references:
- Deitmar A. A first course in harmonic analysis[M]. 2005.
- Ideal quotient | encyclopedia article by TheFreeDictionary
- Fractional ideal | encyclopedia article by TheFreeDictionary
- Pontryagin duality | encyclopedia article by TheFreeDictionary
- Algebraic number field | encyclopedia article by TheFreeDictionary
- Duality and Subgroups on JSTOR
- The Duality of Groups - Zhihu (zhihu.com)
- Lattice password 1.6 Dual space and dual lattice - Zhihu (zhihu.com)
- Dual space (1): space, basis and mapping - Zhihu (zhihu.com)
- Explanation of dual lattice - Zhihu (zhihu.com)
- Commutative Ring Theory (6) Noether Ring - Zhihu (zhihu.com)
- Commutative ring theory (8): Dedejin whole ring - Zhihu (zhihu.com)
- [Abstract Algebra] 18. Direct products and direct sums of modules, free modules and projective modules - Zhihu (zhihu.com)
- Continuous mapping and homeomorphism - Zhihu (zhihu.com)
- Metric space (metric space)-CSDN Blog
- Algebraic Structure: Module-CSDN Blog
- Algebraic Geometry: Irreducible Varieties-Prime Ideals-CSDN Blog
Article directory
fractional ideal
Let RRR is an integral ring, and itsfieldof fractions is denoted byqf ( R ) : = { ab ∣ a , b ∈ R } qf(R):=\{\dfrac{a}{b}| a,b \in R\}qf(R):={ ba∣a,b∈R}
Fractional ideal : it is qf ( R ) qf(R)AnRR of q f ( R )R -SubmoduleIII , there exists a ring elementr ∈ R r \in Rr∈R , such thatr I ⊆ R rI \subseteq RrI⊆R is an (integral) ideal.
Principal fraction ideal (principal): as RRR - submodule, consisting of a single element of the fractional domainx ∈ qf ( R ) x \in qf(R)x∈q f ( R ) is generated, in the formI = R x I=RxI=Rx
Attention set I ⊆ qf ( R ) I \subseteq qf(R)I⊆q f ( R ) is not an ideal of the fractional domain (only trivial ideals). Taker = er=er=e is a unit, and all (integral) ideals can be regarded as special fractional ideals. Ifx ∈ R x \in Rx∈R useI = R x I=RxI=R x , then it is an (integer) principal ideal. RR_The fractional ideal contained in R is the (integral) ideal.
Similar to the ideal sum, intersection and product, define the integral ring RRR fraction idealI , JI,JI,J 的运算,
I + J : = { a + b ∣ a ∈ I , b ∈ J } I J : = { ∏ i = 1 n a i b i ∣ a i ∈ I , b i ∈ J , n ∈ Z + } I ∩ J : = { a ∣ a ∈ I , a ∈ J } \begin{aligned} I+J &:= \{a+b| a \in I,b \in J\}\\ IJ &:= \{\prod_{i=1}^na_ib_i| a_i \in I,b_i \in J,n \in \mathbb Z^+\}\\ I \cap J &:= \{a|a \in I,a \in J\} \end{aligned} I+JIJI∩J:={
a+b∣a∈I,b∈J}:={
i=1∏naibi∣ai∈I,bi∈J,n∈Z+}:={
a∣a∈I,a∈J}
In addition, we define the formal quotient of the fractional ideal , (
J : I ) : = { a ∈ qf ( R ) ∣ a I ⊆ J } (J:I) := \{a \in qf(R )| aI \subseteq J\}(J:I):={
a∈qf(R)∣aI⊆J}
易知 I ( J : I ) ⊆ J I(J:I) \subseteq J I(J:I)⊆J. _ Formal quotient( J : I ) (J:I)(J:I ) isRRR - submodule, not necessarily a fractional ideal; ifI, JI,JI,J are all (integer) ideals, then( J : I ) (J:I)(J:I ) is also ideal. If there exists a fractional idealJJJ makesIJ = R IJ=RIJ=R , we call the fractional idealIII isinvertibleideal. Regardless of the fractional idealIIWhether I is invertible, we define itsformal inverse,
I − 1 : = ( R : I ) I^{-1}:=(R:I)I−1:=(R:I)
Zero ideal OOThe formal inverse of O is the fraction fieldqf ( R ) qf(R)q f ( R ) (not a fractional ideal), non-zero fractional idealI ≠ OI \neq OI=The formal inverse of O is a fractional ideal (andI − 1 I ⊆ RI^{-1}I \subseteq RI− 1 I⊆R is ideal).
