Regular ternary triangular forms
Mingyu Kim, Byeong-Kweon Oh

TL;DR
This paper classifies all primitive regular ternary triangular forms, proving that exactly 49 such forms exist, which represent all locally represented positive integers.
Contribution
It provides a complete classification of primitive regular ternary triangular forms, establishing the exact number as 49.
Findings
Exactly 49 primitive regular ternary triangular forms exist.
Each such form represents all positive integers locally represented.
The classification completes the understanding of regular ternary triangular forms.
Abstract
An integer of the form for some positive integer is called a triangular number. A ternary triangular form for positive integers and is called regular if it represents every positive integer that is locally represented. In this article, we prove that there are exactly 49 primitive regular ternary triangular forms.
| 1 | 2 | 3 | 4 | 5 | 6 | 7 | 8 | 9 | 10 | 11 | |
| 1 | 1 | 1 | 1 | 1 | 1 | 1 | 1 | 1 | 1 | 1 | 1 |
| 2 | 1 | 1 | 1 | 1 | 1 | 1 | 1 | 1 | 1 | 1 | 1 |
| 3 | 1 | 1 | 1 | 1 | 1 | 1 | 1 | 1 | 1 | 1 | 1 |
| 4 | 2 | 1 | 1 | 1 | 1 | 1 | 1 | 1 | 1 | 1 | 1 |
| 5 | 2 | 1 | 1 | 1 | 1 | 1 | 1 | 1 | 1 | 1 | 1 |
| 6 | 2 | 2 | 1 | 1 | 1 | 1 | 1 | 1 | 1 | 1 | 1 |
| 7 | 2 | 2 | 1 | 1 | 1 | 1 | 1 | 1 | 1 | 1 | 1 |
| 9 | 2 | 2 | 2 | 1 | 1 | 1 | 1 | 1 | 1 | 1 | 1 |
| 19 | 4 | 3 | 3 | 2 | 2 | 2 | 1 | 1 | 1 | 1 | 1 |
| 20 | 4 | 3 | 3 | 2 | 2 | 2 | 2 | 1 | 1 | 1 | 1 |
| 25 | 4 | 3 | 4 | 3 | 2 | 2 | 2 | 2 | 1 | 1 | 1 |
| 26 | 4 | 4 | 4 | 3 | 2 | 2 | 2 | 2 | 1 | 1 | 1 |
| 29 | 6 | 4 | 4 | 3 | 3 | 2 | 2 | 2 | 1 | 1 | 1 |
| 32 | 6 | 5 | 4 | 3 | 3 | 2 | 2 | 2 | 2 | 2 | 1 |
| 35 | 6 | 5 | 4 | 4 | 3 | 3 | 2 | 2 | 2 | 2 | 1 |
| 41 | 7 | 5 | 4 | 4 | 4 | 3 | 3 | 2 | 2 | 2 | 2 |
| 47 | 8 | 5 | 4 | 5 | 4 | 3 | 3 | 3 | 2 | 2 | 2 |
| 49 | 8 | 5 | 4 | 5 | 4 | 3 | 3 | 3 | 2 | 2 | 2 |
| 83 | 13 | 9 | 7 | 6 | 7 | 5 | 5 | 4 | 3 | 3 | 3 |
| 314 | 41 | 29 | 22 | 16 | 13 | 11 | 10 | 12 | 11 | 11 | 9 |
| (10,5) | (9,5) | (8,7) | (7,7) | (6,7) | (5,9) | (4,13) | (3,29) | |
|---|---|---|---|---|---|---|---|---|
| 10 | 9 | 8 | 7 | 6 | 5 | 4 | 3 | |
|---|---|---|---|---|---|---|---|---|
| Cases | Triangular forms in |
|---|---|
| , | , | , |
|---|---|---|
| , | , | , |
| , | , | , |
| , | , | , |
| , | , | , |
| , | , | , |
| , | , | , |
| , | , | , |
| , | , | , |
| , | , | , |
| , | , | , |
| , | , | , |
| , | , | , |
| , | , | , |
| , | , | , |
| , | , | , |
| . |
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
TopicsAnalytic Number Theory Research · Algebraic Geometry and Number Theory · Advanced Mathematical Identities
Regular ternary triangular forms
Mingyu Kim and Byeong-Kweon Oh
Department of Mathematical Sciences, Seoul National University, Seoul 151-747, Korea
Department of Mathematical Sciences and Research Institute of Mathematics, Seoul National University, Seoul 151-747, Korea
Abstract.
