On results of certain modules over untwisted affine Liesuperalgebras

Repeat the proof of [Yousofzadeh:Finite, weight, 2020, Lemma 3.4].

Just repeat the proof of [Yousofzadeh:Finite, weight, 2020, Lemma 3.5].

Just repeat the proof of [Yousofzadeh:Finite, weight, 2020, Proposition 3.6].

Assume that \zeta (\delta)> 0. Fix p and \lambda as mentioned. Let \dot{\beta}\in \dot R_{re}. Then for large enough k>0, we have

Throughout the proof, t_{\dot{\beta}} shall denote the smallest positive integer with the property \zeta (\dot{\beta}+t_{\dot{\beta}}\delta)>0.

Define

Clearly, {\mathcal S}\subseteq {\mathcal P}\subseteq T_{re}^+. Furthermore, if \mu_0\in\mathcal A, then one can actually infer that the set

is finite. We now break the proof into the following claims.

Claim 1. \mathcal A \neq \emptyset. To see this, we show that \lambda \in \mathcal A. Let \alpha \in T_{re}^+ with \lambda +\alpha \in {\rm supp}(W). We must prove that \alpha \in \mathcal P. Since \alpha \in T_{re}^+, we can assume that \alpha =\dot{\alpha}+m\delta for some \dot{\alpha}\in \dot{T}_{re} and m\in \Bbb Z with m\geq t_{\dot{\alpha}}. We can write m-t_{\dot{\alpha}}=pq+r where q\in \Bbb Z^{\geq 0} and 0\leq r\leq p-1. Assume that q\neq 0. Then, we have

Clearly, -\left(\dot{\alpha}+(t_{\dot{\alpha}}+r)\delta\right)\in T^-_{re}. By the assumption T^-_{re}\subseteq \frak C_M, hence the element

is contained in {\rm supp}(W). On the other hand, pq>0. It follows that \lambda +qp\delta \in (\lambda +\Bbb Z^{>0}p\delta)\cap {\rm supp}(W)=\emptyset, a contradiction. Thus q=0. This implies that \alpha =\dot{\alpha}+(t_{\dot{\alpha}}+r)\delta \in \mathcal P, and so \lambda \in \mathcal A.

Claim 2. For any \mu \in \mathcal A, the set

is finite. Suppose on the contrary that for some \mu_0\in \mathcal A, there are infinitely many m\delta \in \Bbb Z^{>0}\delta with

Suppose that \dot\alpha_*\in \dot T_{re}. Then, \dot\alpha_*+t_{\dot\alpha_*}\delta\in T^+_{re}. Thus, by choosing infinitely many n\delta\in Q with n\geq t_{\dot\alpha_*}, we have \mu_0+n\delta\in {\rm supp}(W). As -(\dot\alpha_*+t_{\dot\alpha_*}\delta)\in T^-_{re} \subseteq {\frak C}_M, we get

for infinitely many n. So, -\dot\alpha_*+(n-t_{\dot\alpha_*})\delta\in T^+_{re} for sufficiently large numbers n. This contradicts the fact (*).

Claim 3. There is \eta\in {\rm supp}(W) such that \eta+m\delta \not \in {\rm supp}(W) for all m\in \Bbb Z^{>0}. To see this, by Claim 1, pick \mu_0 \in {\mathcal A}\subseteq {\rm supp}(W). By Claim 2, we can choose the greatest N\in \Bbb Z^{>0} such that \mu_0+N\delta \in {\rm supp}(W). Set \eta=\mu_0+N\delta. It is clear that \eta+m\delta \not\in {\rm supp}(W) for all m\in\Bbb Z^{>0}.

Claim 4. There is \eta\in X such that (\eta+{\rm span}_{\Bbb Z^{\geq 0}}{\mathcal S})\cap {\rm supp}(W)=\{\eta\} where

First, we note that, by Claim 3, X is a nonempty set. We apply Proposition 3.3(ii) to prove this claim. Assuming this proposition, we must have

as claimed. Now, let us show how to apply this proposition. It is clear that \mathcal S is a nonempty subset of T. Note that R_{im}=\Bbb Z\delta and R_{im}^\pm:= R_{im}\cap R^\pm ={\Bbb Z}^{{\gtrless} 0}\delta, because \zeta (\delta)>0, and \zeta (k\delta)=k\zeta (\delta) for all k\in\Bbb Z. Thus, \mathcal S does not contain imaginary roots. According to the assumption and material mentioned above, we have {\mathcal S}\subseteq {\mathcal P}\subseteq T_{re}^+ \subseteq \frak B_W and T_{re}^-\subseteq \frak C_M, and hence -{\mathcal S}\subseteq \frak C_M. It only remains to show that

is true so that we must meet all the conditions of Proposition 3.3(ii). Assuming the contrary, there are \eta \in X and \beta \in \mathcal S such that \eta +\beta \in {\rm supp}(W) and \eta +\beta \not \in X. Hence, there is m\delta \in Z^{>0}\delta with \eta +\beta +m\delta \in {\rm supp}(W). Since -\beta \in \frak C_W, we can deduce that

which contradicts the fact that \eta \in X.

