The Pontryagin dual of the multiplicative group of positive rational numbers is constructed. Then we study its topological generators and representations.
Citation: Chaochao Sun, Yuanbo Liu. The Pontryagin dual of the multiplicative group of positive rational numbers[J]. AIMS Mathematics, 2023, 8(9): 20351-20360. doi: 10.3934/math.20231037
[1] | Yan Ling Fu, Wei Zhang . Some results on frames by pre-frame operators in Q-Hilbert spaces. AIMS Mathematics, 2023, 8(12): 28878-28896. doi: 10.3934/math.20231480 |
[2] | Faik Babadağ . A new approach to Jacobsthal, Jacobsthal-Lucas numbers and dual vectors. AIMS Mathematics, 2023, 8(8): 18596-18606. doi: 10.3934/math.2023946 |
[3] | Faik Babadağ, Ali Atasoy . A new approach to Leonardo number sequences with the dual vector and dual angle representation. AIMS Mathematics, 2024, 9(6): 14062-14074. doi: 10.3934/math.2024684 |
[4] | Keun Young Lee, Gwanghyun Jo . The dual of a space of compact operators. AIMS Mathematics, 2024, 9(4): 9682-9691. doi: 10.3934/math.2024473 |
[5] | Jin Zhong, Yilin Zhang . Dual group inverses of dual matrices and their applications in solving systems of linear dual equations. AIMS Mathematics, 2022, 7(5): 7606-7624. doi: 10.3934/math.2022427 |
[6] | Semra Kaya Nurkan, İlkay Arslan Güven . Construction of vectorial moments via direction curves. AIMS Mathematics, 2023, 8(6): 12857-12871. doi: 10.3934/math.2023648 |
[7] | Sezer Erdem . Compact operators on the new Motzkin sequence spaces. AIMS Mathematics, 2024, 9(9): 24193-24212. doi: 10.3934/math.20241177 |
[8] | Aifang Liu, Jian Wu . g-frame generator sets for projective unitary representations. AIMS Mathematics, 2024, 9(6): 16506-16525. doi: 10.3934/math.2024800 |
[9] | Rawya A. Hussein, Ali A. Ali . Geometry of the line space associated to a given dual ruled surface. AIMS Mathematics, 2022, 7(5): 8542-8557. doi: 10.3934/math.2022476 |
[10] | Qi Xiao, Jin Zhong . Characterizations and properties of hyper-dual Moore-Penrose generalized inverse. AIMS Mathematics, 2024, 9(12): 35125-35150. doi: 10.3934/math.20241670 |
The Pontryagin dual of the multiplicative group of positive rational numbers is constructed. Then we study its topological generators and representations.
We are inspired by Alain Connes' pioneering and excellent work [1], where he realized just those imaginary parts of zeros lying on Re(s)=12 of L-function to be the spectrums of the differential operator on a Hilbert space. Meanwhile, in loc.cit., he proposed a global trace formula in noncommutative geometry, which is equivalent to the Riemann hypothesis of the L-function L(χ,s). Nevertheless the trace formula may be hard to prove.
When we try to understand his work, we find a key thought is to deal with the action of the group R×+, i.e., the positive real numbers. Since the positive rational number group Q×+ is dense in R×+, the action of R×+ can be reflected by Q×+. On the other hand, Q×+≅⊕ppZ is the expression of the fundamental theorem of arithmetic by group language. Hence, it is an important object for both group theory and arithmetic. The Pontryagin dual ˆQ×+ of Q×+ is a very interesting space and the Hilbert function space on ˆQ×+ can give some information about L-function.
First, we calculate the Pontryagin dual group of the multiplicative group of positive rational numbers. Then we get some topological generators in the dual group. Next, we consider the Hilbert space of the dual group, in which we find a function attached to the Dirichlet L-function. For simplicity, we only consider the rational field. The idea can be generalized for general number fields, but it may be more complicated.
Denote R×+ the multiplicative group of positive real numbers and Q×+ the positive rational numbers. Then Q×+ is dense in R×+ under the usual topology. Since Q×+≅⊕ppZ, where pZ={pn: n∈Z}, the set {p: p is prime} is the set of topological generators of R×+.
The subgroup pZ is discrete in R×+. This fact can be obtained from the topological isomorphism
log: R×+→R, x↦logx. |
Since log(pZ)=Zlogp, we have R+/pZ≅R/Zlogp.
Let p,q be different primes. Then the group ⟨p,q⟩:={pmqn: m,n∈Z} is dense in R×+. This can be deduced by the following proposition.