For any whole ring RRR , has the following properties
-
Fractional field qf ( R ) qf(R)q f ( R ) Any finitely generatedRRR - submodules, all fractional ideals
-
Good luck III is finitely generated, then any( J : I ) (J:I)(J:I ) is a fractional ideal (soI − 1 I^{-1}I− 1 is the inverse ideal)
-
All reversible fractional ideals are finitely generated
-
I + J I+J I+J和IJ IJI and J are still fractional ideals, butI ∩ JI \cap JI∩JNot necessarily
-
令 x ∈ q f ( R ) x \in qf(R) x∈q f ( R ) , principal fraction idealI = R x I=RxI=R x is reversible, and the inverse ideal isI − 1 = R x − 1 I^{-1}=Rx^{-1}I−1=Rx−1
-
Nonzero Fractional Ideals III is reversible if and only ifI ⊆ qf ( R ) I \subseteq qf(R)I⊆q f ( R ) is the projective module
-
Free mold (free): RRR -moduleMMM has a set ofbasis(basis,RRR - linearly independent finite generator)
-
Projective mold (projective): Projective mold PPP , for any two modulesM , NM,NM,N and the full modular homomorphismσ : M → N \sigma:M \to Np:M→N , for any modular homomorphismϕ : P → N \phi:P \to Nϕ:P→N , there exists a modular homomorphismψ \psiψ σ ∘ψ = ϕ \sigma \circ \psi=\phip∘p=ϕ
-
R R R -diePPP is projective if and only if there is a free moduleFFF and modular homomorphismα : F → P , β : P → F \alpha:F \to P, \beta:P \to Fa:F→P,b:P→F solver∘ β = 1 P \soft \circ \beta=1_Pa∘b=1P
-
The projective mode can be written as the direct sum of the free modes
-
For special rings, they have the following properties
- If RRR is a Noether ring, then any fractional ideal is finitely generated.
- Noetherian : commutative ring RRAll (integral) ideals of R are finitely generated
- If RRR is a local ring, then any fractional ideal is principal
- Local ring (Local): exchange ring RRR has only one great ideal
令 F ( R ) \mathcal F(R) F ( R ) is the collection of allnon-zero fractional ideals, letP ( R ) \mathcal P(R)P ( R ) is the ideal collection of allreversible fractions, letP rin ( R ) Prin(R)P r in ( R ) is the ideal collection of allprincipal fractions. Regarding the product of fractional ideals:
- F ( R ) \mathcal F(R) F ( R ) isa commutative monoid group(closed, combined), unitaryRRR
- P ( R ) \mathcal P(R) P ( R ) isan Abelian group(closed, associative, unitary, reversible),P rin ( R ) Prin(R)P r in ( R ) is itssubgroup
Dedekind domains: Integral RRAll nonzero fractional ideals of R are reversible
- Number field KKThe integer ring of K is OK \mathcal O_KOKIt's a Dede gold ring
- Ring of integer: Finite algebra extension K = Q ( e 1 , ⋯ , en ) K=\mathbb Q(e_1,\cdots,e_n)K=Q(e1,⋯,en) all integer elementsz 1 e 1 + ⋯ znen , ∀ zi ∈ Z z_1e_1+\cdots z_n e_n,\forall z_i \in \mathbb Zz1e1+⋯znen,∀zi∈The ring composed of Z is denoted as OK \mathcal O_KOK
- Gaussian rationals Q ( i ) \mathbb Q(i) Q ( i ) ,Cyclotomic field Q ( ζ n ) \mathbb Q(\zeta_n)Q ( gn)