An integer of the form for some positive integer is called a triangular number. A ternary triangular form for positive integers and is called regular if it represents every positive integer that is locally represented. In this article, we prove that there are exactly 49 primitive regular ternary triangular forms.
Key words and phrases:
Representations of ternary quadratic forms, triangular numbers
2010 Mathematics Subject Classification:
Primary 11E12, 11E20
This work of the second author was supported by the National Research Foundation of Korea (NRF-2017R1A2B4003758).
1. Introduction
A quadratic homogeneous polynomial
[TABLE]
is called an integral quadratic form. Throughout this article, we always assume that is positive definite, that is, for any non-zero real vector . Let be any ring containing . For an integer , if there is a solution of the equation , then we say that is represented by over . In particular, if is represented by over the -adic integer ring for any prime , then we say that is locally represented by . From the definition, note that any integer that is represented by over is locally represented by . However, it is known that the converse is not true, in general. A positive definite integral quadratic form is called regular if the converse is also true, that is, it represents every integer over that is locally represented.
Dickson [9] who initiated the study of regular quadratic forms first coined the term regular. Jones and Pall [16] gave the list of all 102 primitive diagonal regular ternary quadratic forms. Watson proved in his thesis [26] that there are only finitely many equivalence classes of primitive positive definite ternary regular forms. Jagy, Kaplansky and Schiemann [14] succeeded Watson’s study on regular quadratic forms and provide the list of 913 candidates of regular positive definite integral ternary quadratic forms. All but 22 of them are already proved to be regular at that time. Recently, the second author [21] proved the regularities of 8 ternary quadratic forms among remaining 22 candidates. A conditional proof for the remaining 14 candidates under the Generalized Riemann Hypothesis was given by Lemke Oliver [20]. Note that there are infinitely many regular positive definite integral quaternary quadratic forms (for this, see [12]).
Now we look into the representations of ternary triangular forms. An integer of the form for some positive integer is called a triangular number. For positive integers with , a polynomial of the form
[TABLE]
is called a ternary triangular form. We say an integer is represented by the triangular form if
[TABLE]
has an integer solution . Similarly to the quadratic form case, if Equation (1.1) has a solution in for any prime , then we say that is locally represented by . We say is regular if it represents every integer that is locally represented. Note that any ternary triangular form is regular if it represents all positive integers. Such a ternary triangular form is said to be universal.
Gauss’ Eureka Theorem says that every positive integer is a sum of at most three triangular numbers, which implies that the ternary triangular form is universal. In 1862, Liouville classified all universal ternary triangular forms, and they are, in fact, the following seven forms:
[TABLE]
As mentioned above, these universal triangular forms are regular. In 2013, Chan and Oh [7] proved that there are only finitely many regular ternary triangular forms. In 2015, Chan and Ricci [8] proved the finiteness of regular ternary triangular forms in a more general setting. They actually proved that for any given positive integer , there are only finitely many inequivalent positive ternary regular primitive complete quadratic polynomials with conductor . From this follows the finiteness of regular ternary -gonal forms. Note that an integer of the form for some integer is called an -gonal number, and a (regular) ternary -gonal form is defined similarly.
In this article, we prove that there are exactly regular ternary triangular forms. In the previous papers [7] and [8], the authors use Burgess’ estimation on character sums (for this, see [4] and [11]) to prove the finiteness of regular ternary triangular forms. It seems to be quite difficult to find an explicit upper bound of the discriminant of regular ternary triangular forms by using Burgess’ estimation. In this article, we use a purely arithmetic method to find such an explicit and effective upper bound of the discriminant of regular ternary triangular forms, and finally, we classify all regular ternary triangular forms.
A -lattice is a finitely generated free -module equipped with a non-degenerate symmetric bilinear form such that . The corresponding quadratic map is defined by for any .
Let be a -lattice. The quadratic form corresponding to is defined by . Furthermore, the corresponding symmetric matrix is defined by , which is called the matrix presentation of . If admits an orthogonal basis , we call diagonal and simply write
[TABLE]
For any odd prime , denotes a non-square unit in .
Any unexplained notations and terminologies can be found in [19] or [24].
