3.13. Lecture 13

3.13.1. Highest weights

The following is our first version of the fundamental weight calculation.

Lemma 3.13.1.

Let (ฯ,V)(\rho,V) be a complex-linear representation of ๐”ฐโข๐”ฉ2,โ„‚\mathfrak{sl}_{2,\mathbb{C}}. Let ฮฑ\alpha be a weight of VV and let ๐ฏโˆˆVฮฑ{\bf v}\in V_{\alpha}. Then

Xโข๐ฏโˆˆVฮฑ+2X{\bf v}\in V_{\alpha+2}

and

Yโข๐ฏโˆˆVฮฑโˆ’2.Y{\bf v}\in V_{\alpha-2}.

Thus we have three maps:

H\displaystyle H :VฮฑโŸถVฮฑ\displaystyle:V_{\alpha}\longrightarrow V_{\alpha}
X\displaystyle X :VฮฑโŸถVฮฑ+2\displaystyle:V_{\alpha}\longrightarrow V_{\alpha+2}
Y\displaystyle Y :VฮฑโŸถVฮฑโˆ’2.\displaystyle:V_{\alpha}\longrightarrow V_{\alpha-2}.
Proof.

We have, for ๐ฏโˆˆVฮฑ{\bf v}\in V_{\alpha},

ฯโข(H)โขฯโข(X)โข๐ฏ\displaystyle\rho(H)\rho(X){\bf v} =[ฯโข(H),ฯโข(X)]โขv+ฯโข(X)โขฯโข(H)โข๐ฏ\displaystyle=[\rho(H),\rho(X)]v+\rho(X)\rho(H){\bf v}
=ฯโข([H,X])โข๐ฏ+ฯโข(X)โขฯโข(H)โข๐ฏ\displaystyle=\rho([H,X]){\bf v}+\rho(X)\rho(H){\bf v}
=2โขฯโข(X)โข๐ฏ+ฮฑโขฯโข(X)โข๐ฏ\displaystyle=2\rho(X){\bf v}+\alpha\rho(X){\bf v}
=(ฮฑ+2)โขฯโข(X)โข๐ฏ.\displaystyle=(\alpha+2)\rho(X){\bf v}.

So ฯโข(X)โข๐ฏโˆˆVฮฑ+2\rho(X){\bf v}\in V_{\alpha+2} as required.

The claim about the action of YY is proved similarly. โˆŽ

Definition 3.13.2.

A vector ๐ฏโˆˆVฮฑ{\bf v}\in V_{\alpha} is a highest weight vector if it is a weight vector and if

Xโข๐ฏ=0.X{\bf v}=0.

In this case we call the weight of ๐ฏ{\bf v} a highest weight.

Lemma 3.13.3.

Any finite-dimensional complex linear representation VV of ๐”ฐโข๐”ฉ2,โ„‚\mathfrak{sl}_{2,\mathbb{C}} has a highest weight vector.

Proof.

Indeed, let ฮฑ\alpha be the numerically greatest weight of VV (there must be one, as VV is finite-dimensional) and let ๐ฏ{\bf v} be a weight vector of weight ฮฑ\alpha. Then XโขvXv has weight ฮฑ+2\alpha+2 by the fundamental weight calculation, so must be zero as ฮฑ\alpha was maximal. โˆŽ

Example 3.13.4.

Let V=โ„‚2โŠ—โ„‚2V=\mathbb{C}^{2}\otimes\mathbb{C}^{2}. Then the highest weight vectors are ๐ž1โŠ—๐ž1{\bf e}_{1}\otimes{\bf e}_{1} and ๐ž1โŠ—๐žโˆ’1โˆ’๐žโˆ’1โŠ—๐ž1{\bf e}_{1}\otimes{\bf e}_{-1}-{\bf e}_{-1}\otimes{\bf e}_{1}.