- square-free integer ddd (anys 2 ∣ ds^2|ds2∣d 都有 s ∈ R s \in R s∈R belongs to the unit),Quadratic field Q ( d ) \mathbb Q(\sqrt d)Q(d),
Duality group
Duality of finite groups
Let GGG isa finite Abelian group,T = { e 2 π ix ∣ x ∈ R } ⊆ C \mathbb T=\{e^{2\pi ix}|x \in \mathbb R\} \subseteq \mathbb CT={ e2πix∣x∈R}⊆C isthe unit torus(continuous multiplicative group), including the unit cyclic group∀ n , ( ζ n ) ⊆ T \forall n,(\zeta_n) \subseteq \mathbb T∀n,( gn)⊆T
Character is a group homomorphism χ : G → T \chi: G \to \mathbb Th:G→T ,
χ ( ab ) = χ ( a ) χ ( b ) \chi(ab) = \chi(a)\chi(b)χ(ab)=x ( a ) x ( b )
Let the set G ^ \hat GG^ is the collection of all features, we define the multiplication (pointwise product) of the group homomorphism( χ , η ) ↦ χ η (\chi,\eta) \mapsto \chi\eta( x ,h )↦χ,
χ η ( a ) = χ ( a ) η ( a ) , ∀ a ∈ G \chi\eta(a) = \chi(a)\eta(a),\,\, \forall a \in Gchi ( a )=x ( a ) h ( a ) ,∀a∈G
Dual group : algebraic structure ( G ^ , ⋅ ) (\hat G,\cdot)(G^,⋅ ) becomes an Abelian group, also called Pontryagin dual
For the cyclic group G = ( g ) G=(g)G=( g ) , iford ( g ) = N ord(g)=Nord(g)=N , then the group homomorphism
χ l ( g ) = ζ N l , l = 0 , 1 , ⋯ , N − 1 \chi_l(g) = \zeta_N^l,\,\, l=0,1,\cdots ,N-1hl(g)=gNl,l=0,1,⋯,N−1
are all the characteristics, the dual group G ^ = ( χ 1 ) \hat G=(\chi_1)G^=( x1) is alsoNNCyclic group of order N , so there isG ≅ G ^ G \cong \hat GG≅G^
Bidual : mapping A → A ^ ^ A \to \widehat{\hat A}A→A^
义为a ↦ δ aa \mapsto \delta_aa↦da, in which δ a \delta_adaδ
a : A ^ → T χ ↦ χ ( a ) \begin{aligned} \delta_a: \hat A &\to \mathbb T\\ \chi &\mapsto \chi(a) \end{aligned } }da:A^h→T↦x ( a )
A → A ^ ^ A \ to \widehat{\hat A}A→A^ is a canonical isomorphism
Compactness & LCA Group
Metric space : set XXX and its measureddStructure composed of d (X, d) (X,d)(X,d)
For an (infinite) sequence (xn) in a set ⊆ X (x_n) \subseteq X(xn)⊆X , we say itconvergestox ∈x∈X,如果
∀ ϵ > 0 , ∃ N ∈ Z , ∀ n > N , d ( x n , x ) < ϵ \forall \epsilon>0,\exists N \in \mathbb Z,\forall n>N,d(x_n,x) < \epsilon ∀ϵ>0,∃N∈Z,∀n>N,d(xn,x)<ϵ
Two metric spaces X, YX,YX,Mapping f between Y : X → Y f:X \to Yf:X→Y , we say it iscontinuousif it: maps the convergent sequence to the convergent sequence and maintains the limit
lim n → ∞ f ( xn ) = f ( lim n → ∞ xn ) \lim_{n \ to \infty} f(x_n) = f(\lim_{n \to \infty} x_n)n→∞limf(xn)=f(n→∞limxn)
We say that the two measures d 1 , d 2 d_1,d_2d1,d2is equivalent if for all (xn) (x_n)(xn) , both have:(xn) (x_n)(xn) ind 1 d_1d1Lower convergence ⟺ \iff⟺ ( x n ) (x_n) (xn) ind 2 d_2d2Lower convergence, denoted as d 1 ∼ d 2 d_1 \sim d_2d1∼d2
Homeomorphism: also called bicontinuous function . Topological space (X, OX), (Y, OY) (X,\mathcal O_X),(Y,\mathcal O_Y)(X,OX),(Y,OYFunction f on ) : X → Y f:X \to Yf:X→Y , satisfies three properties:fff is a bijection,fff is a continuous function,f − 1 f^{-1}f− 1 is a continuous function.