2. Preliminaries
A nonnegative integer of the form for some positive integer is called a triangular number. For example, are triangular numbers. Since , is a triangular number for any integer . For positive integers with , we call a polynomial of the form
[TABLE]
a -ary triangular form. For a triangular form , we define , which is called the discriminant of the triangular form . A triangular form is called primitive if . Unless stated otherwise, we always assume that
every triangular form is primitive.
For an integer and a -ary triangular form , we say that is represented by if the Diophantine equation
[TABLE]
has an integral solution. In this case, we write . We also define
[TABLE]
and to be the cardinality of the above set.
A triangular form is called universal if it represents every positive integer, that is,
[TABLE]
for any positive integer . A triangular form is called regular if it globally represents every integer which is locally represented. In other words, is regular if for any integer such that is soluble in for any prime , the diophantine equation is soluble in . As shown in [7], any primitive triangular form is universal over .
Note that a triangular form represents if and only if the Diophantine equation
[TABLE]
is soluble in . This equivalence shows how the representation of a triangular form is transformed into the representation of a diagonal quadratic form with congruence conditions. Now, we can reformulate the regularity in a practical way. A ternary triangular form is regular if the following implication holds: for any positive integer , if is soluble in for any odd prime , then there exist odd integers and such that .
Let be a positive definite integral quadratic form of rank and let be an integer. We define
[TABLE]
We say that is represented by if . For a vector , we also define
[TABLE]
and .
For an integer and a diagonal quadratic form , we write
[TABLE]
if there is a vector with such that . We also use the notation
[TABLE]
if there does not exist such a vector . Under these notations, the followings are all equivalent:
- (i)
; 2. (ii)
; 3. (iii)
; 4. (iv)
.
Let be a -lattice and let be a positive integer. Watson transformation of modulo is defined by
[TABLE]
We denote by the primitive -lattice obtained from by scaling by a suitable rational number. Let be an odd prime. Let be a ternary -lattice, where and . Then one may easily check
[TABLE]
For a ternary triangular form and an odd prime , we define
[TABLE]
where .
Lemma 2.1**.**
Let be an odd prime and let be positive integers which are not divisible by . Let be positive integers. If the ternary triangular form is regular, then so is .
Proof.
See [7, Lemma 3.3]. ∎
Though the proof of the next lemma is quite similar to the proof of Lemma 2.1, we provide the proof for completeness.
Lemma 2.2**.**
Let be an odd prime and let be a positive integer. Let , and be positive integers such that and , where is the Legendre symbol modulo . If the ternary triangular form is regular, then so is .
Proof.
It is enough to show that is regular. Let be a positive integer such that the equation
[TABLE]
is soluble in for any prime . Then
[TABLE]
Thus
[TABLE]
Since is regular, there is a vector with such that . Since is divisible by , we have . From the assumption , we have . So
[TABLE]
with . Thus Equation is soluble in . This completes the proof. ∎
For an odd prime and a ternary -lattice , we say that is -stable if
[TABLE]
for some . We say that is stable if is -stable for every odd prime . A ternary triangular form is called -stable (stable) if the corresponding quadratic form is -stable (stable, respectively). Let be a regular ternary triangular form. Then by taking -transformations to repeatedly, if possible, for any odd prime dividing the discriminant, we may obtain a stable regular ternary triangular form by Lemmas 2.1 and 2.2. Note that the corresponding quadratic form has a smaller discriminant and a simpler local structure than .
3. Stable regular ternary triangular forms
In this section, we prove that there are exactly 17 stable regular ternary triangular forms. Throughout this section, denotes the -th odd prime so that is the set of all odd primes. Let be a stable regular ternary triangular form. We always assume that .
Lemma 3.1**.**
For an integer greater than , let be odd primes. Let be an integer with and let be an arbitrary integer. Then there is an integer with such that .
Proof.
See [17, Lemma 3]. ∎
Though Lemma 3.1 gives, in general, a nice upper bound of the longitude of arithmetic progression satisfying the assumption, there is a shaper bound in some restricted situation.
Lemma 3.2**.**
Under the same notations given in Lemma 3.1, if , then there is an integer with such that .
Proof.
