-negative integer . If we admit that is a negative integer, then is called a generalized polygonal number of order (or a generalized -gonal number). Then, Lagrangeβs four square theorem implies that the diophantine equation
[TABLE]
has an integer solution and , for any non-negative integer .
Recently, Sun in [11] proved that every positive integer can be written as a sum of four generalized octagonal numbers, which is also considered as a generalization of Lagrangeβs four square theorem. He also defined that for positive integers , a quaternary sum (simply, ) of generalized octagonal numbers is universal over if the diophantine equation
[TABLE]
has an integer solution for any non-negative integer . Then, he showed that if is universal over , then , and is one of the following 40 triples:
[TABLE]
and proved the universalities over except the following 5 triples:
[TABLE]
He conjectured that for each of these 5 triples, the sum is also universal over .
The aim of this article is to prove this conjecture. In Section 3, we further prove that the sum of generalized octagonal numbers is universal over if and only if it represents
[TABLE]
This might be considered as a generalization of β-theoremβ of Conway and Schneeberger. For universal sums of triangular numbers, see [3].
For a quadratic form , the corresponding symmetric matrix of is defined by . In particular, for a diagonal quadratic form , we simply write . For an integer , if the diophantine equation has an integer solution, then we say is represented by , and we write . The genus of , denoted by , is the set of all quadratic forms that are locally isometric to . The number of isometry classes in is called the class number of .
Any unexplained notations and terminologies can be found in [7] or [10].
2. Quaternary universal sums of generalized octagonal numbers
In this section, we prove that if is one of the triples in (1.1), then the quaternary sum of generalized octagonal numbers is universal over .
Let be positive integers. Recall that a quaternary sum of generalized octagonal numbers is said to be universal over if the diophantine equation
[TABLE]
has an integer solution for any non-negative integer . Note that is universal over if and only if the equation
[TABLE]
has an integer solution for any non-negative integer . This is equivalent to the existence of an integer solution of the diophantine equation
[TABLE]
such that .
Theorem 2.1**.**
The quaternary sum of generalized octagonal numbers is universal over .
Proof.
It is enough to show that has an integer solution such that . First, assume that . Note that the class number of is and has an integer solution such that . Hence
[TABLE]
Furthermore, we also have
[TABLE]
Now assume that . Note that for any , the diophatine equation has an integer solution such that . Therefore we assume that . Similarly as above, has an integer solution such that . Hence
[TABLE]
Therefore, the diophantine equation has always an integer solution. β
Theorem 2.2**.**
The quaternary sum of generalized octagonal numbers is universal over .
Proof.
First, we show that has an integer solution such that . If , then one may directly check that such an integer solution exists. Note that the class number of is and it represents all integers such that and for any non-negative integers . Note that for any ,
[TABLE]
Hence has an integer solution such that and or . Since
[TABLE]
has an integer solution . Now, by Theorem 9 of [6] (see also [5], and for more generalization see [9]), there are integers such that and . β
Theorem 2.3**.**
The quaternary sum of generalized octagonal numbers is universal over .
Proof.
First, we show that the diophantine equation has an integer solution such that and . If , then one may directly check that such an integer solution exists. From now on, we assume that .
Assume that an integer , for some positive integer is represented by the genus of . Since the class number of is one, is represented by , that is, has an integer solution . Note that or is divisible by . Therefore is represented by . Similarly, every integer of the form , for some positive integer that is represented by the genus of is represented by itself.
If and , then both and are represented by by the above observation. Furthermore, at least one of them is not of the form for any integer . If and , then both and are represented by . Furthermore, at least one of them is not of the form for any integer . Finally, assume that , where is odd, or , and . Then the assertion follows from the fact that the assertion holds if or and .
Now, since
[TABLE]
and
[TABLE]
the diophantine equation has always an integer solution by Theorem 9 of [6]. β
Theorem 2.4**.**
The quaternary sums and of generalized octagonal numbers are all universal over .
Proof.
Since the proofs are quite similar to each other, we only provide the proof of the former case.
It suffices to show that the diophantine equation has an integer solution such that for any non-negative integer . If , then one may directly check that such an integer solution exists. Therefore we assume that . First, we show that the diophantine equation has an integer solution such that
[TABLE]
Note that an integer is represented by if and only if is divisible by and is not of the form for any non-negative integers . Then one may easily show that there is an integer such that is represented by . Furthermore, one may also show that there is an integer such that is represented by . Hence By taking a suitable integer , we assume that there are integers and such that
[TABLE]
Since , we may assume that
[TABLE]
Hence there are integers such that and by Theorem 9 of [6]. Therefore, if , then we are done. Assume that . Since , we have
[TABLE]
where . Therefore, the diophantine equation has an integer solution for any non-negative integer . β
3. The octagonal theorem of sixty
Let be any positive integers. For an integer , if the diophantine equation
[TABLE]
has an integer solution, then we say the sum of generalized octagonal numbers represents and we write . The sum of generalized octagonal numbers is called universal over if it represents all non-negative integers. In this section, we prove:
Theorem 3.1**.**
The sum of generalized octagonal numbers is universal over if and only if it represents the integers
[TABLE]
To prove this, we need the following six lemmas:
Lemma 3.2**.**
The quaternary sum of generalized octagonal numbers represents all positive integers except .
Proof.
