Homology of Yang-Baxter modules Yin Tian, Xiao Wang, Yuxin Zhang Abstract . We study the Yang-Baxter operator for the vector representation V m V_{m} of the quantum group U q ​ ( s ​ l m ) U_{q}(sl_{m}) . We consider the one-term Yang-Baxter homology with coefficients in V m V_{m} -modules and provide a direct sum decomposition of the one term Yang-Baxter chain complex. The homology is explicitly computed for some specific V m V_{m} -modules. 2020 Mathematics Subject Classification Primary: 16T25. Secondary: 57K10. 1. Introduction Yang-Baxter equation is a key object both in mathematics and physics. It was discovered by C. N. Yang [ 12 ] and R. J. Baxter [ 2 ] independently. Solutions to the Yang-Baxter equation are deeply related with quantum groups, and can be used to construct invariants of links such as Jones polynomial [ 3 ] and of manifolds such as Rushtikin-Turaev invariants [ 9 ] . For any Yang-Baxter operator, there is a homology theory, which generalizes the quandle homology theory and the latter has been proved useful in knot theory. In this article, we study the one-term homology theory of Yang-Baxter operators, in particular, focusing on the ones R_{m} coming from quantum groups U_{q}(sl_{m}) . We expect our study can help understanding better the two-term Yang-Baxter homology, which is promising in constructing new invariants for links and manifolds. Besides, during our work, we have found interesting algebraic structures coming from the Yang-Baxter operators themselves. We hope our results have reflections on the theory of quantum groups as well. Given a Yang-Baxter operator (V,R) , there is a unital assoiciative algebra F(V) . The notion of V -modules(see Definition 2.6 ) coincides with that of F(V) -modules. For each V -module M , we have a one-term Yang-Baxter chain complex C(M) : \cdots\to M\otimes V^{\otimes n}\to M\otimes V^{\otimes n-1}\to\cdots\to M\otimes V\to M, where the differential is determined by the Yang-Baxter operator R . In this paper, we focus on the Yang-Baxter operators (V_{m},R_{m}) , where V_{m} is the vector representation of U_{q}(sl_{m}) . The groundfield is \mathbb{K}=\mathbb{C}(y) , where y is the quantum parameter. The associative algebra F(V_{m}) is the polynomial algebra in m variables. Our main result is an explicit computation of the one-term Yang-Baxter homology. Theorem 1.1 . For the Yang-Baxter operator (V_{m},R_{m}) , and any V_{m} -module M , we have the followings. (1) The one-term Yang-Baxter complex C(M) is isomorphic to a tensor product C^{f}(M)\otimes B(V_{m}) , where C^{f}(M) is a finite complex of length m , and B(V_{m}) is a graded vector space only depending on (V_{m},R_{m}) . (2) The finite