Claim 5. There is \mu \in {\rm supp}(W) such that

To prove this, we can deduce from Claim 4 that there is at least one \mu with the following property

Now, suppose on the contrary that there is \alpha \in T^+_{re}\cup \Bbb Z^{>0}\delta with \mu +\alpha \in {\rm supp}(W). If \alpha \in \Bbb Z^{>0}\delta, then \mu +\alpha =\mu by (**). This implies that \alpha =0, a contradiction. Hence \alpha \in T^+_{re}, and so \alpha =\dot{\alpha}+m\delta for some \dot{\alpha}\in \dot{T}^\times _{re} and m\geq t_{\dot{\alpha}}. Since -(\dot{\alpha}+t_{\dot{\alpha}}\delta) \in T^-_{re}\subseteq \frak C_W, we have

Thus \mu+(m-t_{\dot{\alpha}})\delta =\mu by (**). This implies that m=t_{\dot{\alpha}}. Hence, \alpha \in \mathcal S. This again leads us to \mu+\alpha=\mu by (**), a contradiction. Therefore, there is no \alpha \in T^+_{re}\cup \Bbb Z^{>0}\delta with \mu +\alpha \in {\rm supp}(W), and the proof is complete.
If \zeta (\delta)<0, then we use -\delta instead of \delta and repeat the above process.

Note that

Let 0\neq \beta\in R. If \beta \in R_{ns}, then \beta=\dot{\beta}+r_{\dot\beta}\sigma+n\delta in which \dot{\beta}\in \dot{R}_{ns} and n\in \Bbb Z.

Since \zeta (\delta)>0, we can find k\in \Bbb Z such that

The same argument holds for \beta \not\in R_{ns}. Throughout the proof, m_{\dot{\beta}} shall denote the smallest integer with the property

and we define \beta_{\dot{\alpha}}:=\dot{\alpha}+r_{\dot\alpha}\sigma +m_{\dot{\alpha}}\delta, and we set \Phi as the set of all such \beta_{\dot{\alpha}}s. We now break the proof into the following claims.

Claim 1. B is equal to

To see this, let v\in B. Thus, for each \dot{\alpha}\in \dot{R}_{ns}, there is N_{\dot{\alpha}}\in \Bbb Z^{\geq 0} with {\frak L}^{\dot \alpha+r_{\dot\alpha}\sigma+n\delta}v=0 for all n\geq N_{\dot{\alpha}}. Set N:=\max\{N_{\dot{\alpha}}-m_{\dot{\alpha}}~|~\dot{\alpha}\in \dot{R}_{ns}\}. Then, {\frak L}^{\beta_{\dot{\alpha}}+n\delta}v=0 for all n\geq N and \dot{\alpha}\in \dot{R}_{ns}. This means B\subseteq B'. Conversely, suppose v\in B' and pick N\in \Bbb Z ^{\geq 0} with {\frak L}^{\beta_{\dot{\alpha}}+n\delta}v=0 for all n\geq N and \dot{\alpha}\in \dot{R}_{ns}. Hence, for each \dot{\alpha}\in \dot{R}_{ns} and n\geq N+m_{\dot{\alpha}}, we have {\frak L}^{\beta_{\dot{\alpha}}+n\delta}v=0, that is v\in B.

Claim 1 leads us to define the

and

Claim 2. Assume v\in B, N\in \Bbb Z^{\geq 0} and \alpha \in C_v with the following properties

• (1) {\frak L}^{{\alpha}+N\delta}v\neq 0,

• (2) If \alpha'\in C_v and {\frak L}^{{\alpha'}+N\delta}v\neq 0, then \zeta (\alpha')\leq \zeta (\alpha),

• (3) For all m\in \Bbb Z^{\geq 1} and \alpha'\in C_v, {\frak L}^{{\alpha'}+N\delta+m\delta}v=0.

