The first fundamental theorem of invariant theory for the quantum queer superalgebra

Recall that \mathsf{A}_{r,n} is generated by 1, t_{i,a} with i\in I_{r|r}, a\in I_{n|n} and t_{-i,-a}=t_{ia}, we know that \mathsf{A}_{r,n} is generated by 1, t_{ia} with i=1,\ldots,r and a\in I_{n|n}.

We next show that the relation (?) is isomorphic to (3.2) provided that t_{-i,-a}=t_{ia}. In fact, the relation (?) is equivalent to

for i,j\in I_{r|r}, a,b\in I_{n|n}. When i,j>0, it is reduced to

which is the expansion of (3.2).

Conversely, it suffices to show the relation (?) also implies (?). We consider the following four cases separately:

Case1: i>0, j>0. In this situation, the relations (?) and (?) are same.

Case2: i<0, j>0. We deduce from (?) that

Since t_{-i,-a}=t_{ia}, we obtain that

which coincides with (?) in the situation of i<0 and j>0.

Case 3: i>0, j<0. We deduce from (?) that

Then t_{-j,-a}=t_{ja} implies that

Hence,

which yields with (?) in the situation of i>0 and j<0.

Case 4: i<0, j<0. We deduce from (?) that

It follows from t_{-j,-a}=t_{j,a} that

Hence,

which yields with (?) in the situation of i<0 and j<0.

Now, we have shown that the relations (?) and (?) are equivalent to (3.2). Hence, the generators 1, t_{ia}, i=1,\ldots, r and a\in I_{n|n} and the relations (3.2) give an alternative presentation of the superalgebra \mathsf{A}_{r,n}.

Since \mathsf{A}_{r,n} is generated by 1 and t_{ia} with i=1,\ldots, r, a\in I_{n|n}, while \bar{\mathsf{A}}_{s,n} is generated by 1 and \bar{t}_{\alpha, b} with \alpha\in I_{s|s}, b\in I_{n|n}, the equations (3.3) and (3.3) determine \Upsilon(\bar{g}\otimes f) when f and \bar{g} are one of those generators of \mathsf{A}_{r,n} and \bar{\mathsf{A}}_{s,n}, respectively. The definition is then extended to the whole \bar{\mathsf{A}}_{s,n}\otimes\mathsf{A}_{r,n} according to diagrams (3.1) and (3.1). It suffices to show \Upsilon is well-defined.

By Lemma 3.2, the generators t_{ia}, i=1,\ldots, r, a\in I_{n|n} of \mathsf{A}_{r,n} satisfy

We calculate using the diagram (3.1) that

Since the matrix S satisfies the quantum Yang-Baxter equation, we conclude that

The generators \bar{t}_{\alpha b}’s of \bar{\mathsf{A}}_{s,n} satisfy

We similarly verify that

Moreover, \bar{t}_{-\alpha,-b}=(-1)^{|\alpha|+|b|}\bar{t}_{\alpha, b} for \alpha\in I_{s|s} and b\in I_{n|n}. It can be written in the matrix form \bar{T}=(J_n\otimes 1)\bar{T}(J_s\otimes1), where

Since (J_n\otimes 1)S=S(J_n\otimes1), we also verify that

which yields that

Now, the well-definedness of \Upsilon follows from (?), (?) and (?).

For u\in\mathrm{U}_q(\mathfrak{q}_n) and f\in\mathsf{A}_{r,n}, if \Delta(u)=\sum\limits_{(u)}u_{(1)}\otimes u_{(2)}, then

which yields

for f\in\mathsf{A}_{r,n}. A similar calculation also shows

for \bar{g}\in\bar{\mathsf{A}}_{s,n}.

In order to show that

for a,b\in I_{n|n}, \alpha\in I_{s|s} and k=1,\ldots, r, it suffices to consider the case where u=L_{ij} with i\leqslant j. Then the verification can be done using matrix computations as follows:

Using the quantum Yang-Baxter equation for S again, we obtain

Now, \mathsf{A}_{r,n} is generated by 1, t_{ka}, k=1,\ldots, r and a\in I_{n|n}, while \bar{\mathsf{A}}_{s,n} is generated by 1, \bar{t}_{\alpha,b}, \alpha\in I_{s|s} and b\in I_{n|n}. Using the commutative diagrams (3.1) and (3.1), we inductively deduce that

for all f\in\mathsf{A}_{r,n} and \bar{g}\in\bar{\mathsf{A}}_{s,n}. This shows \Upsilon is a homomorphism of \mathrm{U}_q(\mathfrak{q}_n)-supermodules.