Proposition 2.1. Let a,b∈R with b≠0, ab∉Q. Then Za+Zb={ma+nb: m,n∈Z} is dense in R.
Proof. Consider the map
f:Za+Zb→R≥0, x↦|x|, |
where R≥0 is the set of the nonnegative real numbers. If f(Za+Zb)∖{0} has a least absolute value, then Za+Zb is discrete in R. But any discrete subgroup of R is of the form Zx, x∈R. Since ab∉Q, the free abelian group Za+Zb is of rank 2. Hence, Za+Zb≠Zx for any x∈R. This means Za+Zb is dense in R.
Corollary 2.2. Let p,q be different primes. Then the group ⟨p,q⟩ is dense in R×+.
Proof. Under the isomorphism log: R×+→R, we have
log⟨p,q⟩≅Zlogp+Zlogq. |
Since logp/logq∉Q, we have that Zlogp+Zlogq is dense in R by Proposition 2.1. Hence ⟨p,q⟩ is dense in R×+.
Corollary 2.3. The cyclic group ⟨e2πiθ⟩={e2πinθ: n∈Z} is dense in S1:={z∈C: |z|=1} for those θ∉Q.
Proof. Consider the homomorphism
fθ: R→S1, x↦e2πiθx. |
Then the kernel of fθ is Z1θ. Since θ∉Q, the group Z+Z1θ is dense in R by Proposition 2.1. Hence ⟨e2πiθ⟩ is dense in S1.
Remark 2.4 (Kronecker's approximation theorem). There is a general form of Kronecker's approximation theorem, that is, the real numbers 1,a1,⋯,an are linearly independent over Q if and only if the set {(ma1,⋯,man): m∈Z} is dense in the torus Rn/Zn. One can see [2] for details or see Corollary 3.4.
Let G be a topological group. We denote by ˆG the Pontryagin dual of G. The topological group R×+ is topologically isomorphic to R. Since R is self-dual, so is R×+. The group pZ is the discrete topology. Under the commutative diagram
![]() |
we have ^pZ≅R×+/pZ. Further, the dual of Q×+ with the discrete topology is the compact group ∏pR×+/pZ. This can be obtained by the following well known theorem (see [3, Theorem 23.21], which is also a special case of [4, Proposition 1.1.13])
Theorem 2.5. Let {Gi}i be a set of compact groups. Then
^∏iGi≅⨁iˆGi. |
Corollary 2.6. Let Q×+ be given by the discrete topology. Then we have
^Q×+≅∏pR×+/pZ≅∏pR/Zlogp≅∏pR/Z1logp. |
Proof. Since ^pZ≅R×+/pZ, we have that ^R×+/pZ≅pZ by the property of Pontryagin dual. Moreover,
^∏pR×+/pZ≅⨁p^R×+/pZ≅⨁ppZ≅Q×+. |
Hence, we have ^Q×+≅∏pR×+/pZ. The second isomorphism in this corollary is deduced locally by the isomorphisms
R×+/pZlog→R/Zlogp. |
The third isomorphism in this corollary is deduced locally by the following isomorphism restricting on the group Zlogp
R⟶R, x⟼x(logp)2. |
The inverse limit of Z/mZ with respect to the natural maps is ˆZ:=lim←mZ/mZ=∏pZp, where Zp is the p-adic integers. We know that Z is dense in ˆZ and a topological generator is 1 in ˆZ.
There is a canonical homomorphism from Z to R/Zlogp defined by
τ: Z⟶R/Zlogp≅→S1, m⟼e2πim/logp. |
Then Z is dense in R/Zlogp by Corollary 2.3, this is because logp∉Q. Similarly, for the map
τ′: Z⟶R/Z1logp≅→S1, m⟼e2πimlogp, |
Z is dense in R/Z1logp. In general, we have the following theorem.
Theorem 3.1. For r∈Q∖{0}, let τr be the canonical homomorphism
τr: Z⟶∏pR/Z1logp, m⟼(rm)p, |
where (rm)p means that the p-th component is rm∈R/Z1logp. Then τr(Z) is dense in ∏pR/Z1logp. Hence, any nonzero rational number r is a topological generator of ∏pR/Z1logp.
