Let $V$ be a $m+1$-dim $K$-vector space with char$K=0$. Let $(v_0,v_1,\dots,v_m)$ be a basis of $V(m)$. Now suppose I construct a representation of $sl(2,K)$ on this representation.

How do I show that this representation is irreducible?

A rep is called irreducible if the only representation subspace $U\subset V$ left invariant by the action of my algebra is $V$ itself and the zero subspace. But in general how does one show that this is indeed true for a given example?

For example, I have a representation. Now I can create a proper subspace $U$ by removing at least one basis vector from $V$, say $v_i$, so $U$ is spanned by $$(v_0,v_1,\dots,v_{i-1},v_{i+1},\dots, v_m).$$ Now I can show that this subspace is not closed under the action of my algebra. Then I can continue and remove all possible combinations of the basis vectors and find that none are invariant until I get to the zero space.

Is this a valid approach? Is there a more direct way?

EDIT: Here is a specific example. Let $X,Y,H$ be the basis of $sl(2,K)$ with $$X=\left( \begin{array}{cc} 0 & 1 \\ 0 & 0 \end{array}\right), \quad Y=\left( \begin{array}{cc} 0 & 0 \\ 1 & 0 \end{array}\right), \quad H=\left( \begin{array}{cc} 1 & 0 \\ 0 & -1 \end{array}\right).$$

I define $v_{-1}=0=v_{m+1}$ and $v_i=\frac{1}{i!}Y^iv_0$, where $v_0$ is a maximal vector. $\lambda$ is a highest weight of $H$. I represent the action of $X,Y,H$ on $V$ by $$X.v_i=(\lambda-i+1)v_{i-1}$$ $$Y.v_i=(i+1)v_{i+1}$$ $$H.v_i=(\lambda-2i)v_i$$ Now I can create proper subspaces by removing basis vectors from $V$. Remove $v_i$ and define $U$ to be the subspace spanned by $$(v_0,\dots,v_{i-1},v_{i+1},\dots,v_m)$$ Then it is clear that by acting with $X$ on $v_{i+1}$ I will obtain $v_i$ so this subspace is not closed. I can repeat this and remove all possible combinations of basis vectors from $V$, each time defining a new subspace. It is clear that unless I remove none (ie. start with $V$) or all (i.e start with zero space) then no subspace will be invariant, as I can always act with $X$ or $Y$ a sufficient number of times to get a vector not in this subspace.

Hence the representation is irreducible.

Is this correct?

  • $\begingroup$ I think the action of $X$ and $Y$ should shift the index $i$ of $v_i$, in other words have $v_{i+1}$ or $v_{i-1}$ in the right hand side. $\endgroup$ – Marc van Leeuwen Jun 14 '16 at 20:45
  • $\begingroup$ Yes absolutely, a typo on my behalf. $\endgroup$ – qftey Jun 14 '16 at 21:39

In the example a key fact is that the vectors $v_i$ already form a basis of diagonalisation for the action of $H$, and that all the eigenvalues are distinct. This implies that the only subspaces invariant for $H$ are those spanned by a subset of the vectors$~v_i$. If such a subspace contains some $v_i$, then in order to be stable for $X$ and $Y$ it must (with the proper formulas) also contain $v_{i-1}$ and $v_{i+1}$ if these exist. Then the only subsets of the $v_i$ that span a subspace invariant under $H,X,Y$ are the empty set and the set of all the $v_i$. This shows the representation is irreducible.


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.