Then for 0\neq w\in {\frak L}^{{\alpha}+N\delta}v, w\in B.
To prove this, we first show that {\frak L}^{m\delta}w=0 for all m\in \Bbb Z^{\geq 1}. Now, we have

Now, let \beta \in R_{re} with \zeta (\beta)>0. Since v\in \mathcal A, we have {\frak L}^{\beta}v=0. Thus

If \alpha+\beta+N\delta is not a root, then {\frak L}^{\alpha+\beta+N\delta}v=0, and so {\frak L}^{\beta}w=0. Assume that \alpha+\beta+N\delta \in R_{re}. Since \alpha \in C_v, one can show that \zeta (\alpha)>0. On the other hand, \zeta(\beta)>0 by the assumption. Thus

It follows that \alpha+\beta+N\delta\in R^+\cap R_{re}=R_{re}^+, and hence {\frak L}^{\alpha+\beta+N\delta}v=0 (because v\in\mathcal A). This implies that {\frak L}^{\beta}w=0.
Now, assume that \alpha+\beta+N\delta\in R_{ns}. This forces that \alpha+\beta \in R_{ns}. Thus \alpha +\beta =\dot{\gamma}+ r_{\dot\gamma}\sigma +k\delta where \dot{\gamma}\in \dot{R}_{ns} and k\in \Bbb Z. Since \zeta (\alpha+\beta)>0, we can write \alpha +\beta=\dot{\gamma}+r_{\dot\gamma}\sigma+(m_{\dot{\gamma}}+m)\delta where m\in \Bbb Z^{\geq 0}. If m=0, then \alpha +\beta \in C_v. However, \zeta (\alpha +\beta)=\zeta (\alpha)+\zeta (\beta)> \zeta (\alpha). By (2), we get {\frak L}^{\alpha+\beta+N\delta}v= 0, and so {\frak L}^{\beta}w=0. If m>0, then by (3), {\frak L}^{\alpha+\beta+N\delta}v= 0, and so {\frak L}^{\beta}w=0. Thus, {\frak L}^{\beta}w=0 for all 0\neq \beta \in R^+\cap R_{re}.

Now, we show that w\in B=B' (see Claim 1). Thus by the above argument, it is enough to prove that there is a positive integer p such that for all \eta \in \Phi \cap R_{ns} and n\geq p, {\frak L}^{\eta+n\delta}w=0. Since v\in B=B', we can choose p\in \Bbb Z^{>0} such that {\frak L}^{\eta+n\delta}v=0 for all \eta \in \Phi \cap R_{ns} and n\geq p. Hence for all \eta \in \Phi \cap R_{ns} and n\geq p,

Since C_v\subseteq R_{ns}, it follows that \alpha and \eta belong to R_{ns}. If \eta+\alpha+n\delta+N\delta is not a root, then {\frak L}^{\eta+\alpha+n\delta+N\delta}v=0, and so {\frak L}^{\eta+n\delta}w=0. Hence, assume that \eta+\alpha+n\delta+N\delta\in R. We know that the sum of two nonsingular roots is either a real root or an imaginary root. If \eta+\alpha+n\delta+N\delta\in R_{im}, then we must have \eta+\alpha=0. Since \alpha,\eta\in \Phi, it follows that \zeta(\alpha+\eta)>0, which contradicts the condition \alpha+\eta=0. Thus, \eta+\alpha+n\delta+N\delta\in R_{re}. On the other hand, \zeta (\eta+\alpha+n\delta+N\delta)>0, that is \eta+\alpha+n\delta+N\delta\in R^+_{re}. This implies that {\frak L}^{\eta+\alpha+n\delta+N\delta}v=0, because v\in \mathcal A. Therefore, {\frak L}^{\eta+n\delta}w=0.

Now, just repeat Claim 3, Claim 4, Claim 5 and Claim 6 of [Yousofzadeh:Finite, weight, 2020, Proposition 3.8].

Repeat the proof of [Yousofzadeh:Finite, weight, 2020, Proposition 4.4].

(i). If \alpha \in {\frak C}_M, then by Lemma 3.2(ii),

for any t\in \Bbb Z^{>0}. To prove the converse, let t\alpha \in {\frak C}_M for some t\in \Bbb Z^{>0}. Since \alpha \in R_{re}=R^{in}\cup R^{ln}, we have \alpha \in R^{in} or \alpha \in R^{ln}. If \alpha \in R^{ln}, then \alpha \in \frak B_M. This implies that there are finitely many positive integers k_1, \cdots, k_m with \eta +k_i\alpha \in {\rm supp}(M), 1\leq i\leq m, for some \eta \in {\rm supp}(M). As t\alpha \in {\frak C}_M, we have

Again, as t\alpha \in {\frak C}_M, we have

This process can continue unabated. Thus for large enough i\in \Bbb N, \eta +(it+k_1)\alpha is distinct from \eta +k_1\alpha, \eta +k_2\alpha, \cdots, and \eta +k_m\alpha, which is contradiction. Thus \alpha \in R^{in}\subseteq \frak C_M.
(ii). We first note that, according to Table ? and Table ?, in types B(m,n)^{(1)} and G(3)^{(1)}, some roots belong to both R_{re} and R_1. Let \alpha \in R_{re}\cap R_1. It is clear that 2\alpha\in R_{re}\cap R_0. On the other hand, there are x\in \frak L^\alpha and y\in \frak L^{-\alpha} such that {\rm span}_{\Bbb C}\{x, y,h := [x, y], [x, x], [y, y]\} is a Lie superalgebra isomorphic to \frak {osp}(1, 2) with \alpha(h) =2 (we refer to the material before Lemma 3.1). This leads us to the \frak{sl}_2-triple (\dfrac{1}{4}[x,x],\dfrac{-1}{4}[y,y], \dfrac{1}{2}h) for 2\alpha, and consequently r_\alpha=r_{2a}. Thus, by (i), it is enough to prove the statement (2) for \alpha\in R_{re}\cap R_0. As well, we note that if \alpha, \gamma\in R_{re} then r_\alpha(\gamma)\in R_{re}.