These are easily checked to be highest weight vectors โ€” the first is killed by XX since Xโข๐ž1=0X{\bf e}_{1}=0, the second becomes

๐ž1โŠ—๐ž1โˆ’๐ž1โŠ—๐ž1=0.{\bf e}_{1}\otimes{\bf e}_{1}-{\bf e}_{1}\otimes{\bf e}_{1}=0.

It is left to you to check that there are no further highest weight vectors.

3.13.2. Classification of irreducible representations of ๐”ฐโข๐”ฉ2,โ„‚\mathfrak{sl}_{2,\mathbb{C}}.

The key point here is that a highest weight vector must have non-negative integer weight nn, and it then generates an irreducible representation of dimension n+1n+1 whose isomorphism class is determined by nn and which has a very natural basis of weight vectors.

Let (ฯ,V)(\rho,V) be a complex-linear representation of ๐”ฐโข๐”ฉ2,โ„‚\mathfrak{sl}_{2,\mathbb{C}}.

Lemma 3.13.5.

Suppose that VV has a highest weight vector ๐ฏ{\bf v} of weight nn. Then the subspace WW spanned by the vectors

๐ฏ,Yโข๐ฏ,Y2โข๐ฏ=Yโข(Yโข๐ฏ),โ€ฆ{\bf v},Y{\bf v},Y^{2}{\bf v}=Y(Y{\bf v}),\dots

is an ๐”ฐโข๐”ฉ2,โ„‚\mathfrak{sl}_{2,\mathbb{C}}-invariant subspace of VV.

Moreover, for nโ‰ฅ0n\geq 0, the dimension of WW is n+1n+1, with basis ๐ฏ,Yโข๐ฏ,โ€ฆ,Ynโข๐ฏ{\bf v},Y{\bf v},\ldots,Y^{n}{\bf v}, where Yn+1โข๐ฏ=0Y^{n+1}{\bf v}=0.

Proof.

Let WW be the span of the Ykโข๐ฏY^{k}{\bf v}. Since the Ykโข๐ฏY^{k}{\bf v} are weight vectors, their span is HH-invariant. It is clearly also YY-invariant. So we only need to check the invariance under the XX-action. We claim that for all mโ‰ฅ1m\geq 1,

XโขYmโข๐ฏ=mโข(nโˆ’m+1)โขYmโˆ’1โข๐ฏ.XY^{m}{\bf v}=m(n-m+1)Y^{m-1}{\bf v}. (3.13.6)

The proof is by induction. The case m=1m=1 is:

XโขYโข๐ฏ=([X,Y]+YโขX)โข๐ฏ=Hโข๐ฏ+Yโข(Xโข๐ฏ)=Hโข๐ฏ=nโข๐ฏ.XY{\bf v}=([X,Y]+YX){\bf v}=H{\bf v}+Y(X{\bf v})=H{\bf v}=n{\bf v}.

If the formula holds for mm, then

XโขYm+1โข๐ฏ\displaystyle XY^{m+1}{\bf v} =([X,Y]+YโขX)โขYmโข๐ฏ\displaystyle=([X,Y]+YX)Y^{m}{\bf v}
=HโขYmโข๐ฏ+Yโข(XโขYmโข๐ฏ)\displaystyle=HY^{m}{\bf v}+Y(XY^{m}{\bf v})
=(nโˆ’2โขm)โขYmโข๐ฏ+mโข(nโˆ’m+1)โขYmโข๐ฏ\displaystyle=(n-2m)Y^{m}{\bf v}+m(n-m+1)Y^{m}{\bf v} (induction hypothesis)
=(m+1)โข(nโˆ’m)โขYmโข๐ฏ\displaystyle=(m+1)(n-m)Y^{m}{\bf v}

as required.

If nn is not a nonnegative integer, then mโข(nโˆ’m+1)โ‰ 0m(n-m+1)\neq 0 for all mm. So

Ymโˆ’1โข๐ฏโ‰ 0โŸนXโขYmโข๐ฏโ‰ 0โŸนYmโข๐ฏโ‰ 0,Y^{m-1}{\bf v}\neq 0\implies XY^{m}{\bf v}\neq 0\implies Y^{m}{\bf v}\neq 0,

whence Ymโข๐ฏโ‰ 0Y^{m}{\bf v}\neq 0 for all mm. As these are weight vectors with distinct weights, they are linearly independent and so span an infinite dimensional subspace.