complex C^{f}(M) is isomorphic to M\otimes_{F(V_{m})}K , where K is the Koszul resolution of the one dimensional left F(V_{m}) -module \mathbb{K} . (3) The graded vector space B(V_{m}) is a free algebra generated by b_{i} elements of degree i , for 2\leq i\leq m+1 , where 1-\sum\limits_{i=2}^{m+1}b_{i}q^{i}=(1-mq)(1+q)^{m} . Further discussions: The generators of B(V_{m}) are related to some canonical basis of V_{m}^{\otimes n} ; It is interesting to discuss the two-term Yang-Baxter homology using similar ideas in this paper; It is natural to consider Yang-Baxter operators for other representations of U_{q}(sl_{m}) . The paper is organised as follows. In Section 2, we define the notion of V -modules and the associative algebra F(V) . Then, we show the equivalence between V -modules and F(V) -modules. We also recall the definition of Yang-Baxter equation and that of the one-term Yang-Baxter homology of (V,R) with coefficients in a V -module M . In Section 3, we define an operator \sigma_{n} that depends only on the Yang-Baxter operators (V,R) themselves and study its properties in the case of (V_{m},R_{m}) . Through \sigma_{n} , we obtain an eigenspace decomposition of V_{m}^{\otimes n} , which leads to a decomposition of the one-term Yang-Baxter chain complex of (V_{m},R_{m}) . Furthermore, the above-mentioned B(V_{m}):=\oplus_{n}\ker\sigma_{n} has a graded algebraic structure, and we know the dimension of each \ker\sigma_{n} . For m=2,3 , we explicitly provide generators. We also compute the one-term Yang-Baxter homology of R_{m} with coefficients in F(V_{m}) and certain finite dimensional V_{m} -modules. 2. Preliminary Definition 2.1 . Let k be a commutative ring and V be a k -module. If a k -linear map, R: V\otimes V\to V\otimes V , satisfies the following equation (R\otimes\mathrm{id}_{V})\circ(\mathrm{id}_{V}\otimes R)\circ(R\otimes\mathrm{id}_{V})=(\mathrm{id}_{V}\otimes R)\circ(R\otimes\mathrm{id}_{V})\circ(\mathrm{id}_{V}\otimes R), then we say (V,R) is a pre-Yang-Baxter operator. The equation above is called a Yang-Baxter equation. If, in addition, R is invertible, then we say (V,R) is a Yang-Baxter operator. Remark 2.2 . In this article, we usually abbreviate the notation (V,R) simply by R when V is clear from context. The following is the main example of Yang-Baxter operators we consider in this article. Example 2.3 . [ 8 ] Let \mathbb{K}=\mathbb{C}(y) and V_{m}=\mathbb{K}\{v_{1},\cdots v_{m}\} be an m dimension \mathbb{K} -space. We give a family of Yang-Baxter operators \displaystyle R_{m}:V_{m}\otimes V_{m} \displaystyle\to V_{m}\otimes V_{m} \displaystyle v_{i}\otimes v_{j} \displaystyle\mapsto(1-y^{2})v_{i}\otimes v_{j}+y^{2}v_{j}\otimes