Proof. It is easy to see that τr is injective. Let V be the closure of τr(Z) in ∏pR/Z1logp. If V≠∏pR/Z1logp, then ∏pR/Z1logp∖V is a nonempty open set. Let J be a finite set of primes such that for each p∈J, there is a nonempty open set Up satisfying (∏p∈JUp×∏p∉JR/Z1logp)∩V=∅. Consider the group ∏p∈JR/Z1logp and its open set ∏p∈JUp. Let VJ be the closure of τr(Z) in ∏p∈JR/Z1logp. Then we have VJ∩∏p∈JUp=∅. This means that τr(Z) is not dense in ∏p∈JR/Z1logp, which is a contradiction by the following claim.
Claim: τr(Z) is dense in ∏p∈JR/Z1logp for each finite set J of primes.
Consider the isomorphism
f:Rn→Rn, (a1,⋯,an)↦(a1logp1,⋯,anlogpn). |
Then we have f(∏ni=1Z1logpi)=∏ni=1Z. Also, f(1,⋯,1)=(logp1,⋯,logpn). By Kronecker's approximation theorem(see Remark 2.4), we just need to show that the numbers 1,logp1,⋯,logpn are linearly independent over Q. Suppose that there is
a0+a1logp1+⋯+anlogpn=0, |
where ai∈Q. Multiplying the above equation by a suitable integer, we can assume ai∈Z. Then we have
log(pa11⋯pann)=−a0. |
Taking exponential function on both sides, we have
pa11⋯pann=e−a0. |
Since we are assuming that a0∈Z, e−a0 is rational only when a0=0. Then by the fundamental theorem of arithmetic, we have a1=⋯=an=0, that is, the real numbers 1,logp1,⋯,logpn are linearly independent over Q. Hence, τr(Z) is dense in ∏p∈JR/Z1logp for each finite set J of primes.
This theorem shows that the group ∏pR/Z1logp is similar to the group ∏pZp. Both have a topological generator, but their topologies are very different. ∏pR/Z1logp is a torus with rank ∞, which is a compact connected Hausdorff space, while ∏pZp is a compact totally disconnected Hausdorff space.
Corollary 3.2. A topological generator of ∏pR/Zlogp is (⋯,(logp)2,⋯).
Proof. This follows from Theorem 3.1 and the isomorphism
∏pR/Z1logp⟶∏pR/Zlogp, (xp)p⟼(xp(logp)2)p. |
The following theorem is very useful.
Theorem 3.3. Let G be a compact group and ˆG be its character group. Then g∈G is a topological generator if and only if χ(g)≠1 for each nontrivial character χ∈ˆG.
Proof. See [5, Page 66,Paragraph 3], or [6, Theorem Ⅰ].
In fact, the above result is equivalent to Kronecker's approximation theorem in the torus case.
Corollary 3.4. The real numbers 1,a1,⋯,an are linearly independent over Q if and only if the set {(ma1,⋯,man): m∈Z} is dense in the torus Rn/Zn.
Proof. The character group of Rn/Zn is Zn. Let χ∈^Rn/Zn. Then χ can be expressed by
χ(t1,…,tn)=n∏j=1e2πimjtj=e2πi∑nj=1mjtj, |
where mj∈Z,tj∈R/Z. We can view this χ as the vector mχ=(m1,⋯,mn). Then by Theorem 3.3, (a1,⋯,an) is a topological generator of Rn/Zn if and only if
χ(a1,⋯,an)=e2πi∑nj=1mjaj≠1 |
for every nontrivial character χ.
Let (a1,⋯,an) be a topological generator of Rn/Zn. Suppose m0+∑nj=1mjaj=0, where mj∈Q. Moreover, we can assume mj∈Z. If m1=⋯=mn=0 does not hold, then we get a nontrivial character mχ=(m1,⋯,mn). For this nontrivial character χ, we have χ(a1,⋯,an)=1, which is a contradiction. Hence, m1=⋯=mn=0, implying that m0=0. This means 1,a1,⋯,an are linearly independent over Q.
Let 1,a1,⋯,an be linearly independent over Q. If (a1,⋯,an) is not a topological generator of Rn/Zn, then there exists a nontrivial character χ, i.e., mχ=(m1,⋯,mn)≠0 such that
χ(a1,⋯,an)=e2πin∑j=1mjaj=1. |
Then we have ∑nj=1mjaj∈Z. Hence there exists some m∈Z such that m+∑nj=1mjaj=0. Since m1,⋯,mn are not all 0, this means 1,a1,⋯,an are not linearly independent over Q, which is a contradiction. Therefore, (a1,⋯,an) is a topological generator of Rn/Zn.