Now, assume that \pm\alpha\in R^{ln}.

Then,

Because the opposite direction of the arrows is also correct, the proof of the case \pm\alpha\in R^{ln} is complete.

Now, assume that \pm\alpha\in R^{in}. If \gamma\in R^{in}, then by Lemma 3.2(ii), r_\alpha(\gamma)=\gamma +m\alpha \in \frak C_M. Hence, r_\alpha(\gamma)\in R^{in}. Conversely, assume that \gamma +m\alpha=r_\alpha(\gamma)\in R^{in}. Since \pm\alpha\in R^{in}, we have \pm m\alpha\in \frak C_M by Lemma 3.2(ii). Using Lemma 3.2(ii) again, \gamma=\underbrace{-m\alpha}_{\in \frak C_m}+\underbrace{\gamma +m\alpha}_{\in \frak C_m}\in \frak C_M. So, \gamma \in R^{in}.

Repeat the proof of [Yousofzadeh:Finite, weight, 2020, Theorem 4.7] and use the above results (presented for the untwisted cases) in the body of that proof.

We can write \beta=\dot{\beta}+n\delta in which \dot{\beta}\in \dot{R}_{re} and n\in \Bbb Z. Clearly,

Thus, if (i) and (ii) do not occur, then

or

for some k\in \Bbb Z. We claim that if (*) holds, then (iii) happens. Before starting the main body of the proof of the claim, we show that the situation

cannot happen. Suppose on the contrary that -\gamma \in R^{ln} and -\gamma -\delta \in R^{in}. Then, according to (*), we have \pm\gamma\in R^{ln} or \pm(\gamma+\delta)\in R^{in}. Thus by Lemma 3.7(ii)

Let \mu \in {\rm supp}(M). As -\gamma+\delta\in R^{in}, it follows that \mu+(-\gamma+\delta)\in {\rm supp}(M). On the other hand, since \gamma -\delta\in R^{in}\subseteq \frak C_M, we have \mu\in {\rm supp}(M). So, by similar arguments, we can easily show that

Now, let \eta \in {\rm supp}(M). Since \gamma \in R^{ln}\subseteq \frak B_M, we can choose k\in \Bbb Z^{\geq 1} such that

which is a contradiction. We now proceed with the remaining cases for -\gamma and -\gamma -\delta.

Case 1. -\gamma \in R^{ln} and -\gamma -\delta \in R^{ln}.
Assume that there is p\in\Bbb Z such that \gamma+p\delta\in R^{in}. AS \pm\gamma \in R^{ln}, it follows from Lemma 3.7(ii) that

Again by Lemma 3.7(ii),

Thus, we can prove that

for all p\in \Bbb Z. Particularly,

On the other hand, if \pm \gamma +2\delta\in R^{ln}, then by Theorem 3.8

and so by Lemma 3.2(ii)

This contradicts (?). Hence, \pm \gamma +2\delta\in R^{in}. Now, assume that n\in \Bbb Z^{\geq 1}. Then by Lemma 3.2(ii)

Furthermore, by Lemma 3.2(ii)

Now, assume that n\in \Bbb Z^{\leq 0}. If \pm \gamma +2n\delta\in R^{in}, then by Lemma 3.2(ii)

This contradicts the assumption \pm\gamma \in R^{ln}. Hence, \pm \gamma +2n\delta\in R^{ln}. Therefore, we have showed that

This means that (iii) holds.

Case 2. -\gamma \in R^{in} and -\gamma -\delta \in R^{in}.
Assume that \gamma-n\delta\in R^{in} for some n\in\Bbb Z^{>0}. Then by Lemma 3.2(ii) and (*)

This contradicts the assumption \gamma \in R^{ln}, see (*). Thus \gamma-n\delta\in R^{ln} for all n\in\Bbb Z^{\geq 0}. Based on (*), we have \gamma +\delta\in R^{in}. So by Lemma 3.7,

belongs to R^{ln}, because \gamma \in R^{ln}. If -\gamma-n\delta\in R^{in} for n\geq 3, then consider the following equation

By Lemma 3.2(ii) and (*), the right hand side of the equation belongs to R^{in}, while the left hand side of the equation belongs to R^{ln}, a contradiction. Thus, -\gamma-n\delta\in R^{ln} for all n\in\Bbb Z^{\geq 3}.
Now, assume that n\in \Bbb Z^{\geq 0}. Then by Lemma 3.2(ii) and (*)

Note that by Lemma 3.2(ii), we have

in which n\geq 1. Therefore, we show that

This means that (iii) holds.