v_{i} \displaystyle v_{i}\otimes v_{i} \displaystyle\mapsto v_{i}\otimes v_{i} \displaystyle v_{j}\otimes v_{i} \displaystyle\mapsto v_{i}\otimes v_{j} where 1\leq i1 d^{n}_{k}:=(R\otimes\mathrm{id}_{V^{\otimes n-2}})\circ(\mathrm{id}_{V}\otimes R\otimes\mathrm{id}_{V^{\otimes n-3}})\circ\cdots\circ(\mathrm{id}_{V^{\otimes k-2}}\otimes R\otimes\mathrm{id}_{V^{\otimes n-k}}). We define a linear map \sigma_{n}:V^{\otimes n}\to V^{\otimes n} by \sigma_{n}=\sum_{i=1}^{n}(-1)^{i-1}d^{n}_{i}. A graphical illustration of them is shown in Figure 3.1 : Figure 3.1. \sigma_{n} and {d_{k}^{n}} Remark 3.2 . We note that \partial^{YB}_{n}=(R_{M}\otimes\mathrm{id}_{V^{\otimes n-1}})\circ(\mathrm{id}_{M}\otimes\sigma_{n}) . Here we provide some properties of \sigma_{n} for general Yang-Baxter operators. Lemma 3.3 . The following identities hold for any Yang-Baxter operator R : (1) \sigma_{n}=\sigma_{k}\otimes\mathrm{id}_{V^{\otimes n-k}}+(-1)^{k}({d_{k+1}^{n}}\otimes\mathrm{id}_{V^{\otimes n-k-1}})\circ(\mathrm{id}_{V^{\otimes n-k}}\otimes\sigma_{n-k}). (2) \sigma_{n}\mid_{V^{\otimes k}\otimes\ker\sigma_{n-k}}=\sigma_{k}\otimes\mathrm{id}_{V^{\otimes n-k}}\ and\ \sigma_{n}({V^{\otimes k}\otimes\ker\sigma_{n-k}})\subset{V^{\otimes k}\otimes\ker\sigma_{n-k}}. (3) \forall i\leq j,\ (\mathrm{id}_{V}\otimes d_{j}^{n-1})\circ d_{i}^{n}-(\mathrm{id}_{V}\otimes d_{i}^{n-1})\circ d_{j+1}^{n}=(\sigma_{2}\otimes\mathrm{id}_{V^{\otimes n-2}})\circ(\mathrm{id}_{V}\otimes d_{j}^{n-1})\circ d_{i}^{n}. (4) (\mathrm{id}_{V}\otimes\sigma_{n-1})\circ\sigma_{n}=(\sigma_{2}\otimes\mathrm{id}_{V^{\otimes n-2}})\circ\psi_{n} , where \psi_{n}=\sum_{i\leq j}(-1)^{i+j}(\mathrm{id}_{V}\otimes d_{j}^{n-1})\circ d_{i}^{n}. Proof. (1)(2) Follows directly from the definition of \sigma_{n} (3) This is true since R is an Yang-Baxter operator and the Reidemeister move III can be applied on the diagram representing d_{k}^{n} as in Figure 3.1 . (4) By the definition of \sigma_{n} and (3), we have \displaystyle(\mathrm{id}_{V}\otimes\sigma_{n-1})\circ\sigma_{n} \displaystyle=\sum_{j=1}^{n-1}\sum_{i=1}^{n}(-1)^{i+j}(\mathrm{id}_{V}\otimes d_{j}^{n-1})\circ d_{i}^{n} \displaystyle=\sum_{i\leq j}(-1)^{i+j}((\mathrm{id}_{V}\otimes d_{j}^{n-1})\circ d_{i}^{n}-(\mathrm{id}_{V}\otimes d_{i}^{n-1})\circ d_{j+1}^{n}) \displaystyle=(\sigma_{2}\otimes\mathrm{id}_{V^{\otimes n-2}})\circ\sum_{i\leq j}(-1)^{i+j}(\mathrm{id}_{V}\otimes d_{j}^{n-1})\circ d_{i}^{n}. ∎ Remark 3.4 . Now we fix our Yang-Baxter operator as R_{m} . For the purpose to use less subscriptions, we abbreviate V_{m} as V . We also write F for F(V_{m}) which is actually polynomial algebra in m variables over \mathbb{K} . For later convenience, we introduce