Trivial. ∎
Lemma 3.3**.**
Let be a prime and let be a positive integer with . Let be a -stable -lattice that is anisotropic over . Then there is an integer such that
- (i)
;
- (ii)
* over ;*
- (iii)
* over ;*
- (iv)
.
Proof.
Since is -stable and is anisotropic over by assumption, we have
[TABLE]
for some . First, we assume that divides . Since , it does not represent satisfying . Since , there exists a positive integer with such that
[TABLE]
Then one may easily check that satisfies all conditions given above. Now, assume that divides . Without loss of generality, we may assume that divides . Since , there exists an integer with such that is not a square modulo and . We take a positive integer with such that . One may easily show that satisfies all conditions given above, which completes the proof. ∎
Let be the set of odd primes such that the diagonal ternary quadratic form is anisotropic over . Since such primes are only finitely many, we let
[TABLE]
Let
[TABLE]
Note that if , and otherwise.
Lemma 3.4**.**
Under the assumptions given above, we have .
Proof.
Note that represents every integer of the form over . Let be a positive integer satisfying Lemma 3.3 in the case when and .
By Lemma 3.1, there is an integer with such that . If we let , then one may easily show that
[TABLE]
and
[TABLE]
Since is regular, there is a vector with such that . From Equation (3.1), we have . So and we have . Now
[TABLE]
Assume to the contrary that . Then one may easily show that
[TABLE]
Since for any , we have
[TABLE]
which is a contradiction. Therefore we have . This completes the proof. ∎
If we are able to use Lemma 3.2 instead of Lemma 3.1, then we may have more effective upper bound of than the previous lemma.
Lemma 3.5**.**
Under the same notations given above, if for some such that , then we have
[TABLE]
Proof.
Note that represents every integer of the form over . Let be a positive integer satisfying Lemma 3.3 in the case when and . Let
[TABLE]
where for each , is suitably chosen in so that
[TABLE]
for any . Note that and for any . Since by assumption, we apply Lemma 3.2 with odd primes , and so that we may conclude that there is an integer with such that
[TABLE]
Therefore, by a similar reasoning to Lemma 3.4, we have . The lemma follows directly from this. ∎
Lemma 3.6**.**
Under the assumptions given above, we have .
Proof.
By Lemma 3.4, we may assume that . First, assume that . Since , we may apply Lemma 3.5 so that
[TABLE]
From this, one may easily show that .
Now, assume that . Since , we may apply Lemma 3.5 so that we may conclude that
[TABLE]
Suppose that . Since , one may directly show that
[TABLE]
Since for any , we have
[TABLE]
which is a contradiction. Therefore we have . Now, since , we deduce, similarly to the above, that
[TABLE]
and thus .
Assume that . Since in this case, one may deduce that
[TABLE]
and thus we have . Now, since , we may have
[TABLE]
and hence . Since ,
[TABLE]
Therefore, we have .
Finally, assume that . Since , we have
[TABLE]
Now, since , we have
[TABLE]
Then, since , we have
[TABLE]
The lemma follows directly from this. ∎
Recall that we are assuming that is stable. Hence for any odd prime ,
[TABLE]
for some . In the former case, every element in is represented by over . In the latter case,
[TABLE]
Recall that is the -th odd prime. Let be a positive integer not divisible by and let be an integer. Let . For a positive integer , we define
[TABLE]
We also define
[TABLE]
Let be the base- representation of , that is,
[TABLE]
with for and . We define
[TABLE]
We also define
[TABLE]
Lemma 3.7**.**
Under the notations and assumptions given above, we have
[TABLE]
Proof.
Since both cases can be done in a similar manner, we only provide the proof of the case when for some positive integer . Without loss of generality, we may assume that . We have to show that the number of integers of the form () in the set is less than or equal to the right hand side, where () is a square (nonsquare, respectively) in .
For any integer such that , let
[TABLE]
Let be the smallest integer greater than that is divisible by . Then any integer in the set is less than or equal to . Note that there is at most one more integer other than these integers that is divisible by , and that is less than or equal to . Note that such an integer exists only when (or ). Furthermore, if such an integer exists, then it must be . Note that there are exactly quadratic residues and quadratic non-residues in the consecutive integers. Therefore there are exactly quadratic residues and quadratic non-residues in
[TABLE]
Note that for any . The remaining multiples of are contained in
[TABLE]
Among them, there are at most quadratic residues, and at most quadratic non-residues. Note that there is at most one multiple of in which is, if exists, contained in the set
[TABLE]
Note that there are at most quadratic residues or a multiple of , and at most quadratic non-residues or a multiple of in the set . The lemma follows from this. ∎
For the sake of brevity, we let
[TABLE]
for positive integers and .