Case 3. -\gamma \in R^{in} and -\gamma -\delta \in R^{ln}.
Let n\in \Bbb Z^{\geq 0}. Then for n\geq 0, we have

Assume that \gamma -k\delta\in R^{in} for some k\in \Bbb Z ^{\geq 1}. Since -\gamma+k\delta=-(\gamma-k\delta)\in R^{in} (above), we have

by Lemma 3.7. However, according to the above, the left side belongs to R^{in}, which is a contradiction. Thus, \gamma -n\delta\in R^{ln} for all n\in \Bbb Z ^{\geq 0}. On the other hand, by using the above and Theorem 3.8

for all n\in \Bbb Z ^{\geq 1}. Therefore, we show that

This means that (iii) holds.
Regarding (iv), in (**), we have

This implies that (**) turns to (*), and we get (iv). The proof is complete.

(i) Choose an element \alpha=\dot\alpha+m\delta\in S_{re} for some m\in \Bbb Z. Note that \dot\alpha\neq 0. If \alpha\in R_0, then \dot\alpha\in \dot S. If not, according to Table ?, Table ? and Table ?, we have \alpha\in R_1\cap R_{re}\subseteq R_0. Since S is closed and 2(R_1\cap R_{re})\subseteq R_0, we have 2\alpha \in S\cap R_0. Hence, we get \dot S\neq \{0\}. Being a finite root system \dot S is easily verified.
(ii) We prove the statement for type A(n,n)^{(1)} and the proof can be repeated for type X\neq A(n,n)^{(1)} with a similar process. So, we use the notations related to type A(n,n)^{(1)} in the proof process freely.
Assume that (rc_2+sd_2,\Bbb Cc_2\oplus \Bbb Cd_2)=\{0\} where r,s\in \Bbb C. Since (rc_2+sd_2,c_2)=0 and (rc_2+sd_2,d_2)=0, we have r=s=0. Thus, the restriction of form on \Bbb Cc_2\oplus \Bbb Cd_2 is nondegenerate. According to (i), \dot S(i) is an irreducible finite root system. If \sum_{\dot\alpha\in \dot S(i)}\frak L^{\dot \alpha} is a simple Lie algebra, then \sum_{\dot\alpha\in \dot S(i)}\Bbb Ct_{\dot\alpha} is contained in the Cartan subalgebra. Hence, form (-,-) is nondegenerate on \sum_{\dot\alpha\in \dot S(i)}\Bbb Ct_{\dot\alpha}. Otherwise, the Lie algebra \sum_{\dot\alpha\in \dot S(i)}\frak L^{\dot \alpha} is of type BC(n), and it contains \sum_{\dot\alpha\in B(n)}\frak L^{\dot \alpha}. Hence, \sum_{\dot\alpha\in \dot S(i)}\Bbb Ct_{\dot\alpha}\subseteq \sum_{\dot\alpha\in B(n)}\frak L^{\dot \alpha}. This leads us that the form (\cdot,\cdot) is nondegenerate on \sum_{\dot\alpha\in \dot S(i)}\Bbb Ct_{\dot\alpha}. Recalling (?), we have

Let \alpha=\dot \alpha +k\delta \in S(i) for some k\in \Bbb Z. Note that t_\alpha=r_1c_1+r_2c_2+s_1d_1+s_2d_2+h (we identify h\otimes 1 by h). This implies that

and

Thus

So, t_\alpha=r_2c_2+h=kc_2+t_{\dot\alpha}. Therefore, actually, with the help of (Proof), we have shown that

Define

It follows that the form is nondegenerate on \frak H(i) and so, there is an orthogonal complement T_i for \frak H(i) in \frak H. Now, for \alpha\in S(i), x\in \frak L^\alpha and h\in T_i, we have

that is, T_i is contained in the center of \frak L(i). Define

We show that

Clearly, \frak H(i)\oplus\raisebox{0.2ex}{\scalebox{.8}{_S(i)^ }}\frak{L}^{\alpha}\oplus\raisebox{0.2ex}{\scalebox{.8}{_0kZ }}(\sum_{ \dot\alpha\in \dot S(i)^\times}\sum_{t\in{\mathbb Z}}[\frak L^{\dot\alpha+t\delta},\frak L^{-\dot\alpha+(k-t)\delta}])\subseteq \mathfrak{k}(i). Conversely, assume that u\in [\frak L^\alpha,\frak L^{\beta}]\subseteq \frak L^{\alpha+\beta} for some \alpha,\beta \in S{(i)}^\times. Let \alpha=\dot\alpha+m\delta and \beta=\dot\beta+n\delta for some m,n\in \Bbb Z. Now, if \dot\alpha+\dot \beta\neq 0, then \alpha+\beta\in S(i)^\times, because S(i) is closed. So, \frak L^{\alpha+\beta}\subseteq \raisebox{0.2ex}{\scalebox{.8}{_S(i)^ }}\frak{L}^{\alpha}. This shows that