the following notation. Definition 3.5 . Consider the finite dimensional vector space V with basis \{v_{1},...,v_{m}\} .For each n , the n -bracket on V^{\otimes n} is a linear map defined on standard tensor basis, \displaystyle{}_{n}:V^{\otimes n}\longrightarrow \displaystyle V^{\otimes n} \displaystyle v_{i_{1}}\otimes\cdots\otimes v_{i_{n}}\mapsto[v_{i_{1}},\cdots,v_{i_{n}}]_{n} \displaystyle:=\sum_{\tau\in S_{n}}(-1)^{{\rm sgn}(\tau)}v_{i_{\tau_{1}}}\otimes v_{{}_{\tau_{2}}}\otimes\cdots\otimes v_{i_{\tau_{n}}}. We denote [V]_{n} to be the image of V^{\otimes n} under [\ ,\,]_{n} . It is clear that [V]_{n}=\{0\} when n>\dim(V) . Remark 3.6 . From now on, when n is clear in the context, we will use the notation [v_{i_{1}},\cdots,v_{i_{n}}] instead of [v_{i_{1}},\cdots,v_{i_{n}}]_{n} . Lemma 3.7 . [ 1 ] The n -bracket [\ ,\,]_{n} is antisymmetric and \displaystyle=\sum_{s=1}^{n}(-1)^{s+1}v_{i_{s}}\otimes[v_{i_{1}},\cdots,\hat{v}_{i_{s}},\cdots,v_{i_{n}}] \displaystyle=\sum_{1\leq s0 . ∎ 3.2. The proof of Theorem 3.8 In this subsection, we will establish the proof of Theorem 3.8 . For that, we will first prove several lemmas. Lemma 3.12 . (F\otimes[V]_{k},\delta_{k}:=(R_{F}\otimes{\rm id}_{V^{\otimes k-1}})\mid_{F\otimes[V]_{k}}) is a chain complex and is isomorphic to the chain complex (F\otimes_{F}\bigwedge^{k}F^{m},{\rm id}_{F}\otimes d_{k}) : {\lx@inpgf@ignorespaces 0} {\lx@inpgf@ignorespaces F\otimes_{F}\bigwedge^{m}F^{m}} {\lx@inpgf@ignorespaces\cdots} {\lx@inpgf@ignorespaces F\otimes_{F}\bigwedge^{2}F^{m}} {\lx@inpgf@ignorespaces F\otimes_{F}F^{m}} {\lx@inpgf@ignorespaces F\otimes_{F}F} {\lx@inpgf@ignorespaces 0} {\lx@inpgf@ignorespaces F\otimes_{\mathbb{K}}[V]_{m}} {\lx@inpgf@ignorespaces\cdots} {\lx@inpgf@ignorespaces F\otimes_{\mathbb{K}}[V]_{2}} {\lx@inpgf@ignorespaces F\otimes_{\mathbb{K}}[V]_{1}} {\lx@inpgf@ignorespaces F} \scriptstyle{\lx@inpgf@ignorespaces\bar{f}_{m}} \scriptstyle{\lx@inpgf@ignorespaces{\rm id}_{F}\otimes d_{m}} \scriptstyle{\lx@inpgf@ignorespaces\bar{f}_{2}} \scriptstyle{\lx@inpgf@ignorespaces\mathrm{id}_{F}\otimes\mathop{d}_{2}} \scriptstyle{\lx@inpgf@ignorespaces\bar{f}_{1}} \scriptstyle{\lx@inpgf@ignorespaces\mathrm{id}_{F}\otimes\mathop{d}_{1}} \scriptstyle{\lx@inpgf@ignorespaces\simeq} \scriptstyle{\lx@inpgf@ignorespaces\delta_{m}} \scriptstyle{\lx@inpgf@ignorespaces\delta_{2}} \scriptstyle{\lx@inpgf@ignorespaces\delta_{1}} where \bar{f}_{k} is defined on the basis by \bar{f}_{k}(x\otimes e_{i_{1}}\wedge\cdots\wedge e_{i_{k}}):=x\otimes[v_{i_{1}},\cdots,v_{i_{k}}]. Therefore, both of them are exact for k\geq 1 . Proof. For k\geq 2 , we have [V]_{k}\subset[V]_{2}\otimes V^{\otimes k-2}={\rm im}\,({\rm id}_{V^{\otimes 