For the metric space (X, d) (X,d)(X,Automorphism fon d ) : X → X f:X \to Xf:X→X , then define a new metric
d ′ ( x , y ) : = d ( f ( x ) , f ( y ) ) d'(x,y):=d(f(x),f(y))d′(x,y):=d(f(x),f(y))
It satisfies d ′ ∼ d d' \sim dd′∼d is equivalent.
Since f ( x ) = x / ( x + 1 ) f(x)=x/(x+1)f(x)=x/(x+1 ) is[ 0 , ∞ ) → [ 0 , 1 ) [0,\infty) \to [0,1)[0,∞)→[0,1 ) Monotone homeomorphismon, then: any metricddd , there exists an equivalent metricd ′ ( x , y ) : = d ( x , y ) d ( x , y ) + 1 d'(x,y):=\dfrac{d(x,y)} {d(x,y)+1}d′(x,y):=d(x,y)+1d(x,y), such that I m ( d ′ ) = [ 0 , 1 ) Im(d')=[0,1)I m ( d′)=[0,1 ) . According to the equivalence relationship, we combine all measuresddd is divided into equivalence classes[d] [d][ d ] , we call( X , [ d ] ) (X,[d])(X,[ d ]) isa measurable space(metrizable space)
A measurable space ( X , [ d ] ) (X,[d])(X,[ d ]) iscompactif any sequence(xn) (x_n)(xn) , both contain a convergedsubsequence(fromxn x_nxnPick some points infinitely). Example: R n \mathbb R^nRA bounded closed subset of n ; a discrete space is compact if and only if it is finite.
Two weaker compacts:
- We say XXX is σ \sigmaσ - compact, if there exists a compact subset sequenceK n ⊂ K n + 1 K_n \subset K_{n+1}Kn⊂Kn+1Such that X = ⋃ n K n X=\bigcup_n K_nX=⋃nKn, this sequence (K n) (K_n)(Kn) is calledcompact exhaustion. Easy to knowR \mathbb RR can chooseK n = [ − n , n ] K_n=[-n,n]Kn=[−n,n ] , this sequence also exists in countable discrete spaces.
- We say XXX iscompactif for any pointx ∈x∈X , there existsr > 0 r>0r>0 makes the closed ballB ˉ r ( x ) \bar B_r(x)Bˉr( x ) is compact. Example: spaceR n \mathbb R^nRn , discrete space; counterexample: infinite-dimensional Hilbert space.
- If the metric space simultaneously σ \sigmaσ -compact and locally compact, it is called σ \sigmaσ - locally compact. Example:R \mathbb RR, R / Z \mathbb R/\mathbb Z R / Z , a countable discrete space.
Measurable abelian group : Abelian group AAA , metric equivalence class[d] [d][ d ] , and satisfy( x , y ) ↦ xy (x,y) \mapsto xy(x,y)↦x y andx ↦ x − 1 x \mapsto x^{-1}x↦x− 1 are all continuous functions (map the convergence sequence to the convergence sequence and maintain the limit)
LCA group : measurable σ \sigmaσ - locally compact Abelian group. Example: countable Abelian group + discrete metric, real number fieldR \mathbb RR , real torusR / Z \mathbb R/\mathbb ZR/Z
-
A compact exhaustion sequence (K n) (K_n)(Kn) is calledabsorbing, if for any compact subsetK ⊂ AK \subset AK⊂A , there exists a certainn ∈ N n \in \mathbb Nn∈N makesK ⊂ K n K \subset K_nK⊂Kn
LCA groups always contain an absorbing exhaustion
-
Subset D ⊂ XD \subset X of measurable spaceD⊂X is calleda densesubset. For anyx ∈x∈X , there always existsxn ∈ D x_n \in Dxn∈D makes the sequence(xn) (x_n)(xn) converges toxxx
The LCA group always contains a countably dense subset
Note that continuous functions and dense subsets are relative to the "convergence sequence", and the measure between different points is d (x, y) d(x,y)d(x,y ) can be discrete as long asn → ∞ n \to \inftyn→When ∞ the metric tends to 0 00 is enough.
