Previous Up Next
Creative Commons License
This work is licensed under a Creative Commons Attribution-NonCommercial-ShareAlike 4.0 International License.

Lecture 2 The Heisenberg Group

The relations, which define the Heisenberg group or its Lie algebra, are of a fundamental nature and appeared in very different areas. For example, the basic operators of differentiation and multiplication by an independent variable in analysis satisfy to the same commutation relations as observables of momentum and coordinate in quantum mechanics.

It is very easy to oversee those common structures. Roger Howe said in [138]:

An investigator might be able to get what he wanted out of a situation while overlooking the extra structure imposed by the Heisenberg group, structure which might enable him to get much more.

We shall start from the general properties the Heisenberg group and its representations. Many important applications will follow.

Remark 1 It is worth to mention that the Heisenberg group is also known as the Weyl (or Heisenberg–Weyl) group in physical literature. It is another illustration of its perception as an extraneous object: physicists call it by the name of a mathematician, and mathematicians by the name of a physicists.

To add more confusion the Lie algebra of the Heisenberg group is called Weyl algebra. However, the commutator relations [Q,P]=I in the Weyl algebra are called Heisenberg commutator relations, however the commutation relation sm=q ms in the representation the Heisenberg group are known as the Weyl commutation relation. The most obvious simplification would be to call every above object the Heisenberg–Weyl. However we will use the most common names in this work.

2.1 The Symplectic Form and the Heisenberg group

Let n≥ 1 be an integer. For two real n-vectors x, y∈ℝn, we write xy for their inner product:

xy=x1y1+x2y2+⋯+xnyn,   where x=(x1,x2,…,xn), y=(y1,y2,…,yn). (1)

Similarly for complex vectors z, w∈ℂn, we define:

zw=z1w1+z2w2+⋯+znwn,   where z=(z1,z2,…,zn), w=(w1,w2,…,wn). (2)

The following notion is the central for Hamiltonian formulation of classical mechanics [11]*§ 37.

Definition 2 The symplectic form ω on 2n is a function of two vectors such that:
ω(x,y;x′,y′)=xy′−xy,     where  (x,y), (x′,y′)∈ℝ2n. (3)
Exercise 3 Check the following properties:
  1. ω is anti-symmetric ω(x,y;x′,y′)=− ω(x′,y′;x,y).
  2. ω is bilinear:
          ω( x, y;α x′,α y′)=      α ω(x,y;x′,y′),
          ω( x, y;  x′+x″, y′+x″)=      α ω(x,y;x′,y′)+ω(x,y;x″,y″).
  3. Let z=x+i y and w=x′+i y then ω can be expressed through the complex inner product (2) as ω(x,y;x′,y′)=−ℑ (zw).
  4. The symplectic form on 2 is equal to det (
          xx
    yy
    ). Consequently it vanishes if and only if (x,y) and (x′,y′) are collinear.
  5. Let ASL2(ℝ) be a real 2× 2 matrix with the unit determinant. Define:


            x
    y


    =A


            x
    y


      and  


            x′′
    y′′


    =A


            x
    y


    . (4)
    Then, ω(x′,y′;x′′,y′′)=ω(x,y;x′,y′). Moreover, the symplectic group [2]—the set of all linear transformations of 2 preserving ω—coincides with SL2(ℝ).

Now we define the main object of our consideration.

Definition 4 An element of the n-dimensional Heisenberg group ℍn [104, 139] is (s,x,y)∈ℝ2n+1, where s∈ℝ and x, y∈ ℝn. The group law on n is given as follows:
(s,x,y)·(s′,x′,y′)=(s+s′+
1
2
ω(x,y;x′,y′),x+x′,y+y′),  (5)
where ω the symplectic form.

For the sake of simplicity, we will work with the one-dimensional Heisenberg group ℍ1 on several occasions. This shall be mainly in the cases involving the symplectic group, since we want to the stay the basic case [2]∼ SL2(ℝ). However, consideration of the general case of ℍn is similar in most respects.

Exercise 5 For the Heisenberg group n, check that:
  1. The unit is (0,0,0) and the inverse (s,x,y)−1=(−s,−x,−y).
  2. It is a non-commutative Lie group. Hint: Use the properties of the symplectic form from the Exercise 3. For example, the associativity follows from the linearity of ω.
  3. It has the centre
    Z={(s,0,0)∈ ℍn,  s ∈ ℝ}. (6)

The group law on ℍn can be expressed in several equivalent forms.