2}}-R)\otimes V^{\otimes k-2} by Lemma 3.7 . Thus \delta_{k-1}\circ\delta_{k}=(R_{F}\otimes{\rm id}_{V^{\otimes k-2}})\circ(R_{F}\otimes{\rm id}_{V^{\otimes k-1}})([V]_{k})=0. It is clear that \bar{f}_{k} is an isomorphism and the diagram commutes. We note that H_{k}(F\otimes_{F}\bigwedge^{*}F^{m},{\rm id}_{F}\otimes d_{*}) is Tor_{k}^{F}(F,\mathbb{K}) which is trivial for k\geq 1 . Thus, the lower chain complex is also exact for k\geq 1 . ∎ Lemma 3.13 . The following assertions are true. (1) [V]_{n}=V\otimes[V]_{n-1}\cap[V]_{2}\otimes V^{\otimes n-2}=[V]_{2}\otimes V^{\otimes n-2}\cap V\otimes[V]_{2}\otimes V^{\otimes n-3}\cap\cdots\cap V^{\otimes n-2}\otimes[V]_{2} . (2) Define \phi_{n}^{i}:V^{\otimes n}\to V^{\otimes n} by \phi_{n}^{i}=(\mathrm{id}_{V^{\otimes n-i}}\otimes\sigma_{i})\circ(\mathrm{id}_{V^{\otimes n-i-1}}\otimes\sigma_{i+1})\circ\cdots\circ\sigma_{n}. We have (a) {\rm im}(\phi_{n}^{i})\subset[V]_{n-i+1}\otimes V^{\otimes i-1} . (b) \phi_{n}^{1} is a surjective linear map onto [V]_{n} and \phi_{n}^{1}(v_{i_{1}},\cdots,v_{i_{n}})=y^{2\,{\rm inv}(i_{n},\cdots,i_{1})}[v_{i_{1}},\cdots,v_{i_{n}}], where {\rm inv}(i_{n},\cdots,i_{i}) is the inversion number of the sequence. Proof. (1) Here we denote (R_{F}\otimes{\rm id}_{V^{\otimes n-1}})(F\otimes[V]_{n}) by F[V]_{n} for convenience. By Lemma 3.12 , we have \ker R_{F}=\ker\delta_{1}={\rm im}\,\delta_{2}=F[V]_{2} and F[V]_{n}={\rm im}\,\delta_{n}=\ker\delta_{n-1}=F\otimes[V]_{n-1}\cap F[V]_{2}\otimes V^{\otimes n-2}. We can identify V^{\otimes n} with (R_{F}\otimes{\rm id}_{V^{\otimes n-1}})(\mathbb{K}\otimes V^{\otimes n}) as a subspace of F\otimes V^{\otimes n-1} and obtain the first equality 2 2 2 Here we identify both V^{\otimes n} and its subspaces as subspaces of F\otimes V^{\otimes n-1} . \displaystyle\ [V]_{n} \displaystyle=F[V]_{n}\cap V^{\otimes n} \displaystyle=(F\otimes[V]_{n-1}\cap V^{\otimes n})\cap(F[V]_{2}\otimes V^{\otimes n-2}\cap V^{\otimes n}) \displaystyle=V\otimes[V]_{n-1}\cap[V]_{2}\otimes V^{\otimes n-2}. The second equality can obtain from the first one clearly. (2) (a) Note that {\rm im}\,\sigma_{2}={\rm im}\,({\rm id}_{V^{\otimes 2}}-R)=[V]_{2} . Applying Lemma 3.3 (4) to (\mathrm{id}_{V^{\otimes k}}\otimes\sigma_{n-k})\circ(\mathrm{id}_{V^{\otimes k-1}}\otimes\sigma_{n-k+1}) in \phi_{n}^{i} , we obtain {\rm im}\,\phi_{n}^{i}\subset V^{\otimes k-1}\otimes[V]_{2}\otimes V^{\otimes n-k-1} . As k taking values from 1 to n-i , we have \qquad\qquad{\rm im}\,\phi_{n}^{i}\subset[V]_{2}\otimes V^{\otimes n-2}\cap\cdots\cap V^{\otimes n-1-i}\otimes[V]_{2}\otimes V^{\otimes i-1}=[V]_{n-i+1}\otimes V^{\otimes i-1}. (b) Next we show the formula \phi_{n}^{1}(v_{i_{1}},\cdots,v_{i_{n}})=y^{2\,{\rm