It suffices to verify

We first check it on generators.

for f\in\mathsf{A}_{r,n} and \bar{g}\in\bar{\mathsf{A}}_{s,n}. We also have

which implies

for k=1,\ldots,r, \alpha\in I_{s|s} and a,b\in I_{n|n}. Using the commutative diagrams (3.1) and (3.1), we deduce the following equation by induction

for f\in\mathsf{A}_{r,n} and \bar{g}\in\bar{\mathsf{A}}_{s,n}. This completes the proof.

We set

Then X^{1[2]}=T_+^{1[2]}\bar{T}^{1[2']}. Since \Delta(L^{[2]3})=L^{2[3]}\otimes L^{[2']3}, the action of generating matrix L on X under \Phi is

Since the counit \varepsilon of \mathrm{U}_q(\mathfrak{q}_n) satisfies \varepsilon(L)=I, we obtain x_{i\alpha}\in\left(\mathcal{O}_{r,s}\right)^{\mathrm{U}_q(\mathfrak{q}_n)}.

Note that X^{1[3]}=T_+^{1[3]}\bar{T}^{1[3']}. We have already known from Lemma 3.2 and Theorem 3.3 that

in \mathcal{O}_{r,s}. Hence, we deduce that

Similarly,

This completes the proof.

Since X^{2[3]}=T_+^{2[3]}\bar{T}^{2[3']}, it follows from (4.2) that

By (4.2), we have

This implies (4.3).

Since the superalgebra \mathsf{A}_{r,n} (resp. \bar{\mathsf{A}}_{s,n}) is a \mathrm{U}_q(\mathfrak{q}_r)\otimes\mathrm{U}_q(\mathfrak{q}_n)-supermodule (resp. a \mathrm{U}_q(\mathfrak{q}_s)\otimes\mathrm{U}_q(\mathfrak{q}_n)-supermodule) and has the multiplicity free decomposition (?) (resp. (?)), the superalgebra \mathcal{O}_{r,s}=\mathsf{A}_{r,n}\otimes\bar{\mathsf{A}}_{s,n} is decomposed into the direct sum

as a \mathrm{U}_q(\mathfrak{q}_r)\otimes\mathrm{U}_q(\mathfrak{q}_n)\otimes\mathrm{U}_q(\mathfrak{q}_s)\otimes\mathrm{U}_q(\mathfrak{q}_n)-supermodule.

We denote

Then Schur’s Lemma for queer superalgebra states that

which implies that

Recall from [GJKKK15] that \mathsf{L}_r^*(\lambda)\otimes \mathsf{L}_n(\lambda) (resp. \mathsf{L}_s(\mu)\otimes \mathsf{L}_n^*(\mu)) is the direct sum of 2^{\delta(\lambda)} (resp. 2^{\delta(\mu)}) copies of \mathsf{L}_r^*(\lambda)\circledast \mathsf{L}_n(\lambda) (resp. \mathsf{L}_s(\mu)\circledast \mathsf{L}_n^*(\mu)). Hence,

which yields

Hence,

This proves the decomposition (4.4).

For f\in\mathsf{A}_{r,s} and u\in\mathrm{U}_q(\mathfrak{q}_r),

Then we deduce that

It shows \tilde{\Delta}_{r,s}\circ\Psi_u(f)=(\Psi_u\otimes1)\circ\tilde{\Delta}_{r,s}(f), and hence \tilde{\Delta}_{r,s} is a homomorphism of \mathrm{U}_q(\mathfrak{q}_r)-supermodules.

We need to check \tilde{\Delta}_{r,s}\circ\tilde{\Phi}_u=(1\otimes \bar{\Psi}_u)\circ\tilde{\Delta}_{r,s}.

Using the pairing \langle\mathsf{A}_n,\mathrm{U}_q(\mathfrak{q}_n)\rangle, the action \tilde{\Phi} of \mathrm{U}_q(\mathfrak{q}_s) on \mathsf{A}_{r,s} and the action \bar{\Psi} of \mathrm{U}_q(\mathfrak{q}_s) on \bar{\mathsf{A}}_{s,n} can be written as

for u\in\mathrm{U}_q(\mathfrak{q}_s), f\in\mathsf{A}_{r,n} and \bar{g}\in\bar{\mathsf{A}}_{s,n}. Since

we have

Then we compute that

which completes the proof.