If Yiโข๐ฏ=0Y^{i}{\bf v}=0 for some 0<iโ‰คn0<i\leq n, then

0=XโขYiโข๐ฏ=iโข(nโˆ’i+1)โขYiโˆ’1โข๐ฏ0=XY^{i}{\bf v}=i(n-i+1)Y^{i-1}{\bf v}

and so Yiโˆ’1โข๐ฏ=0Y^{i-1}{\bf v}=0. Repeating gives that ๐ฏ=Y0โข๐ฏ=0{\bf v}=Y^{0}{\bf v}=0, a contradiction.

Now,

XโขYn+1โข๐ฏ=(n+1)โข(nโˆ’n)โข๐ฏ=0XY^{n+1}{\bf v}=(n+1)(n-n){\bf v}=0

and so Yn+1โข๐ฏY^{n+1}{\bf v} is either zero or a highest weight vector of weight โˆ’(n+2)<0-(n+2)<0. We have already seen that the second possibility cannot happen, so Yn+1โข๐ฏ=0Y^{n+1}{\bf v}=0.

Thus WW is spanned by the (non-zero) weight vectors ๐ฏ,Yโข๐ฏ,โ€ฆ,Ynโข๐ฏ{\bf v},Y{\bf v},\ldots,Y^{n}{\bf v} with distinct weights, which are therefore linearly independent and so a basis for WW. โˆŽ

Remark 3.13.7.

If we didnโ€™t assume that VV was finite-dimensional, then the first part of the previous lemma would still be true.

Corollary 3.13.8.

In the situation of the previous lemma, WW is irreducible.

Proof.

Suppose that Wโ€ฒโІWW^{\prime}\subseteq W is a non-zero subrepresentation. Then it has a highest weight vector, which must be (proportional to) YiโขvY^{i}v for some 0โ‰คiโ‰คn0\leq i\leq n. But then

XโขYiโขv=iโข(nโˆ’i+1)โขYiโˆ’1โขvโ‰ 0XY^{i}v=i(n-i+1)Y^{i-1}v\neq 0

if i>0i>0 and so i=0i=0, meaning that ๐ฏโˆˆWโ€ฒ{\bf v}\in W^{\prime}. But then YiโขvโˆˆWโ€ฒY^{i}v\in W^{\prime} for all ii, so Wโ€ฒ=WW^{\prime}=W as required. โˆŽ

The weights of WW are illustrated in Figure 3.13.2.

[Uncaptioned image]
Theorem 3.13.9.

Suppose VV is an irreducible finite-dimensional complex-linear representation of ๐”ฐโข๐”ฉ2,โ„‚\mathfrak{sl}_{2,\mathbb{C}}. Then:

  1. (i)
    โ€‹

    There is a unique (up to scalar) highest weight vector, with highest weight nโ‰ฅ0n\geq 0.

  2. (ii)
    โ€‹

    The weights of VV are n,nโˆ’2,โ€ฆ,2โˆ’n,โˆ’nn,n-2,\ldots,2-n,-n.

  3. (iii)
    โ€‹

    All weight spaces VฮฑV_{\alpha} are one-dimensional (we say that VV is โ€˜multiplicity freeโ€™).

  4. (iv)
    โ€‹

    The dimension of VV is n+1n+1.

  5. (v)
    โ€‹

    For every nโ‰ฅ0n\geq 0, the unique irreducible complex-linear representation of ๐”ฐโข๐”ฉ2,โ„‚\mathfrak{sl}_{2,\mathbb{C}} with highest weight nn is Symnโก(โ„‚2)\operatorname{Sym}^{n}(\mathbb{C}^{2}) (up to isomorphism).

Proof.