inv}(i_{n},\cdots,i_{1})}[v_{i_{1}},\cdots,v_{i_{n}}], which implies \phi_{n}^{1} is surjective. Since {\rm im}\,\phi_{n}^{1}\subset[V]_{n} , we know \phi_{n}^{1}(v_{i_{1}},\cdots,v_{i_{n}})=\alpha[v_{i_{1}},\cdots,v_{i_{n}}], where \alpha\in\mathbb{K} . By direct computation, it is easy to see that the basis v_{i_{n}}\otimes\cdots\otimes v_{i_{1}} in \phi_{n}^{1}(v_{i_{1}},\cdots,v_{i_{n}}) appears only once and whose coefficient is y^{2\,{\rm inv}(i_{n},\cdots,i_{1})} . Therefore, we have \alpha=y^{2\,{\rm inv}(i_{n},\cdots,i_{1})} . ∎ Lemma 3.14 . [V]_{n} is a subspace of the eigenspace of \sigma_{n} with the eigenvalue [n]_{y^{2}}=1+y^{2}+\cdots+y^{2n-2} . Proof. Note that [V]_{2} is the eigenspace of R with the eigenvalue -y^{2} . And we have [V]_{n}=[V]_{2}\otimes V^{\otimes n-2}\cap V\otimes[V]_{2}\otimes V^{\otimes n-3}\cap\cdots\cap V^{\otimes n-2}\otimes[V]_{2}. Thus, \forall k\in\{1,\cdots,n\} and \forall v\in[V]_{n} d^{n}_{k}(v)=(-1)^{k-1}y^{2k-2}v and \sigma_{n}(v):=\sum_{k=1}^{n}(-1)^{k-1}d^{n}_{k}(v)=(1+y^{2}+\cdots+y^{2n-2})v. ∎ Corollary 3.15 . [V]_{k}\otimes\ker\sigma_{n-k} is a subspace of the eigenspace of \sigma_{n} with the eigenvalue [k]_{y^{2}}=1+y^{2}+\cdots+y^{2k-2} . Lemma 3.16 . We observe the following way to count the dimension of V^{\otimes n} : \dim(V^{\otimes n})=\dim(\ker\sigma_{n})+\dim(V\otimes\ker\sigma_{n-1})+\dim([V]_{2}\otimes\ker\sigma_{n-2})+\cdots+\dim([V]_{n}) Proof. Consider the map \phi_{n}^{1}=\phi_{n}^{2}=(\mathrm{id}_{V^{\otimes n-2}}\otimes\sigma_{2})\circ\cdots\circ\sigma_{n} , we have \dim(V^{\otimes n})=\dim((\phi_{n}^{1})^{-1}({\rm im}\,\phi_{n}^{1}))=\dim({\rm im}\,\phi_{n}^{1})+\dim(\ker\phi_{n}^{1}). Then let us further decompose \dim(\ker\phi_{n}^{2})=\dim(\ker\phi_{n}^{1}) , \dim(\ker\phi_{n}^{2})=\dim((\phi_{n}^{3})^{-1}(\ker(\mathrm{id}_{V^{\otimes n-2}}\otimes\sigma_{2})))=\dim\ker\phi_{n}^{3}+\dim({\rm im}\,\phi_{n}^{3}\cap\ker(\mathrm{id}_{V^{\otimes n-2}}\otimes\sigma_{2})) Recurvisely, we have \dim(V^{\otimes n})=\dim({\rm im}\,\phi_{n}^{1})+\sum_{i=2}^{n}\dim({\rm im}\,\phi_{n}^{i}\cap\ker(\mathrm{id}_{V^{\otimes n+1-i}}\otimes\sigma_{i-1}))+\dim(\ker\sigma_{n}). Next, we claim {\rm im}\,\phi_{n}^{i}\cap\ker(\mathrm{id}_{V^{\otimes n+1-i}}\otimes\sigma_{i-1})=[V]_{n+1-i}\otimes\ker\sigma_{i-1} which proves Lemma directly. Firstly, {\rm im}\,\phi_{n}^{i}\subset[V]_{n+1-i}\otimes V^{\otimes i-1} from Lemma 3.13 (3)(a), thus {\rm im}\,\phi_{n}^{i}\cap\ker(\mathrm{id}_{V^{\otimes n+1-i}}\otimes\sigma_{i-1})\subset[V]_{n+1-i}\otimes\ker\sigma_{i-1}. Secondly, \phi_{n}^{i} restricts to V^{\otimes n+1-i}\otimes\ker\sigma_{i-1} , \phi_{n}^{i} will be the same as \phi_{n+1-i}^{1}\otimes\mathrm{id}_{V^{\otimes i-1}} which is surjective to [V]_{n+1-i}\otimes\ker\sigma_{i-1} . Thus we have {\rm im}\,\phi_{n}^{i}\cap\ker(\mathrm{id}_{V^{\otimes n+1-i}}\otimes\sigma_{i-1})\supset[V]_{n+1-i}\otimes\ker\sigma_{i-1}. ∎ Now, let us provide the proof of Theorem 3.8 . Proof of Theorem 3.8 . By Corollary 3.15 , [V]_{k}\otimes\ker\sigma_{n-k} is a subspace of the eigenspace of the eigenvalue [k]_{y^{2}}=1+y^{2}+\cdots+y^{2k-2} . Thus we have the direct sum decompostion since they all belongs to different eigenspaces, \ker\sigma_{n}\oplus V\otimes\ker\sigma_{n-1}\oplus[V]_{2}\otimes\ker\sigma_{n-2}\oplus\cdots\oplus[V]_{n}\subset V^{\otimes n}. And we have \dim(V^{\otimes n})=\dim(\ker\sigma_{n})+\dim(V\otimes\ker\sigma_{n-1})+\dim([V]_{2}\otimes\ker\sigma_{n-2})+\cdots+\dim([V]_{n}) Thus V^{\otimes n}=\ker\sigma_{n}\oplus V\otimes\ker\sigma_{n-1}\oplus[V]_{2}\otimes\ker\sigma_{n-2}\oplus\cdots\oplus[V]_{n}. ∎ 3.3. The structure of the kernel of \sigma_{n} Let m be a fixed integer and M(n)=\dim(\ker\sigma_{n}) . We define M(0)=1 and M(n)=0 when n<0 . In this subsection, we will show that B(V)=\bigoplus_{n}\ker\sigma_{n} is an algebra. In particular, a generating set of B(V) is provided inductively when m=2,3 . First, let us determine M(n) . From the eigenspace decomposition in Theorem 3.8 , we can get some formulas of M(n) . Lemma 3.17 . For M(n) , we have the following: (1) m^{n}=\sum_{i=0}^{m}\binom{m}{i}M(n-i) (2) 0=\sum_{i=0}^{m}(i-1)\binom{m+1}{i}M(n-i) Proof. (1) It follows from Theorem 3.8 directly. (2) \displaystyle 0 \displaystyle=m\cdot m^{n-1}-m^{n}=\sum_{i=0}^{m}m\binom{m}{i}M(n-1-i)-\sum_{i=0}^{m}(m-i)M(n-i) \displaystyle=-M(n)+\sum_{i=0}^{m-1}(m\binom{m}{i}-\binom{m}{i+1})M(n-1-i)+m\binom{m}{m}M(n-m-1) \displaystyle=-M(n)+\sum_{i=0}^{m-1}i\binom{m+1}{i+1}M(n-1-i)+m\binom{m}{m}M(n-m-1) \displaystyle=\sum_{i=0}^{m+1}(i-1)\binom{m+1}{i}M(n-i) ∎ Proposition 3.18 . (B(V),\otimes) is a graded algebra, i.e. \ker\sigma_{s}\otimes\ker\sigma_{t}\subset\ker\sigma_{s+t}. Proof. It follows directly from the Lemma 3.3 (1). ∎ Definition 3.19 . Inductively, for each n , we choose a subspace {\widetilde{\ker}\sigma_{n}} of \ker\sigma_{n} so that ({\ker\sigma_{n-1}}\otimes{\widetilde{\ker}\sigma_{1}}+\cdots+{\ker\sigma_{1}}\otimes{\widetilde{\ker}\sigma_{n-1}})\oplus{\widetilde{\ker}\sigma_{n}}=\ker\sigma_{n} with initial conditions {\widetilde{\ker}\sigma_{1}}=\ker\sigma_{1}=\{0\} . Proposition 3.20 . There is a direct sum : (V^{\otimes n-1}\otimes{\widetilde{\ker}\sigma_{1}})\oplus(V^{\otimes n-2}\otimes{\widetilde{\ker}\sigma_{2}})\oplus\cdots\oplus{\widetilde{\ker}\sigma_{n}} Proof. For n=3 , V\otimes{\widetilde{\ker}\sigma_{2}}\cap{\widetilde{\ker}\sigma_{3}}=\{0\} since