Recall from Lemma 3.2 that \mathsf{A}_{r,s} is generated by 1 and t_{i,\alpha} with i=1,\ldots, r and \alpha\in I_{s|s}, the equality (5.5) determines \Omega(f\otimes g) when f,g are those generators of \mathsf{A}_{r,s}. Then the definition of \Omega is extended to the whole vector \mathsf{A}_{r,s}\otimes\mathsf{A}_{r,s} according to (5.5). It suffices to show \Omega is well-defined. Thus, we need to check the following two equations

and

By (5.5), we have

On the other hand,

The equality (Proof) follows from the matrix S satisfies quantum Yang-Baxter equation. Similarly, one can prove (Proof).

The case where p=1 and k=2 is given by (5.5), we prove the statement by induction.

Firstly, we consider the case where k=p+1. By (5.5), we have

For k\geqslant p+2, we deduce by the second equality in (5.5) that

It completes the proof.

We first observe that

which shows (5.8) holds when f=1 or g=1.

The definition of \tilde{\Delta}_{r,s} implies that

On the other hand, Lemma 5.7 yields that

Since the multiplication in \mathcal{O}_{r,s} satisfies

and

we deduce that

This implies that (5.8) holds when f and g are monomials in t_{i\alpha}’s. Now, \mathsf{A}_{r,s} is spanned by all monomials in t_{i\alpha}’s, thus (5.8) holds for all f,g\in\mathsf{A}_{r,s}.

The proof of this theorem splits into the following two cases: n\geqslant \text{max}(r,s) or otherwise.

Case 1: n\geqslant\max(r,s).

In this situation, we have shown in Lemma 5.1 that \tilde{\Delta}_{r,s}:\mathsf{A}_{r,s}\rightarrow\mathcal{O}_{r,s} is an injective linear map. Moreover, if \mathsf{A}_{r,s} (resp. \mathcal{O}_{r,s}) is regarded as the \mathrm{U}_q(\mathfrak{q}_r)\otimes\mathrm{U}_q(\mathfrak{q}_s)-supermodule via the action \Psi\otimes\tilde{\Phi} (resp. \Psi\otimes\bar{\Psi}), the linear map \tilde{\Delta}_{r,s} is also a homomorphism of \mathrm{U}_q(\mathfrak{q}_r)\otimes\mathrm{U}_q(\mathfrak{q}_s)-supermodules.

As a \mathrm{U}_q(\mathfrak{q}_r)\otimes\mathrm{U}_q(\mathfrak{q}_s)-supermodule under the action \Psi\otimes\Phi, the superalgebra \mathsf{A}_{r,s} admits the multiplicity free decomposition

The action \tilde{\Phi} of \mathrm{U}_q(\mathfrak{q}_s) on \mathsf{A}_{r,s} is obtained by twisting \Phi via the automorphism \sigma of \mathrm{U}_q(\mathfrak{q}_s). If we twist both-sides of the decomposition by the automorphism \sigma, we obtain the same decomposition where \mathsf{A}_{r,s} is the \mathrm{U}_q(\mathfrak{q}_r)\otimes\mathrm{U}_q(\mathfrak{q}_s)-supermodule under the action \Psi\otimes\tilde{\Phi}.

On the other hand, Theorem 4.4 implies that the invariant sub-superalgebra \left(\mathcal{O}_{r,s}\right)^{\mathrm{U}_q(\mathfrak{q}_n)} also has the decomposition

provided that n\geqslant\max(r,s).

Now, \tilde{\Delta}_{r,s} is an injective homomorphism of \mathrm{U}_q(\mathfrak{q}_s)-supermodules, where \mathsf{A}_{r,s} and \mathcal{O}_{r,s} are under the \mathrm{U}_q(\mathfrak{q}_s)-action \tilde{\Phi} and \bar{\Psi}, respectively. The decomposition (?) and (?) shows that every irreducible highest weight \mathrm{U}_q(\mathfrak{q}_s)-supermodule with highest weight \lambda has the same multiplicities in both \mathsf{A}_{r,s} and \left(\mathcal{O}_{r,s}\right)^{\mathrm{U}_q(\mathfrak{q}_n)}. Hence, \tilde{\Delta}_{r,s} is surjective.