Remark 3.8*.*
One may easily show that for any positive integers and , where is the ceiling function. It is a little bit complicate to compute an upper bound of by using Lemma 3.7. Instead of that, one may easily show that
[TABLE]
Recall that is the set of all odd primes at which is anisotropic, and by Lemma 3.6.
Lemma 3.9**.**
Let be a positive integer. For any integer , we define for . Then we have
[TABLE]
Proof.
Let be any integer greater than and let . We also let , , and . Note that and for any , by assumption. From Remark 3.8, for any , we have . Thus we have
[TABLE]
Since is stable regular, we have
[TABLE]
This completes the proof. ∎
Remark 3.10*.*
In the remaining of this section, we need the exact values of ’s for some integers and . We provide some of these values in Table 1 below.
Lemma 3.11**.**
Under the assumptions given above, we have .
Proof.
By Lemma 3.9 with and , one may easily show that for some . From our assumption of , we have , and thus we have . To prove the lemma, we will use Lemma 3.5 repeatedly.
First, assume that . Since , we may apply Lemma 3.5 so that
[TABLE]
This is possible only when . Now, assume that . Since , one may deduce that
[TABLE]
and thus . Finally, since , we have
[TABLE]
and thus . This completes the proof. ∎
Lemma 3.12**.**
For any stable regular ternary triangular form with , we have or .
Proof.
For any positive integer , we define . Since
[TABLE]
we have
[TABLE]
On the other hand, by Lemma 3.9 with and , one may check that
[TABLE]
By comparing these two inequalities, we have .
Now, we will show that if , then is bounded. For each positive odd integer , we let
[TABLE]
and we also let and . Note that does not depend on . For each integer with , we will choose an integer suitably so that . Note that if this inequality holds, then and therefore, we have
[TABLE]
In fact, we choose
[TABLE]
Now, by using Lemma 3.9 with and , one may easily compute the lower bound of :
To compute an upper bound of , note that
[TABLE]
Hence one may easily show that
[TABLE]
By comparing the lower bound for and the upper bound for , we have an upper bound of for each , as follows:
Now, by using MAPLE program, one may check that there is no stable regular ternary triangular form for . Therefore, we have . ∎
Lemma 3.13**.**
Under the assumptions given above, we have .
Proof.
By the proof of Lemma 3.11, we have . First, assume that . By Lemma 3.9 with and , one may easily show, by using Table 1, that
[TABLE]
On the other hand,
[TABLE]
Thus we have and . Let be a positive integer satisfying Lemma 3.3 in the case when and . Note that
[TABLE]
for any integer . For any positive integer , define
[TABLE]
Clearly is represented by over for any . Note that
[TABLE]
where is an integer defined in Lemma 3.9. From this, similarly with the proof of Lemma 3.9, one may easily show that there exists a positive integer with such that is represented by over for any . Therefore, we have
[TABLE]
Furthermore, since is regular, we have
[TABLE]
From our choices of and , we have . Thus, , which implies that . Therefore we have
[TABLE]
This implies that .
Now, assume that . By Lemma 3.9 with and , one may easily show that
[TABLE]
On the other hand,
[TABLE]
Thus we have and . Similarly to the case when , one may deduce that . Therefore, we have
[TABLE]
which implies that . This completes the proof. ∎
Lemma 3.14**.**
For any stable regular ternary triangular form with , we have .
Proof.
Note that or by Lemma 3.12. First, assume that . By Lemma 3.9 with and , one may easily show that
[TABLE]
On the other hand,
[TABLE]
Thus we have , and . Now, assume that . By Lemma 3.9 with and , one may check that
[TABLE]
On the other hand,
[TABLE]
Thus we have , and . ∎
Now, we are ready to classify all stable regular ternary triangular forms. The following lemma is very useful to prove the regularity.
Lemma 3.15**.**
Let be a positive integer congruent to 4 modulo 8. Then
[TABLE]
Proof.
See [18, Lemma 3.1(iii)]. ∎
Theorem 3.16**.**
There are exactly stable regular ternary triangular forms.