Exercise 6
  1. Introduce complexified coordinates (s,z) on 1 with z=x+i y. Then the group law can be written as:
           (s,z)·(s′,z′)=(s+s′+
    1
    2
    ℑ(zz),z+z′). 
  2. Show that the set 3 with the group law
    (s,x,y)·(s′,x′,y′)=(s+s′+xy′,x+x′,y+y′) (7)
    is isomorphic to the Heisenberg group 1. It is called polarised Heisenberg group [104]*§ 1.2. Hint: Use the explicit form of the homomorphism (s,x,y)↦ (s+1/2xy,x,y).
  3. Define the map φ: ℍ1M3(ℝ) by
    φ(s,x,y)=






              1x
    s+
    1
    2
    xy
              01y
              001






    . (8)
    This is a group homomorphism from 1 to the group of 3× 3 matrices with the unit determinant and the matrix multiplication as the group operation. Write also a group homomorphism from the polarised Heisenberg group to M3(ℝ).
  4. Expand the above items from this Exercise to n.

2.2 Lie algebra of the Heisenberg group

The Lie algebra of the Heisenberg group h1 is also called Weyl algebra. From the general theory we know, that h1 is a three-dimensional real vector space, thus, it can be identified as a set with the group ℍ1∼ℝ3 itself.

There are several standard possibilities to realise h1, cf. Sect. 2.3. Firstly, we link h1 with one-parameter subgroups as in Sect. 2.3.1.

Exercise 7
  1. Show that 3× 3 matrices from (8) are created by the following exponential map:
    exp



            0xs
            00y
            000



    =






            1x
    s+
    1
    2
    xy
            01y
            001






    . (9)
    Thus h1 isomorphic to the vector space of matrices in the left-hand side. We can define the explicit identification exp: h1 → ℍ1 by (9), which is also known as the exponential coordinates on 1.
  2. Define the basis of h1:
    S=



              001
              000
              000



    ,  X=



              010
              000
              000



    ,  Y=



              000
              001
              000



    . (10)
    Write the one-parameter subgroups of 1 generated by S, X and Y.

Another possibility is a description of h1 as the collection of invariant vector fields, see Sect. 2.3.2.

Exercise 8
  1. Check that the following vector fields on 1 are left (right) invariant:
      Sl(r)=±∂s,   Xl(r)=±∂x
    1
    2
    ys,    Yl(r)=±∂y+
    1
    2
    xs. (11)
    Show also that they are linearly independent and, thus, are bases the Lie algebra h1 in two different realisations.
  2. Calculate one-parameter groups of right (left) shifts on 1 generated by these vector fields.

The principal operation on a Lie algebra, besides the linear structure, is the commutator [A,B]=ABBA, see Sect. 2.3.3. In the above exercises we can define the commutator for matrices and vector fields through the corresponding algebraic operations in these algebras.

Exercise 9
  1. Check that bases from (10) and (10) satisfy the Heisenberg commutator relation
    [Xl(r),Yl(r)]=Sl(r)(12)
    and all other commutators vanishing. More generally:
    [A,A′]=ω(x,y;x′,y′) S,   where A(′)=s(′)S+x(′)X+y(′)Y, (13)
    and ω is the symplectic form.
  2. Show that any second (and, thus, any higher) commutator [[A,B],C] on h1 vanishes. This property can be stated as “the Heisenberg group is a step 2 nilpotent Lie group”.
  3. Check the formula
    exp(A)exp(B)=exp(A+B+
    1
    2
    [A,B]),    where  A,Bh1. (14)
    The formula is also true for any step 2 nilpotent Lie group and is a particular case of the Baker–Campbell–Hausdorff formula. Hint: In the case of 1 you can use the explicit form of the exponential map (9).
  4. Define the vector space decomposition
    h1=V0⊕ V1,     such that  V0=[V1,V1]. (15)

Consequently, we can start from definition of the Lie algebra hn through the commutation relations (12). Thereafter, ℍn and the group law (5) can be derived from the exponentiation of hn.

We note that any element Ah1 defines an adjoint map ad(A): B↦ [A,B] on h1.

Exercise 10 Write matrices corresponding to transformations ad(S), ad(X), ad(Y) of h1 in the basis S, X, Y.

2.3 Automorphisms of the Heisenberg group

Erlangen programme suggest investigate invariants under group action. This recipe can be applied recursively to groups themselves. Transformations of a group which preserve its structure are called group automorphisms.

Exercise 11 Check that the following are automorphisms of 1:
  1. Inner automorphisms or conjugation with (s,x,y)∈ℍ1:
         
          (s′,x′,y′)↦(s,x,y)· (s′,x′,y′) · (s,x,y)−1= (s′+ω(x,y;x′,y′),x′,y′)   
           =(s′+xy′−xy,x′,y′). (16)
  2. Symplectic maps (s,x,y)=(s,x′,y′), where (
            x
    y
    )=A (
            x
    y
    ) with A from the symplectic group [2] ∼SL2(ℝ), see Exercise 5.
  3. Dilations: (s,x,y)↦(r2s,rx,ry) for a positive real r.
  4. Inversion: (s,x,y)↦(−s,y,x).
The last three types of transformations are outer automorphisms.

In fact we listed all ingredients of the automorphism group.