Lemma 3.5. Let J be a finite set of primes, kp∈N, and np∈Z such that np=0 or (p,np)=1. If
∑p∈Jnp/pkp=n |
for some n∈Z, then we have np=0 for all p∈J.
Proof. Consider the equation
1=e2πi∑p∈Jnp/pkp=∏p∈Je2πinp/pkp. |
Take q∈J and let
mq=∏p∈J,p≠qpkp. |
Then we have
1mq=e2πinqmq/qkq. |
Since (mq,q)=1, we have nq=0 for all q∈J.
Remark 3.6. Another proof of Lemma 3.5 suggested by one of reviewers is as following: Multiply ∏p∈Jpkp on the equation ∑p∈Jnp/pkp=n. Then one has ∑p∈Jnpmp=n∏p∈Jpkp, where mp=1pkp∏p∈Jpkp. Modulo p for this equation, one has npmp≡0modp. Since (mp,p)=1, we have np=0 for all p∈J by the condition on np.
Denote
Tp=R/Z1logp. |
Then we have the following corollary.
Corollary 3.7. Consider the compact group
∏pTp×Zp. |
Then the image of the diagonal map
Z⟶∏pTp×Zp, n⟼(⋯,n,n,⋯) |
is dense in ∏pTp×Zp.
Proof. The character group of Tp (resp. Zp) is Z1logp (resp. Z(p∞), see [3,(25.2)], which is isomorphic to Qp/Zp. Here Qp is the p-adic field). A character χ1 of Tp can be expressed by χ1(t)=e2πimtlogp, where m∈Z,t∈Tp. However, a character χ2 of Zp can be expressed by χ2(t)=e2πim˜t/pk, where m∈Z,k∈N,t∈Zp, ˜t∈Z such that ˜t≡tmodpk, (m,p)=1 or m=0. Hence, by Theorem 2.5, any character χ of ∏pTp×Zp can be written as follows
χ((tp)p,(sp)p)=∏p∈Je2πimptplogpe2πinp˜sp/pkp, |
where J is some finite set of primes, mp,np∈Z,kp∈N,tp∈Tp,sp∈Zp,˜sp∈Z such that ˜sp≡spmodpkp.
Supposing
1=χ((1,1⋯))=e2πi(∑p∈Jmplogp+∑p∈Jnp/pkp), |
we obtain
∑p∈Jmplogp+∑p∈Jnp/pkp∈Z. |
So, for some n∈Z, there is
log(∏p∈Jpmp)=n−∑p∈Jnp/pkp, |
that is,
∏p∈Jpmp=en−∑p∈Jnp/pkp. | (3.1) |
Since e is a transcendental number, the right side of the above equation is a rational number only if
n−∑p∈Jnp/pkp=0. |
From Lemma 3.5, we have np=0 for all p∈J.
Furthermore, by (3.1), we obtain
∏p∈Jpmp=1. |
The fundamental theorem of arithmetic implies that mp=0, ∀p∈J.
This indicates that χ is a trivial character. Hence, by Theorem 3.3, the image of Z is dense in ∏pTp×Zp.
Let G be a compact abelian Lie group. All the irreducible representations of G are of dimension 1. Let L2(G) be the Hilbert space of G, which consists of square integrable complex functions. Let ˆG denote the character group of G. Each function f∈L2(G) can be uniquely expressed by the series
f=∑φ∈ˆGcφ⋅φ, |
where cφ∈C (see [7,§17.B]).
For G=R/Z, the character group ˆG≅Z, because the characters φn: G→S1 are of the form
φn: R/Z→S1, θ↦e2πinθ. |
For G=R/Zlogp, the characters of G are
φnp,p: R/Zlogp→S1, θ↦e2πinpθlogp. | (4.1) |
Hence, for G=∏pR/Zlogp, the character of G is of the form
φ: ∏pR/Zlogp→S1, (θp)p↦∏pe2πinpθplogp=e2πi∑pnpθplogp, | (4.2) |
where np=0 for almost all p by Theorem 2.5. The character of ∏pR/Zlogp is φ as in Eq (4.2). All the characters {φ} form a complete orthogonal base of L2(∏pR/Zlogp). Then for each integrable function f∈L2(∏pR/Zlogp), we obtain its Fourier series (see [8, §17])
f((θp)p)=∑φ∈ˆGcφe2πi∑pnp,φθplogp | (4.3) |
where G=∏pR/Zlogp, cφ=∫Gf(g)e−2πi∑pnp,φθplogpdg, the Haar measure dg of G such that ∫Gdg=1. An integrable function f∈L2(∏pR/Zlogp) is uniquely determined by its Fourier series. Furthermore, for each f∈L2(∏pR/Zlogp), we have the Parseval equation (see [8, §17])
∫G|f(g)|2dg=∑φ∈ˆG|cφ|2, |
where G=∏pR/Zlogp.