[TABLE]
Proof.
By Lemmas 3.12, 3.13 and 3.14, we have
[TABLE]
First, we want to find an upper bound for for each possible pair . Since all the other cases can be done in a similar manner, we only consider 3 representative cases here.
(i)** **. Let . Suppose that . For any , is not represented by . Furthermore, since by assumption, . Since is stable regular, there is an odd prime divisor of such that is anisotropic over . Therefore, divides and also divides . Furthermore, since , there are at least six such odd primes. This is a contradiction to the fact that . Thus, we have if .
(ii)** **. Let . Suppose that . Since we are assuming that is 3-stable, is not divisible by 3. Any element of is of the from for some positive integer , and the elements of share no odd prime divisors other than . Let . From the assumption that , one may easily check that . Since is stable regular, there is an odd prime dividing and is anisotropic over . Hence is greater than and divides . Thus there are at least six such odd primes. This is a contradiction, and we have .
(iii)** **. Since is 3-stable, is not a multiple of . Note that . Thus there is an odd prime dividing and is anisotropic over . Therefore, is divisible by , which is a contradiction. Therefore, the pair is impossible.
All the other cases can be done in a similar manner to one of the above three cases, and one may obtain an upper bound for in each case. After that, with the help of MAPLE program, one may show that there are candidates of stable regular ternary triangular forms given above.
For each , we write and . For any , it is well known that is universal (see [10, p.23]). Hence we may assume that . Let be any positive integer such that
[TABLE]
Note that has class number 1 for any and thus .
For , one may easily check that
[TABLE]
that is, if , then . Assume that . Since the class number of is 1 and it primitively represents over , there is a vector with . One may easily check that implies in this case. If , then one may easily show that
[TABLE]
Similarly to the previous case, the existence of a vector with implies that
[TABLE]
By Lemma 3.15,
[TABLE]
Therefore we have . If , then one may easily check that
[TABLE]
By Lemma 3.15, we have
[TABLE]
Thus we have . Finally, assume that . Note that if , then and . By Lemma 3.15 again, we have
[TABLE]
This completes the proof. ∎
4. Regular ternary triangular forms
In this section, we prove that there are exactly 49 regular ternary triangular forms. Let be a regular ternary triangular form and let be the stable regular ternary triangular form obtained from it by taking -transformations, if necessary, repeatedly. Here, we are not assuming that . It might happen that there is an odd prime dividing such that . We call such a prime a missing prime. Note that for any odd primes and . Thus if is a missing prime, then one of the followings holds:
- (i)
is regular. 2. (ii)
is regular and .
Lemma 4.1**.**
There is no missing prime greater than .
Proof.
Let be a missing prime. Then there is a stable regular ternary triangular form such that , and (i) or (ii) given above holds.
Assume that the case (i) holds, that is, is regular. We let
[TABLE]
First, we prove that . Assume to the contrary that . One may easily check that if
[TABLE]
with odd integers and , then . Thus we have
[TABLE]
On the other hand, by Theorem 3.16, the set of odd primes at which is anisotropic is
[TABLE]
From Remark 3.8, we have
[TABLE]
From the assumption that , we have . Since
[TABLE]
we must have
[TABLE]
However, one may directly show that if , then . This is a contradiction and hence we have . Now, by a direct calculation with the help of MAPLE, one may check that for any prime and any stable regular ternary triangular form , all of the triangular forms are not regular.
Now, assume that is regular with . By Theorem 3.16, is one of the following pairs:
[TABLE]
First, suppose that . Since all the other cases can be done in a similar manner, we only consider the cases when or . Assume that . Since
[TABLE]
and is regular, there is a vector with such that . From the assumption that , we have . This is a contradiction, for is not a sum of two integer squares. Next, assume that . Note that
[TABLE]
Since we are assuming that is regular, there is a vector with such that . Since , we have . This is a contradiction, for is not represented by . Therefore, we have . Now, by a direct calculation with the help of MAPLE, one may check that for any prime and any stable regular ternary triangular form , all of the forms are not regular. This completes the proof. ∎
Remark 4.2*.*
By Theorem 3.16 and Lemma 4.1, any prime divisor of the discriminant of a regular ternary triangular form is less than or equal to .