Let VV be an irreducible finite-dimensional complex-linear representation of ๐”ฐโข๐”ฉ2,โ„‚\mathfrak{sl}_{2,\mathbb{C}}. Let ๐ฏโˆˆV{\bf v}\in V be a highest weight vector. By Lemma 3.13.5 its weight is a nonnegative integer nn, and by Corollary 3.13.8 the vectors ๐ฏ,Yโข๐ฏ,โ€ฆ,Ynโข๐ฏ{\bf v},Y{\bf v},\ldots,Y^{n}{\bf v} span an irreducible subrepresentation of VV of dimension n+1n+1, which must therefore be the whole of VV. The claims (i) to (iv) follow immediately.

Moreover, the actions of XX, YY and HH on this basis are given by explicit matrices depending only on nn, so the isomorphism class of VV is determined by nn. It remains only to show that a representation with highest weight nn exists for all nโ‰ฅ0n\geq 0. Consider Symnโก(โ„‚2)\operatorname{Sym}^{n}(\mathbb{C}^{2}). The vector ๐ž1n{\bf e}_{1}^{n} is a highest weight vector of weight nn, so we are done. In fact, in this case Yaโข๐ž1nY^{a}{\bf e}_{1}^{n} is proportional to ๐ž1nโˆ’aโข๐žโˆ’1a{\bf e}_{1}^{n-a}{\bf e}_{-1}^{a}, so the irreducible representation generated by ๐ž1n{\bf e}_{1}^{n} is in fact the whole of Symnโก(โ„‚2)\operatorname{Sym}^{n}(\mathbb{C}^{2}). โˆŽ

In fact, if (ฯ,V)(\rho,V) is the irreducible representation with highest weight nn, highest weight vector ๐ฏ{\bf v}, then the matrices of HH, YY, and XX with respect to the basis

{๐ฏ,Yโข(๐ฏ),โ€ฆ,Ynโข(๐ฏ)}\{{\bf v},Y({\bf v}),\ldots,Y^{n}({\bf v})\}

are respectively

ฯโข(H)\displaystyle\rho(H) =(nnโˆ’2โ‹ฑโˆ’n),\displaystyle=\begin{pmatrix}n&&&\\ &n-2&&\\ &&\ddots&\\ &&&-n\end{pmatrix},
ฯโข(Y)\displaystyle\rho(Y) =(010โ‹ฑโ‹ฑ10),\displaystyle=\begin{pmatrix}0&&&\\ 1&0&&\\ &\ddots&\ddots&\\ &&1&0\end{pmatrix},
and
ฯโข(X)\displaystyle\rho(X) =(0n02โข(nโˆ’1)03โข(nโˆ’2)โ‹ฑโ‹ฑโ‹ฑn0).\displaystyle=\begin{pmatrix}0&n&&&&\\ &0&2(n-1)&&&\\ &&0&3(n-2)&&\\ &&&\ddots&\ddots&\\ &&&&\ddots&n\\ &&&&&0\end{pmatrix}.
Theorem 3.13.10.

Every finite-dimensional irreducible complex-linear representation of SL2โก(โ„‚)\operatorname{SL}_{2}(\mathbb{C}) or ๐”ฐโข๐”ฉ2,โ„‚\mathfrak{sl}_{2,\mathbb{C}} is isomorphic to Symnโก(โ„‚2)\operatorname{Sym}^{n}(\mathbb{C}^{2}), the symmetric product of the standard representation.

Proof.

We already proved this for ๐”ฐโข๐”ฉ2,โ„‚\mathfrak{sl}_{2,\mathbb{C}}, Theoremย 3.13.9(v). If VV is a representation of SL2โก(โ„‚)\operatorname{SL}_{2}(\mathbb{C}), then its derivative will be isomorphic to Symnโก(โ„‚2)\operatorname{Sym}^{n}(\mathbb{C}^{2}). Since SL2โก(โ„‚)\operatorname{SL}_{2}(\mathbb{C}) is connected, this implies that Vโ‰…Symnโก(โ„‚2)V\cong\operatorname{Sym}^{n}(\mathbb{C}^{2}). โˆŽ

3.13.3. Exercises

.