Take a Dirichlet character χ: (Z/mZ)×⟶S1. The L-function of χ is
L(χ,s)=∞∑n=1χ(n)ns=∏p11−χ(p)p−s, |
which is absolutely convergent on Re(s)>1. For p dividing m, we set χ(p)=0. So, taking the logarithm, we have
logL(χ,s)=−∑plog(1−χ(p)p−s)=∑p∞∑n=1χ(pn)npns,Re(s)>1, | (4.4) |
where we take a branch of the logarithm such that log(R×+)=R.
From (4.4), we have
−L′(χ,s)L(χ,s)=∑p∞∑n=1χ(pn)logppns,Re(s)>1. |
Theorem 4.1. For each s∈C with Re(s)>12, the value logL(χ,s) corresponds to a function
fs,χ(θp)p=∑p∞∑n=1χ(pn)√logpnpnse2πinθplogp√logp∈L2(∏pR/Zlogp). |
Furthermore, fs,χ(0)=logL(χ,s) for Re(s)>1.
Proof. Consider the Eq (4.3). For each character φn,p in Eq (4.1), which can be seen as the character of ∏pR/Zlogp through the projection ∏pR/Zlogp→R/Zlogp, we take cp,n=χ(pn)√logpnpns; for the other remaining character φ of ∏pR/Zlogp, we let cφ=0. Then we have
∑p∞∑n=1χ(pn)√logpnpns¯(χ(pn)√logpnpns)≤∑p∞∑n=1logpp2nRe(s)=−ζ′(2Re(s))ζ(2Re(s)), |
which is absolutely convergent when Re(s)>1/2. Hence, we have an integrable function
fs,χ=∑p∞∑n=1χ(pn)√logpnpnse2πinθplogp√logp∈L2(∏pR/Zlogp). |
It is easy to see fs,χ(0)=logL(χ,s) for Re(s)>1.
We calculate the Pontryagin dual of the multiplicative group of positive rational numbers under the discrete topology. We find some canonical topological generators in the dual group. Besides, we find some canonical generator in other compact group. We study the representation of the dual group the multiplicative group of positive rational numbers. Furthermore, in the Hilbert function space of the dual group we find a function attached to the Dirichlet L-function, which may be a new way to understand the Dirichlet L-function.
The authors declare they have not used Artificial Intelligence (AI) tools in the creation of this article.
We would like to thank Xin Zhang for discussing the first version of the manuscript. We also appreciate the referee's comments to improve the previous version.
The authors declare that there are no conflicts of interest regarding the publication of this paper.
[1] |
A. Connes, Trace formula in noncommutative geometry and the zeros of the Riemann zeta function, Sel. Math., 5 (1999), 29. https://doi.org/10.1007/s000290050042 doi: 10.1007/s000290050042
![]() |
[2] |
S. M. Gonek, H. L. Montgomery, Kronecker's approximation theorem, Indag. Math., 27 (2016), 506–523. https://doi.org/10.1016/j.indag.2016.02.002 doi: 10.1016/j.indag.2016.02.002
![]() |
[3] | E. Hewitt, K. A. Ross, Abstract Harmonic Analysis-Volume I: Structure of Topological Groups Integration Theory Group Representations, Berlin, Heidelberg: Springer, 1963. |
[4] | J. Neukirch, A. Schmidt, K. Wingberg, Cohomology of Number Fields, 2 Eds., Springer, 2013. |
[5] | S. Hartman, Remarks on equidistribution on non-compact groups, Compos. Math., 16 (1964), 66–71. |
[6] |
P. R. Halmos, H. Samelson, On monothetic groups, Proc. Natl. Acad. Sci. USA, 28 (1942), 254–258. https://doi.org/10.1073/pnas.28.6.254 doi: 10.1073/pnas.28.6.254
![]() |
[7] | I. I. Shafarevich, Basic Notation of Algebra, Higher Education Press, 2017. |
[8] |
B. Jessen, The theory of integration in a space of an infinite number of dimensions, Acta Math., 63 (1934), 249–323. https://doi.org/10.1007/BF02547355 doi: 10.1007/BF02547355
![]() |