Although \tilde{\Delta}_{r,s} is not a homomorphism of superalgebras, Proposition 5.8 shows that \tilde{\Delta}_{r,s} sends a product in \mathsf{A}_{r,s} to a linear combinations of products in \mathcal{O}_{r,s}. Since \mathsf{A}_{r,s} is generated by 1 and t_{i\alpha} with i=1,\ldots, r and \alpha\in I_{s|s} and \tilde{\Delta}_{r,s}(t_{i\alpha})=x_{i\alpha}, we conclude that \left(\mathcal{O}_{r,s}\right)^{\mathrm{U}_q(\mathfrak{q}_n)} is generated by x_{i\alpha} with i=1,\ldots, r and \alpha\in I_{s|s}.

Case 2: n<\max(r,s). Let m=\min(n, r,s). We follow the strategy of [LZZ11, Theorem 6.10] that reduces \mathcal{O}_{r,s} to a small superalgebra where the result of Case 1 is applied.

For m\leqslant m', the sub-superalgebra of \mathrm{U}_q(\mathfrak{q}_{m'}) generated by L_{ij} with i\leqslant j and i, j\in I_{m|m}\subseteq I_{m'|m'} is isomorphic to \mathrm{U}_q(\mathfrak{q}_m). Following the standard technique in [LZZ11] and [CW20], for each \mathrm{U}_q(\mathfrak{q}_{m'})-supermodule \mathsf{M}, the subspace

is a \mathrm{U}_q(\mathfrak{q}_{m})-supermodule.

Note that the superalgebra \mathcal{O}_{r,s}=\mathsf{A}_{r,n}\otimes\bar{\mathsf{A}}_{s,n} is a \mathrm{U}_q(\mathfrak{q}_r)\otimes\mathrm{U}_q(\mathfrak{q}_s)\otimes\mathrm{U}_q(\mathfrak{q}_n)-supermodule, we consider the appropriate \mathrm{U}_q(\mathfrak{q}_m)\otimes\mathrm{U}_q(\mathfrak{q}_m)\otimes\mathrm{U}_q(\mathfrak{q}_n)-sub- supermodule

where

and

It is shown in [CW20] that \mathsf{A}_{r,n}^{[m]} (resp. \bar{\mathsf{A}}_{s,n}^{[m]}) is isomorphic \mathsf{A}_{m,m} (resp. \bar{\mathsf{A}}_{m,m}). Hence,

that is generated by x_{i\alpha} with i=1,\ldots,m and \alpha\in I_{m|m} as shown in Case 1.

By Theorem 4.4, the invariant sub-superalgebra has the decomposition

as a \mathrm{U}_q(\mathfrak{q}_r)\otimes\mathrm{U}_q(\mathfrak{q}_s)-supermodule under the action \Psi\otimes\bar{\Psi}. It follows from [CW20, Lemma 4.3] that the irreducible highest weight \mathrm{U}_q(\mathfrak{q}_{m'})-supermodule \mathsf{L}_{m'}(\lambda) gives the irreducible highest weight \mathrm{U}_q(\mathfrak{q}_m)-supermodule \mathsf{L}_{m'}^{[m]}(\lambda)\cong \mathsf{L}_m(\lambda) if \ell(\lambda)\leqslant m, and a highest weight vector in \mathsf{L}_{m'}(\lambda) is also a highest weight vector in \mathsf{L}^{[m]}_{m'}(\lambda). Hence,

It follows that

which shows \left(\mathcal{O}_{r,s}\right)^{\mathrm{U}_q(\mathfrak{q}_n)} is generated by \left(\mathrm{U}_q(\mathfrak{q}_r)\otimes\mathrm{U}_q(\mathfrak{q}_s)\right). x_{i\alpha} with i=1,\ldots, m and \alpha\in I_{m|m}. Therefore, \left(\mathcal{O}_{r,s}\right)^{\mathrm{U}_q(\mathfrak{q}_n)} is generated by x_{i\alpha} with i=1,\ldots, r and \alpha\in I_{s|s} since

and

This completes the proof in the case n<\max(r,s).