Dual of LCA group
ReiAA _A is the LCA group,and its characteristicsare defined as the continuous group homomorphismχ : A → T \chi:A \to \mathbb Th:A→T (maps convergence sequences to convergence sequences while maintaining limits). Collection of all featuresA ^ \hat AA^,附带上乘法乘法χ η ( a ) = χ ( a ) η ( a ) \chi\eta(a) = \chi(a)\eta(a)chi ( a )=χ ( a ) η ( a ) is calledthe dual group.
- 群 Z \mathbb Z Z function:x ↦ e 2 π ixy , y ∈ R / Z x \mapsto e^{2\pi i xy},\,\, y \in \mathbb R/\mathbb Zx↦e2πixy,y∈R / Z , thenZ ^ ≅ R / Z \hat{\mathbb Z} \cong \mathbb R/\mathbb ZZ^≅R/Z
- 群 R / Z \mathbb R/\mathbb Z The function of R / Z :x ↦ e 2 π ixy , y ∈ Z x \mapsto e^{2\pi i xy},\,\, y \in \mathbb Zx↦e2πixy,y∈Z , then there isR / Z ^ ≅ Z \widehat{\mathbb R/\mathbb Z} \cong \mathbb ZR/Z ≅Z
- Group R \mathbb RR function:x ↦ e 2 π ixy , y ∈ R x \mapsto e^{2\pi i xy},\,\, y \in \mathbb Rx↦e2πixy,y∈R , soR ^ ≅ R \hat{\mathbb R} \cong \mathbb RR^≅R
The dual of the LAC group is still the LAC group . Given AAAn absorbing compact exhaustion of A A = ⋃ n K n A=\bigcup_n K_nA=⋃nKn, for the elements χ , η \chi,\eta in the dual groupx ,η , defineA ^ \hat AA^ Definition:
d ^ n ( χ , η ) = sup x ∈ K n ∣ χ ( x ) − η ( x ) ∣ d ^ ( χ , η ) = ∑ nd ^ n ( χ , η ) 2 n \hat d_n(\chi,\eta) = \sup_{x\in K_n} |\chi(x)-\eta(x)|\\ \hat d(\chi,\eta) = \sum_n \dfrac{ \hat d_n(\chi,\eta)}{2^n}d^n( x ,h )=x∈Knsup∣χ(x)−h ( x ) ∣d^ (x,h )=n∑2nd^n( x ,h ).
Group isomorphism R / Z → Z ^ \mathbb R/\mathbb Z \to \widehat{\mathbb Z}R/Z→Z , Z → R / Z ^ \mathbb Z \to \widehat{\mathbb R/\mathbb Z} Z→R/Z , R → R ^ \mathbb R \to \widehat{\mathbb R} R→R All are homeomorphic (bicontinuous).
ReiAA _A is the LCA group (σ \sigmaσ -compact + locally compact): ifAAA is compact, thenA ^ \hat AA^ discrete; ifAAA is discrete, thenA ^ \hat AA^ Compact.
Pontryagin Duality : Any LAC group, canonically isomorphic to its bidual, mapping A → A ^ ^ A \to \widehat{\hat A}A→A^
Let
a ↦ δ a , δ a ( χ ) = χ ( a ) a \mapsto \delta_a,\,\, \delta_a(\type)=\type(a) .a↦da,da( x )=x ( a )
It is two LCA groups A ≅ A ^ ^ A \cong \widehat{\hat A}A≅A^ group homomorphisms between.