they belong to the eigenspace of \sigma_{n} with eigenvalues 1 and 0 respectively. Now we assume that the statement is true for n-1 , we only need to show ((V^{\otimes n-1}\otimes{\widetilde{\ker}\sigma_{1}})\oplus(V^{\otimes n-2}\otimes{\widetilde{\ker}\sigma_{2}})\oplus\cdots\oplus(V\otimes{\widetilde{\ker}\sigma_{n-1}}))\cap{\widetilde{\ker}\sigma_{n}}=\{0\}. We have \displaystyle((V^{\otimes n-1}\otimes{\widetilde{\ker}\sigma_{1}})\oplus(V^{\otimes n-2}\otimes{\widetilde{\ker}\sigma_{2}})\oplus\cdots\oplus(V\otimes{\widetilde{\ker}\sigma_{n-1}}))\cap{\ker\sigma_{n}} \displaystyle{\displaystyle=}((V^{\otimes n-1}\otimes{\widetilde{\ker}\sigma_{1}})\cap\ker\sigma_{n})\oplus((V^{\otimes n-2}\otimes{\widetilde{\ker}\sigma_{2}})\cap{\ker\sigma_{n}})\oplus\cdots\oplus((V\otimes{\widetilde{\ker}\sigma_{n-1}})\cap{\ker\sigma_{n}}) \displaystyle=(\ker\sigma_{n-1}\otimes{\widetilde{\ker}\sigma_{1}})\oplus(\ker\sigma_{n-2}\otimes{\widetilde{\ker}\sigma_{2}})\oplus\cdots\oplus(\ker\sigma_{1}\otimes{\widetilde{\ker}\sigma_{n-1}}), where the first equality holds by Lemma 3.3 (2), which is stating \sigma_{n} preserves the direct sum structure. Therefore, ((V^{\otimes n-1}\otimes{\widetilde{\ker}\sigma_{1}})\oplus(V^{\otimes n-2}\otimes{\widetilde{\ker}\sigma_{2}})\oplus\cdots\oplus(V\otimes{\widetilde{\ker}\sigma_{n-1}}))\cap{\widetilde{\ker}\sigma_{n}}=\{0\}. ∎ Corollary 3.21 . The sums in Definition 3.19 are actually direct sums. Therefore, B(V) , as an algebra, is generated by the chosen basis of \widetilde{\ker}\sigma_{i} . {\widetilde{\ker}\sigma_{n}}\oplus({\ker\sigma_{n-1}}\otimes{\widetilde{\ker}\sigma_{1}})\oplus\cdots\oplus({\ker\sigma_{1}}\otimes{\widetilde{\ker}\sigma_{n-1}})=\ker\sigma_{n} Proof. It is clearly since \ker\sigma_{n-k}\otimes\widetilde{\ker}_{k}\subset V^{\otimes n-k}\otimes\widetilde{\ker}_{k} . ∎ By dimension arguments, we see that {\widetilde{\ker}\sigma_{n}} will vanish for large n . Theorem 3.22 . When n>m+1 , {\widetilde{\ker}\sigma_{n}}=\{0\} i.e. \ker\sigma_{n}=({\ker\sigma_{n-2}}\otimes{\widetilde{\ker}\sigma_{2}})\oplus\cdots\oplus({\ker\sigma_{n-m-1}}\otimes{\widetilde{\ker}\sigma_{m+1}}) Proof. Firstly, it is a direct sum from the Corollary 3.21 . And we have ({\ker\sigma_{n-2}}\otimes{\widetilde{\ker}\sigma_{2}})\oplus\cdots\oplus({\ker\sigma_{n-m-1}}\otimes{\widetilde{\ker}\sigma_{m+1}})\subset\ker\sigma_{n} from Proposition 3.18 . Thus we only need to show the dimensions match, for which, we now prove \dim({\widetilde{\ker}\sigma_{i}})=(i-1)\binom{m+1}{i} . For i=2 , we have {\widetilde{\ker}\sigma_{2}}=\ker\sigma_{2}=span_{\mathbb{K}}\{v_{i}\otimes v_{i}\,,v_{i}\otimes v_{j}+y^{2}v_{j}\otimes v_{i}\mid i