Otherwise, if \dot\alpha+\dot \beta= 0, then there is k\in \Bbb Z such that n=k-m and

Therefore, the desired equality is achieved.
For each element \alpha=\dot\alpha+k\delta\in S(i) where k\in \Bbb Z, we have

This forces that if \alpha\in S(i) and \beta\in S(i) are distinct on \frak H, then they are distinct on \frak H(i). Hence, the decomposition

coincides with the root space decomposition of \mathfrak{k}(i) with respect to \frak H(i).
We claim that \mathfrak{k}(i) is an affine Lie algebra. To prove this, we want to apply Theorem 3.11. Therefore, we will provide the details in the following steps:

• S1. The form (-, -) is nondegenerate and invariant on \frak k(i).
  Clearly, the invariant is inherited from \frak L to \frak k(i). Assume that x\in \frak k(i) and (x,\frak k(i))=0. We can write

  x=\underbrace{r_1t_\alpha+r_2c_2+r_3d_2}_{\in \frak H(i)}+\underbrace{r_4y_{\beta}}_{\in \sum_{ \alpha\in {S(i)^\times}}\frak L^\alpha}+\underbrace{r_5h\otimes t^l}_{\in \sum_{ \dot\alpha\in \dot S(i)^\times}\sum_{m,n\in{\mathbb Z}}[\frak L^{\dot\alpha+m\delta},\frak L^{-\dot\alpha+n\delta}]}.

  As (x,c_2)=(x,d_2)=0, we get r_2=r_3=0. Thus

  x=r_1t_\alpha+r_4y_\beta+r_5h\otimes t^l.

  We know that t_\alpha and y_\beta identify by h'\otimes t^0 and y_{\dot\beta}\otimes t^m respectively. It follows from (x,h'\otimes t^0)=0 that r_1=0. By choosing the appropriate members of \frak k(i) and repeating the above process, it is concluded that r_4=r_5=0. Hence, x=0.

• S2. \frak H(i) is finite dimensional, abelian and self-centralizing. Furthermore, {\rm ad}_h:\frak k(i)\longrightarrow \frak k(i) is diagonalizable for all h\in \frak H(i).
  Clearly, finite dimensional and abelian are inherited from \frak H to \frak H(i). We have

  \frak H(i)\subseteq \{x\in \frak k(i) \mid [x,\frak H(i)]=\{0\}\}.

  Conversely, assume that x\in \frak k(i) and [x,\frak H(i)]=\{0\}. Then, x\in \frak L^\alpha with \alpha\neq 0. Thus for all h\in \frak H(i), we get

  \alpha(h)x=[h,x]=-[x,h]=0.

  It follows that {\rm ad_h} is zero, a contradiction. For the last statement, we just note that {\rm ad}_h is diagonalizable for all h\in \frak H.

• S3. {\rm ad}_x:\frak k(i)\longrightarrow \frak k(i) is locally nilpotent for all x\in \frak k^\alpha and \alpha\in S(i)_{re}.
  For y\in \frak L^\beta, we have

  ({\rm ad}_x)^l(y)=\underbrace{[x[x,\cdots ,\underbrace{x,\underbrace{[x,y]}_{\in \frak L^{\alpha+\beta}}]}_{\in\frak L^{2\alpha+\beta}}\cdots]}_{\in \frak L^{l\alpha+\beta}}.

  This is while l\alpha+\beta is not root for sufficiently large l.

• S4. S(i)_{re} is irreducible.
  Assume that S(i)_{re}=A\cup B with (A,B)=0, A\neq \emptyset and B\neq\emptyset . Then \dot S(i)=\dot A\cup \dot B with (\dot A,\dot B)=0. Since \dot S(i) is irreducible, it follows that either \dot A=\emptyset or \dot B=\emptyset. This leads us to either A=\emptyset or B=\emptyset, a contradiction.

• S5. S(i)_\circ has no isolated roots and {\rm span}_{\Bbb Z}(S(i)_\circ)\cong \Bbb Z.
  Note that S(i)_\circ=\Bbb Z\delta and (\dot\alpha+\Bbb Z\delta)\cap R_0\subseteq S for 0\neq \alpha\in \dot S. So, we have {\rm span}_{\Bbb Z}(S(i)_\circ)=\Bbb Z\delta\cong \Bbb Z.