Pontryagin Duality theorem : an LCA group GGG , orderH⊂ GH \subset GH⊂G is a closed subgroup, letH ⊥ ⊂ G ^ H^\perp \subset \hat GH⊥⊂G^ is the closed subgroup of the dual group, then the duality relation⋅ ^ \hat \cdot⋅^ defines the bijection between the subgroup and the quotient group,
H ^ = G ^ / H ⊥ , G / H ^ = H ⊥ \hat H = \hat G/H^\perp,\,\, \widehat{ G/H}=H^\perpH^=G^/H⊥,G/H
=H⊥
dual space
Let VVV is the domainF \mathbb FLinear space on F (Abelian group, F \mathbb FF -module),the linear functionalis the linear mapV → FV \to \mathbb FV→F,defining operation
( α + β ) ( v ) = α ( v ) + β ( v ) ( k α ) ( v ) = k ⋅ α ( v ) (\alpha+\beta)(v)=\alpha(v )+\beta(v)\\ (k\alpha)(v)=k \cdot \alpha(v)( a+b ) ( v )=a ( v )+b ( v )(kα)(v)=k⋅a ( v )
Dual space : Let V ∗ = H om ( V , F ) V^*=Hom(V,\mathbb F)V∗=Ho m ( V ,F ) is the collection of all linear functionals, and it is easy to verify that it constitutes a linear space.
Dual basis: For a finite-dimensional space dim V = n \dim V=ndimV=n,Nyogaeie_ieiIt’s VVA set of bases of V , thenα i \alpha^iai isV ∗ V^*VA set of basis for ∗ , where α i ( ej ) = δ ij \alpha^i(e_j)=\delta_{ij}ai (ej)=dij, where δ ij \delta_{ij}dijis an indicator function. It can be verified that dim V = dim V ∗ \dim V = \dim V^*dimV=dimV∗ , thenV ≅ V ∗ V \cong V^*V≅V∗ ,and↦ α i e_i \mapsto \alpha^iei↦ai
Dual map: two spaces V, WV, WV,Linear mapping f between W : V → W f:V \to Wf:V→W , α : V → F , β : W → Fin their dual spacesa:V→F,b:W→F can be mappedf ∗ : W ∗ → V ∗ f^*:W^* \to V^*f∗:W∗→V∗ Correlation, defined asf ∗ ( β ) = β ∘ ff^*(\beta)=\beta \circ ff∗ (b)=b∘f , which is also a linear map and satisfies
( f + g ) ∗ = f ∗ + g ∗ ( kf ) ∗ = kf ∗ ( fg ) ∗ = g ∗ f ∗ (f+g)^* = f^*+g ^*\\ (kf)^* = kf^*\\ (fg)^* = g^*f^*(f+g)∗=f∗+g∗(kf)∗=kf∗(fg)∗=g∗f∗
Mapping of linear space to its bidual ϕ : V → ( V ∗ ) ∗ \phi:V \to (V^*)^*ϕ:V→(V∗)∗ , so that it satisfies
( ϕ ( v ) ) ( α ) = α ( v ) (\phi(v))(\alpha)=\alpha(v)( ϕ ( v )) ( a )=a ( v )
An injection that can be proven to be an injection and whose construction does not depend on the choice of basis is called a canonical injection . For finite dimension dim V = dim ( V ∗ ) ∗ \dim V = \dim (V^*)^*dimV=dim ( V∗)∗ , mappingϕ \phiϕ is also a bijection (not infinitely dimensional), which is calledcanonicalisomorphism. V ≅ V ∗ V \cong V^*V≅V∗ is not canonical, butV ≅ ( V ∗ ) ∗ V \cong (V^*)^*V≅(V∗)∗ is exemplary.
Fixed mapping, the original image and the image have a certain "mapping" relationship (each original image only corresponds to one image); similarly, when we fix the original image, there is also a certain mapping relationship between the mapping and the image. Therefore, there is a certain symmetrical relationship between the original image and the mapping. We call this symmetrical relationship "duality" and abstract the space where it is located into the dual space.