It suffices to show that the diophantine equation has an integer solution such that for any non-negative integer . Note that an integer is represented by if and only if it is not of the form for any non-negative integers . If and , then one may directly check that the equation
[TABLE]
has an integer solution. Hence we may assume that . One may easily show that there is an integer
[TABLE]
such that . Furthermore, one may also show that there is an integer
[TABLE]
such that . For all possible cases when both and are squares of integers, one may easily check that the equation has an integer solution such that . Therefore, we may assume that the equation has an integer solution such that and is not a square of an integer. If , then we are done. If , then without loss of generality, we may assume that and . Since , there are integers such that and by Theorem 9 of [6]. The lemma follows from this. β
Lemma 3.3**.**
The quaternary sum of generalized octagonal numbers represents all positive integers except .
Proof.
It suffices to show that the diophantine equation has an integer solution such that for any non-negative integer . If and , then one may directly check that such an integer solution exists. From now on, we assume that . One may easily show that there is an integer such that is represented by . Therefore, the equation has an integer solution such that and . If , then we are done. Suppose that . If , then one may easily check that is also an integer solution of
[TABLE]
Therefore, there is a positive integer such that is an integer solution of (3.1) such that or for any positive integer , is an intger solution of (3.1) each of whose component is divisible by . Since there are only finitely many integer solution of (3.1) and has an infinite order, the latter is impossible unless is an eigenvector of . Note that is an eigenvector of and
[TABLE]
where is not an eigenvector of . Therefore the equation has always an integer solution such that . This completes the proof. β
Lemma 3.4**.**
The quaternary sum of generalized octagonal numbers represents all positive integers except .
Proof.
Since the proof is quite similar to that of Theorem 2.4, it is left to the reader. β
Lemma 3.5**.**
The quinary sum of generalized octagonal numbers is universal over for any integer .
Proof.
Since the proofs are quite similar to each other, we only provide the proof of the sum .
It suffices to show that the diophantine equation has an integer solution such that for any non-negative integer . If , then one may directly check that such an integer solution exists. Therefore we assume that . Note that there are integers such that . Since the class number of is one and , is represented by . Therefore the diophantine equation has an integer solution such that and . If , then we are done. Assume that . Without loss of generalitiy, we may assume that , for . Then we have
[TABLE]
where . Therefore, the equation has an integer solution for any non-negative integer . β
Lemma 3.6**.**
The quinary sum of generalized octagonal numbers is universal over .
Proof.
It suffices to show that the diophantine equation has an integer solution such that for any non-negative integer . If , then one may directly check that such an integer solution exists. Therefore we assume that . First, we show that the the diophantine equation has an integer solution such that and is not a square of an integer. Note that the class number of is one. Then one may easily show that for any integer ,
[TABLE]
Furthermore, in each case, at least one of two integers of the form is not a square of an integer, for we are assuming that . Therefore the equation has an integer solution such that . If , then we are done. Assume that and . Since is not a square of an integers, . Therefore the equation has always an integer solution by Theorem 4.1 of [9]. β
Lemma 3.7**.**
The quinary sum of generalized octagonal numbers is universal over .
Proof.
It suffices to show that the equation has an integer solution such that . If , then one may directly check that such an integer solution exists. Therefore we assume that . First, assume that . Note that the class number of is one and every odd positive integer that is not divisible by is represented by . Therefore or is represented by . Hence the equation
[TABLE]
has an integer solution such that . Assume that . Since or is represented by , the equation
[TABLE]
has an integer solution such that . Assume that . Since or is represented by , the equation
[TABLE]
has an integer solution such that . Therefore the diophantine equation has always an integer solution. β
Proof of Theorem 3.1. Without loss of generality, we may assume that . Since , we have . Since , we have or . Assume that . Since and , we have . Assume that . Since and , we have . Now, note that , where
[TABLE]
Therefore, for each possible case, where is the integer given above.
If is one of possible quadruples given above except the quadruples , , and , then the quaternary sum of generalized octagonal numbers is universal over by [11] and Section 2. Now, note that and for any integer satisfying , where
[TABLE]
Furthermore, for each case, represents all positive integers except by Lemmas 3.2, 3.3 and 3.4. Therefore, the sum of generalized octagonal numbers is universal over for any such that . Note that . Therefore if . The sum of generalized octagonal numbers is universal over for any such that by Lemmas 3.5, 3.6 and 3.7. This completes the proof. β
The reference list from the paper itself. Each links out to its DOI / PubMed record.
- 1[1] M. Bhargava, On the Conway-Schneeberger fifteen theorem , Contem. Math. 272 (2000), 27β38.
- 2[2] M. Bhargava and J. Hanke, Universal quadratic forms and the 290 theorem , to appear in Invent. Math.
- 3[3] W. Bosma and B. Kane, The triangular theorem of eight and representation by quadratic polynomials , Proc. Amer. Math. Soc. 141 (2013), 1473-1486.
- 4[4] L. E. Dickson, Quaternary quadratic forms representing all integers , Amer. J. Math. 49 (1927), 39-56.
- 5[5] W. C. Jagy, Five regular or nearly-regular ternary quadratic forms , Acta Arith. 77 (1996), 361β367.
- 6[6] B. W. Jones, Representation by Positive Ternary Quadratic Forms , unpublished Ph.D. dissertation, University of Chicago, 1928.
- 7[7] Y. Kitaoka, Arithmetic of quadratic forms , Cambridge University Press, 1993.
- 8[8] B.-K. Oh, Regular positive ternary quadratic forms , Acta Arith. 147 (2011), 233-243.