• S6. (\mathfrak{k}(i))_c=\sum_{ \alpha\in {S(i)^\times}}\frak L^\alpha\oplus\sum_{ \dot\alpha\in \dot S(i)^\times}\sum_{m,n\in{\mathbb Z}} [\frak L^{\dot\alpha+m\delta},\frak L^{-\dot\alpha+n\delta}].
  Note that R_0\setminus \{0\}\subseteq R_{re} according to Table ? and Table ?. Thus, if 0\neq \alpha\in S(i)^\times \subseteq R_0, then (\alpha,\alpha)\neq 0, and so \sum_{ \alpha\in {S(i)^\times}}\frak L^\alpha\subseteq (\mathfrak{k}(i))_c. Assume that \dot \alpha, -\dot\alpha \in \dot S(i)^\times. Then \dot\alpha+m\delta, -\dot\alpha+n\delta \in S(i)^\times\subseteq R_0 where m,n\in \Bbb Z. Clearly, \frak L^{\dot\alpha+m\delta}, \frak L^{-\dot\alpha+n\delta}\subseteq (\mathfrak{k}(i))_c. As (\mathfrak{k}(i))_c is algebra, it follows that [\frak L^{\dot\alpha+m\delta}, \frak L^{-\dot\alpha+n\delta}]\subseteq (\mathfrak{k}(i))_c. So far we have shown that

  \sum_{ \alpha\in {S(i)^\times}}\frak L^\alpha\oplus\sum_{ \dot\alpha\in \dot S(i)^\times}\sum_{m,n\in{\mathbb Z}}[\frak L^{\dot\alpha+m\delta},\frak L^{-\dot\alpha+n\delta}]\subseteq(\mathfrak{k}(i))_c.

  To achieve equality, it is enough to note that the materials of (\mathfrak{k}(i))_c are either materials in the form \frak L^\alpha or a summation of materials in the form of [\frak L^\alpha,\frak L^\beta] \subseteq \frak L^{\alpha+\beta}.

• S7. \mathfrak{k}(i)=(\mathfrak{k}(i))_c\oplus \Bbb Cc_2.
  We note that \frak H(i) :=\Bbb Cc_2\oplus \Bbb Cd_2\oplus \sum_{\dot\alpha\in \dot S(i)^\times}\Bbb Ct_{\dot\alpha} and

  \mathfrak{k}(i)=\frak H(i)\oplus\raisebox{0.2ex}{\scalebox{.8}{$\displaystyle \bigoplus_{\alpha \in S{(i)}^\times}\;$}}\frak{L}^{\alpha}\oplus\raisebox{0.2ex}{\scalebox{.8}{$\displaystyle \bigoplus_{0\neq k\in{\mathbb Z}}\;$}}(\sum_{ \dot\alpha\in \dot S(i)^\times}\sum_{t\in{\mathbb Z}}[\frak L^{\dot\alpha+t\delta},\frak L^{-\dot\alpha+(k-t)\delta}]).

  Since [\frak L^{\dot \alpha +m\delta},\frak L^{-\dot\alpha+l\delta}]\subseteq \frak L^{(m+l)\delta}=\frak H(i)\otimes t^{m+l}, we have

  \begin{aligned} [h_1\otimes t^m, h_2\otimes t^l]&=[h_1,h_2]\otimes t^{m+l}+\omega(h_1\otimes t^m,h_2\otimes t^l)\\ &=[h_1,h_2]\otimes t^{m+l}+\frac{1}{n+1}\delta_{m+l,0}(\delta^*(h_1),h_2)c_1+m\delta_{m+l,0}(h_1,h_2)c_2\\ &=[h_1,h_2]\otimes t^{m+l}+m\delta_{m+l,0}(h_1,h_2)c_2. \end{aligned}

  Moreover,

  [d_2,h\otimes t^m]=mh\otimes t^m.

  With a comparison, it is clear that from \frak H(i), only \Bbb Cc_2 is not contained in (\mathfrak{k}(i))_c. Hence, \mathfrak{k}(i)=(\mathfrak{k}(i))_c\oplus \Bbb Cc_2.

• S8. \frak k(i) is tame.
  Assume that x\in (\frak k(i)_c)^\perp. Then x=y+rd_2 when y\in (\frak k(i))c and r\in \Bbb C. If r\neq 0, then

  0=(x,c_2)=(y+rd_2,c_2)=r(d_2,c_2=r,

  which is a contradiction. This implies that x\in (\frak k(i))c, as desired.

We note that the nullity of \frak k(i) is one, according to S5. Therefore,

is an affine Lie algebra up to the central space Z.

It is obtained with appropriate changes in proof [D-G, Proposition 2.17] or repeat the proof of [Yousofzadeh:Finite, weight, 2020, Lemma 5.1].