The dual relationship is symmetric, and the linear functional can be written as f ( x ) = ⟨ f , x f(x) = \langle f,x \ranglef(x)=⟨f,x better represents itsbilinearity(Bilinear). For the finite-dimensional linear spaceVVV , its dual spaceV ∗ V^*V∗ Isomorphism with it, we can loosely definex ∈ V x \in Vx∈V 和 f ∈ V ∗ f \in V^* f∈V∗ Place them in the same space.
Duality
A lattice is a linear space R m \mathbb R^mRdiscrete subspace in m , then it is a linear space (infinite Abelian group).
格Λ ⊆ R m \Lambda \subset \mathbb R^mL⊆Rm ,格基 B ∈ R m × n B \in \mathbb R^{m \times n} B∈Rm × n ,dim Λ = n ≤ m \dim \Lambda=n \le mdimL=n≤m , its dual is defined as the set
Λ ⊥ = { y ∈ S pan ( Λ ) : ∀ x ∈ Λ , ⟨ x , y 〉 ∈ Z } \Lambda^\perp=\{y \in Span(\Lambda): \forall x \in \Lambda,\langle x,y \rangle \in \mathbb Z\}L⊥={
y∈Span ( L ) _:∀x∈L ,⟨x,y⟩∈Z}
It is a dual lattice, the basis of the lattice is D = B ( BTB ) − 1 ∈ R m × n D=B(B^TB)^{-1} \in \mathbb R^{m \times n}D=B(BTB)−1∈Rm × n ,denotedim Λ ⊥ = dim Λ \dim \Lambda^\perp = \dim \LambdadimL⊥=dimL
The relationship between the two:
- The lattice bases are pseudo-inverses of each other, BTD = DTB = IB^TD=D^TB=IBTD=DTB=I
- constant c > 0 c >0c>0 ,free( c ⋅ Λ ) ⊥ = 1 c ⋅ Λ (c\cdot\Lambda)^\perp = \dfrac{1}{c}\cdot\Lambda(c⋅L )⊥=c1⋅L
- Then S on ( Λ 1 ) = S on ( Λ 1 ) Span(\Lambda_1) = Span(\Lambda_1)Span ( L _1)=Span ( L _1) , thenΛ 1 ⊆ Λ 2 ⟺ Λ 1 ⊥ ⊇ Λ 2 ⊥ \Lambda_1 \subseteq \Lambda_2 \iff \Lambda_1^\perp \supseteq \Lambda_2^\perpL1⊆L2⟺L1⊥⊇L2⊥
- The sparser the grid, the denser its dual, det ( Λ ⊥ ) = 1 / det ( Λ ) \det(\Lambda^\perp) = 1/\det(\Lambda)let ( Λ⊥)=1/let ( Λ )
- Unlike the dual space, the bidual of the lattice is exactly itself, ( Λ ⊥ ) ⊥ = Λ (\Lambda^\perp)^\perp = \Lambda( L⊥)⊥=L
Understanding of the dual case,
- Treated as LAC group: linear space G = R m G=\mathbb R^mG=Rm is a LAC group, latticeΛ \LambdaΛ is a closed subgroup, consider the quotient groupP = G / Λ \mathcal P=G/\LambdaP=G /Λ (parallelepiped of lattice). According to Pontryagin’s duality theorem, the dual latticeΛ ⊥ \Lambda^\perpL⊥ is defined as: satisfiesP ^ ≅ Λ ⊥ \hat{\mathcal P} \cong \Lambda^\perpP^≅LDual group closed subgroup Λ of ⊥ ⊥ ⊂ G ^ \Lambda^\perp \subset \hat GL⊥⊂G^
- Considered as a dual space: discrete subspace Λ ⊂ R m \Lambda \subset \mathbb R^mL⊂Rm is aZ \mathbb ZZ - submodule, all linear mapsy : x ∈ Λ ↦ ⟨ x , y 〉 ∈ Z y:y:x∈L↦⟨x,y⟩∈Z forms the dual spaceH om ( R m , R ) Hom(\mathbb R^m,\mathbb R)Ho m ( Rm,Discrete Z of R ) \mathbb ZZ - submodule, defined as dual lattice.