Exercise 12 Show that
  1. Automorphism groups of 1 and h1 coincide as groups of maps of 3 onto itself. Hint: Use the exponent map and the relation (14). The crucial step is a demonstration that any automorphism of 1 is a linear map of 3. See details in [104]*Ch. 1, (1.21).
  2. All transforms from Exercise 11 viewed as automorphisms of h1 preserve the decomposition (15).
  3. Every automorphism of 1 can be written uniquely as composition of a symplectic map, an inner automorphism, a dilation and a power (mod 2) of the inversion from Exercise 11. Hint: Any automorphism is a linear map (by the previous item) of 3 which maps the centre Z to itself. Thus it shall have the form (s,x,y)↦ (cs+ax+by, T(x,y)), where a, b and c are real and T is a linear map of 2, see [104]*Ch. 1, (1.22).

The symplectic automorphisms from Exercise 2 can be characterised as the group of outer automorphisms of ℍ1, which trivially acts on the centre of ℍ1. It is the group of symmetries of the symplectic form ω in (5) [104]*Thm. 1.22 [138]*p. 830. The symplectic group is isomorphic to SL2(ℝ) considered in the first half of this work, see Exercise 5. For future use we will need Sp′(2) which is the double cover of [2].

We can build the semidirect product G=ℍ1Sp′(2) with the standard group law for semidirect products:

(h,g)*(h′,g′)=(h*g(h′),g*g′),    where   h,h′∈ℍ1,   g,g′∈Sp′(2), (17)

and the stars denote the respective group operations while the action g(h′) is defined as the composition of the projection map Sp′(2)→ Sp(2) and the action (4). This group is sometimes called the Schrödinger group and it is known as the maximal kinematical invariance group of both the free Schrödinger equation and the quantum harmonic oscillator [263]. This group is of interest not only in quantum mechanics but also in optics [323, 322].

Consider the Lie algebra sp2 of the group [2]. We again use the basis A, B, Z (13) with commutators (13). Vectors Z, BZ/2 and B are generators of the one-parameter subgroups K, N′ and (35) respectively. Furthermore we can consider the basis {S, X, Y, A, B, Z} of the Lie algebra g of the Schrödinger group G=ℍ1Sp′(2). All non-zero commutators besides those already listed in (12) and (13) are:

     
    [A,X]
=
1
2
X,  
[B,X]
=−
1
2
Y,  
[Z,X]=Y;    (18)
  [A,Y]
=−
1
2
Y,  
[B,Y]
=−
1
2
X,  
[Z,Y]=−X.     (19)

2.4 Subgroups of ℍn and Homogeneous Spaces

We want to classify up to certain equivalences all possible ℍ1–homogeneous spaces. According to Sect. 2.2.2 we will look for continuous subgroups of ℍ1.

One-dimensional continuous subgroups of ℍ1 can be classified up to group automorphism. Two one-dimensional subgroups of ℍ1 are the centre Z (6) and

Hx={(0,t,0)∈ ℍn,  t ∈ ℝ}. (20)
Exercise 13 Show that
  1. There is no an automorphism of 1 which maps Z to Hx.
  2. For any one-parameter continuous subgroup H of 1 there is an automorphism of 1 which maps H either to Z or Hx.

Next, for a subgroup H we wish to describe the respective homogeneous space X=ℍ1/H and actions of ℍ1 on X. See, Section 2.2.2 for the background general theory and definitions of maps p: ℍ1X and s: X → ℍ1.

Exercise 14 Check that:
  1. Both homogeneous spaces 1/Z and 1/Hx are parametrised by 2;
  2. The 1-action on 1/Z is:
    (s,x,y): (x′,y′) ↦ (x+x′,y+y′). (21)
    Hint: Use the following maps: p: (s′,x′,y′)↦ (x′,y′), s:(x′,y′)↦(0,x′,y′).
  3. The 1-action on 1/Hx is:
     (s,x,y): (s′,y′) ↦ (s+s′+
    1
    2
    xy′,y+y′). (22)
    Hint: Use the following maps: p: (s′,x′,y′)↦ x, s: x′↦(0,x′,0).

The classification of two-dimensional subgroups is as follows:

Exercise 15 Show that
  1. For any two-dimensional continuous subgroup of 1 there is an automorphism of 1 which maps the subgroup to
    Hx={(s,0,y)∈ℍ1,  s,y∈ℝ}. (23)
  2. The homogeneous space 1/Hx is parametrised by .
  3. 1-action on 1/Hx is
    (s,x,y): x′ ↦ x+x(24)
    Hint: Use the maps p: (s′,x′,y′)↦ x and s: x′ ↦ (0,x′,0).

Actions (21) and (24) are Euclidean shifts and much more simple than the Möbius action of the group SL2(ℝ) (1). Nevertheless, the associated representations of the Heisenberg group will be far from trivial.

site search by freefind advanced

Last modified: October 28, 2024.
Previous Up Next