Let be a regular ternary triangular form. Then there are nonnegative integers and such that
[TABLE]
is stable regular. Hence, to find all regular ternary triangular forms, it suffices to find all regular ternary triangular forms in the inverse image of the -transformation of each regular triangular form for each . Note that any triangular form in the inverse image , for and , is given in Table 2.
First, we find all regular triangular forms in the inverse images of stable regular ternary triangular forms via -transformation for each , and then we repeat this process again until any inverse image does not contain a regular triangular form. As a sample, ternary triangular forms lying over are given in Table 3. In that table, if the triangular form is not regular, then the smallest positive integer which is represented locally, but not globally by the triangular form is given in parentheses.
Finally, one may have a list of candidates for the regular ternary triangular forms including 17 stable regular forms, which is given in Table 4. The regularities of forms except stable regular forms will be proved here. Before doing that, we need some lemmas.
Let be an odd prime and let be a positive integer relatively prime to . Assume that is represented by the binary quadratic form . In 1928, B. W. Jones proved in his unpublished thesis that if the Diophantine equation has an integral solution, then it also has an integral solution with . The following lemma follows immediately from this.
Lemma 4.3**.**
Let be a positive integer. If for some , then there is a vector such that
[TABLE]
We also need the following lemma which appeared in the middle of the proof of [21, Theorem 3.1].
Lemma 4.4**.**
Let be a positive-definite symmetric matrix and let such that . Let and define
[TABLE]
Assume that
- (i)
* has an infinite order.* 2. (ii)
* for any .*
Then and .
In the following 5 consecutive propositions, we prove the regularities of 5 candidates, all of whose corresponding quadratic forms are not regular(see [16]).
Proposition 4.5**.**
The ternary triangular form is regular.
Proof.
Let be a ternary quadratic form and let be an integer such that . One may easily check that . Thus it suffices to show that . Since
[TABLE]
we may assume that .
First, assume that . Since , there is a vector such that . We have or and thus .
Now, assume that . We assert that there is a vector such that or . Assume to the contrary that there is no such vector. Then, we may assume that there is a vector such that and . Let
[TABLE]
Note that
[TABLE]
If we let
[TABLE]
then one may check that and thus . Thus and by assumption. Since
[TABLE]
we have . From this, one may easily check that satisfies all conditions given in Lemma 4.4 with , and thus we have . Since , we have for some integer and . This is a contradiction to the fact that , and we may conclude that there is a vector such that
[TABLE]
By changing signs of and by interchanging the role of and , if necessary, we may assume that there is a vector such that . If we let
[TABLE]
then one may easily show that . This completes the proof. ∎
Proposition 4.6**.**
The ternary triangular form is regular.
Proof.
Let be a ternary quadratic form and let be an integer such that . Note that
[TABLE]
By [21, Theorem 2.3] one may show that any integer congruent to 7 modulo 8 that is represented by is also represented by . Therefore, is represented by . Note that if , then
[TABLE]
Therefore, if there is a vector with , then we are done by Lemma 3.15. Thus we may assume that for any ,
[TABLE]
Suppose that for any . Let with . For a rational isometry
[TABLE]
of , we apply Lemma 4.4. Then we have . Since , we have for some integer . One may easily check that and . Hence there is a prime such that and . Then
[TABLE]
On the other hand,
[TABLE]
If we let , then
[TABLE]
Note that . By [1, Proposition 1], we have and thus . Thus and we are done with this case.
Now, suppose that there is a vector such that . We define
[TABLE]
Then, one may easily check that . ∎
Proposition 4.7**.**
The ternary triangular form is regular.
Proof.
Let be a ternary quadratic form and let be an integer such that . Note that
[TABLE]
Since , we have
[TABLE]
and thus . If , then
[TABLE]
Thus we may assume that for any ,
[TABLE]
First, suppose that there is a vector with . If we let
[TABLE]
then one may easily check that . Hence we may further assume that for any , .
Now, suppose that there is a vector with . Let . Then we have . Since , we have . By Lemma 4.3, there is a vector with such that . Thus such that or . By changing signs of , if necessary, we may assume that . If we let
[TABLE]
then one may easily check that . Therefore, we further assume that for any , .