Problemย 31. Let VV, WW be representations of ๐”ฐโข๐”ฉ2,โ„‚\mathfrak{sl}_{2,\mathbb{C}}. Let ๐ฏ{\bf v} and ๐ฐ{\bf w} be two weight vectors of VV and WW respectively with respective weights ฮฑ\alpha and ฮฒ\beta. Show that

๐ฏโŠ—๐ฐโˆˆVโŠ—W{\bf v}\otimes{\bf w}\in V\otimes W

is a weight vector with weight ฮฑ+ฮฒ\alpha+\beta, and that if ๐ฏ{\bf v} and ๐ฐ{\bf w} are highest weight vectors then so is ๐ฏโŠ—๐ฐ{\bf v}\otimes{\bf w}.

Problemย 32. If ฮปโˆˆโ„‚\lambda\in\mathbb{C}, show that there is a (possibly infinite dimensional!) representation of ๐”ฐโข๐”ฉ2,โ„‚\mathfrak{sl}_{2,\mathbb{C}} with highest weight ฮป\lambda.

Problemย 33. Consider V=Symnโก(โ„‚2)V=\operatorname{Sym}^{n}(\mathbb{C}^{2}), the irreducible representation of highest weight nn of ๐”ฐโข๐”ฉ2,โ„‚\mathfrak{sl}_{2,\mathbb{C}}. Decompose the following representations into irreducibles, and find highest weight vectors for the irreducible constituents:

  1. (a)
    โ€‹

    Sym2โก(Sym2โก(โ„‚2))\operatorname{Sym}^{2}(\operatorname{Sym}^{2}(\mathbb{C}^{2}));

  2. (b)
    โ€‹

    โ‹€2(Sym2โก(โ„‚2))\bigwedge^{2}\left(\operatorname{Sym}^{2}(\mathbb{C}^{2})\right);

  3. (c)
    โ€‹

    Sym3โก(โ„‚2)โŠ—Sym2โก(โ„‚2)\operatorname{Sym}^{3}(\mathbb{C}^{2})\otimes\operatorname{Sym}^{2}(\mathbb{C}% ^{2});

  4. (d)
    โ€‹

    Sym3โก(Sym2โก(โ„‚2))\operatorname{Sym}^{3}\left(\operatorname{Sym}^{2}(\mathbb{C}^{2})\right).

For the third example, find bases for the irreducible subrepresentations.

Problemย 34. Let (ฯ€,V)(\pi,V) be a finite-dimensional representation of ๐”ฐโข๐”ฉ2,โ„‚\mathfrak{sl}_{2,\mathbb{C}}. Consider the Casimir element11 1 Conventions differ; it might be more usual to call 1+2โข๐’ž1+2\mathcal{C} the Casimir.

๐’ž=ฯ€โข(X)โขฯ€โข(Y)+ฯ€โข(Y)โขฯ€โข(X)+12โขฯ€โข(H)2.\mathcal{C}=\pi(X)\pi(Y)+\pi(Y)\pi(X)+\frac{1}{2}\pi(H)^{2}.
  1. (a)
    โ€‹

    Show ๐’ž\mathcal{C} commutes with the action of ๐”ฐโข๐”ฉ2,โ„‚\mathfrak{sl}_{2,\mathbb{C}}. Conclude that if VV is irreducible then ๐’ž\mathcal{C} acts as a scalar.

  2. (b)
    โ€‹

    What is the scalar for V=Symnโก(โ„‚2)V=\operatorname{Sym}^{n}(\mathbb{C}^{2}), the irreducible representation of highest weight nn?

  3. (c)
    โ€‹

    Compute the action of ๐’ž\mathcal{C} on the space VV of polynomial functions ฯ•\phi on โ„‚2\mathbb{C}^{2}, with action the derivative of (gโขฯ•)โข(๐ฏ)=ฯ•โข(gโˆ’1โขv)(g\phi)({\bf v})=\phi(g^{-1}v) (see problem 2.11.2).