Suppose that there is a vector such that or . Then one may check that by changing signs of , if necessary, we may assume that
[TABLE]
If , then we define
[TABLE]
If , then we define
[TABLE]
Then one may easily check that in each case. Now, we further assume that for any ,
[TABLE]
Suppose that there is a vector such that . By changing signs of and , if necessary, we may assume that and . If we let
[TABLE]
then and . This contradicts to our assumption (4.2). Therefore, we further assume that for any ,
[TABLE]
Take a vector with so that . If we let
[TABLE]
then one may easily check that
[TABLE]
If we let
[TABLE]
then clearly, , and thus . Note that . From this, one may show that satisfies all conditions given in Lemma 4.4 with , and thus we have . Since , we have for some integer with and . Thus there is a prime divisor of . Now for some odd integer . Note that
[TABLE]
Let . Then
[TABLE]
Note that . By [1, Proposition 1], we have , and this completes the proof. ∎
Proposition 4.8**.**
The ternary triangular form is regular.
Proof.
Let be a ternary quadratic form and let be an integer such that . Note that
[TABLE]
Since , we have . Let . We may assume that . Then .
First, assume that and . Then by Lemma 4.3, there is a vector with , and such that . So . By replacing by , if necessary, we may assume . If we let
[TABLE]
then one may easily check that .
Now, assume that and . Note that
[TABLE]
If we let , then
[TABLE]
Then by [1, Proposition 1], for any prime . Note that
[TABLE]
So there is a prime divisor of with . Thus we have .
Finally, assume that . If we let
[TABLE]
then one may easily check that . ∎
Proposition 4.9**.**
The ternary triangular form is regular.
Proof.
Let be a ternary quadratic form and let be an integer such that . Note that
[TABLE]
Since , we have . Let .
First, assume that . Then and thus . So
[TABLE]
Note that and . Since the triangular form is universal, there is a vector and thus .
Now, assume . Note that . Without loss of generality, we may assume that . Then
[TABLE]
Note that , . Since is regular by Proposition 4.8, there is a vector and thus .
Finally, assume that . Since , we may assume that , and . Since , we may further assume that . If we let
[TABLE]
then one may easily check that . ∎
Theorem 4.10**.**
There are exactly regular ternary triangular forms, which are listed in Table 4.
Proof.
For , we write . Let be a ternary quadratic form and let be any integer such that . In Theorem 3.16 and Propositions 4.54.9, we have already proved the regularity of each when
[TABLE]
Hence we may assume that is not contained in the above set. Note that for any integer which is not contained in , which we alreay considered in Propositions 4.54.9, the corresponding quadratic form has class number and thus . If , then one may easily show that . Hence in this case.
Now, we consider the case when . Note that if , then we have and . By Lemma 3.15, we have
[TABLE]
Since the proof of the case when is quite similar to this, we omit the proof.
Assume that . Since the quadratic form has class number and it primitively represents over , there is a vector
[TABLE]
Since , we have .
For the remaining , that is,
[TABLE]
one may check that can be obtained from a ternary triangular form whose regularity is already proved by taking -transformations several times for some . Furthermore, one may easily check that the regularity is preserved during taking the -transformation. This completes the proof. ∎
The reference list from the paper itself. Each links out to its DOI / PubMed record.
- 1[1] J. W. Benham and J. S. Hsia, Spinor equivalence of quadratic forms , J. Number Theory 17 (1983), 337-342.
- 2[2] M. Bhargava, On the Conway-Schneeberger fifteen theorem , Contemp. Math. 272 (2000), 27-38.
- 3[3] M. Bhargava and J. Hanke, Universal quadratic forms and the 290 theorem , preprint.
- 4[4] D. A. Burgess, On character sums and L-series, II, Proc. London Math. Soc. 13 (1963), 524-536.
- 5[5] W. K. Chan and B.-K. Oh, Finiteness theorems for positive definite n 𝑛 n -regular quadratic forms , Trans. Amer. Math. Soc. 355 (2003), 2385-2396.
- 6[6] W. K. Chan and B.-K. Oh, Almost universal ternary sums of triangular numbers , Proc. Amer. Math. Soc. 137 (2009), 3553-3562.
- 7[7] W. K. Chan and B.-K. Oh, Representations of integral quadratic polynomials , Contemp. Math. Am. Math. Soc. 587 (2013), 31-46.
- 8[8] W. K. Chan and J. Ricci, The representation of integers by positive ternary quadratic polynomials J. Number Theory 156 (2015), 75-94.