We know from Table ? and Table ? that if \alpha \in R_{re}\cap R_1, then 2\alpha \in R_{re}\cap R_0. Thus, if \alpha \in R_{re}\cap R_1, then by Theorem 3.8, \alpha\in R^{ln}(V) if and only if 2\alpha\in R^{ln}(V). So, we can assume S\subseteq R_0 without loss of generality. By Proposition 3.13, we have \dot{S}=\{\dot{\alpha}\in \dot R \mid (\dot\alpha+{\mathbb Z}\delta)\cap S\neq \emptyset\} is a finite root system with irreducible components \dot S(1),\ldots, \dot S(k). By the assumption and Remark 3.10 (see Theorem 3.9), we can write

in which “ \uplus ” indicates disjoint union. In the following two cases, we show that K_1\cup \{0\} and K_2\cup \{0\} are symmetric closed subsets of S_{re}\cup \{0\}:

• C1. Let \alpha,\beta \in K_1\cup \{0\}. Then, we prove that \alpha+\beta\in K_1\cup \{0\}. For some N_1 and N_2

  \alpha+\Bbb Z^{\geq N_1}\delta\subseteq R^{ln},~~{\rm and}~~\beta+\Bbb Z^{\geq N_1}\delta\subseteq R^{ln}.

  Set N=N_1+N_2. For n>N, by Theorem 3.8, we have

  \alpha+\beta+n\delta=\underbrace{\alpha+N_1\delta}_{\in R^{ln}}+\underbrace{\beta+(n-N_1)\delta}_{\in R^{ln}}\in R^{ln}.

  Note that \alpha+\beta+n\delta\in R , because S is closed and \alpha+\beta\in S\subseteq R_0. On the other hand, by Remark 3.10 (see Theorem 3.9), -\alpha+\Bbb Z ^{\geq N_3}\subseteq R^{ln} for some N_3. Therefore, K_1\cup \{0\} is a symmetric closed subset of S_{re}\cup \{0\}.

• C2. Let \alpha,\beta \in K_2\cup \{0\}. If \alpha+\beta =0, then there is nothing to prove. Assume that \alpha+\beta \neq 0. Then, for appropriates n and m, \alpha+n\delta,\beta+m\delta \in S_{re}\cap R^{in}\subseteq \frak{C}_M. By Lemma 3.2(ii), \alpha+\beta+(n+m)\delta \in \frak C_M. As S is closed, we get 0\neq \alpha+\beta+(n+m)\delta\in S\subseteq R_0. Hence, \alpha+\beta+(n+m)\delta \in R_{re}, and so \alpha+\beta +(n+m)\delta\in \frak C_M\cap R_{re}=R^{in}. This implies that K_2 is closed. On the other hand, by Remark 3.10 (see Theorem 3.9), -\alpha+\Bbb Z ^{\geq N}\subseteq R^{in} for some N. Thus, K_2\cup \{0\} is a symmetric closed subset of S_{re}\cup \{0\}.

Since S is symmetric and closed, we get \dot S is symmetric and closed. This implies that \dot S(1), \cdots , \dot S(k) are symmetric and closed. A straightforward check shows that S(1), \cdots , S(k) are symmetric and closed. So, for 1\leq i\leq n, S(i)\cap (K_1\cup \{0\}) as well as S(i)\cap (K_2\cup \{0\}) are symmetric closed subsets of S(i)_{re}. Assume that S(i)\cap (K_1\cup \{0\})\neq \emptyset and S(i)\cap (K_2\cup \{0\})\neq \emptyset. Let \alpha\in S(i)\cap (K_1\cup \{0\}) and \beta \in S(i)\cap (K_2\cup \{0\}). If (\alpha,\beta)\neq 0, then by [Yousofzadeh:Extended, affine, 2016, Lemma 3.7], either \alpha-\beta \in R or \alpha+\beta\in R. Suppose that 0\neq \alpha+\beta. As S(i) is closed and S\setminus \{0\}\subseteq R_0\setminus \{0\}\subseteq R_{re}, we get \alpha+\beta\in S(i)_{re}, because S(i) is closed. It follows that \alpha+\beta\in K_1 or \alpha+\beta \in K_2. If \alpha+\beta\in K_1, then \beta=(\alpha+\beta)-\alpha\in K(1) (because K_1 is closed and symmetric), a contradiction. By a similar argument, state \alpha+\beta \in K_2 arrives at a contradiction. If \alpha+\beta=0, then \beta=(\alpha+\beta)-\alpha)\in K_1, which is a contradiction. For the case where \alpha - \beta \in R, we can make a contradiction in a similar way demonstrated above. Thus (\alpha,\beta)=0. So, we can find elements \dot \alpha and \dot \beta in \dot s(i) that have the property (\dot \alpha,\dot \beta)=0. It follows that \dot s(i) is not irreducible, and this is a contradiction. Taking everything into account, we have shown that either S(i)\cap K_1=\emptyset or S(i)\cap K_2=\emptyset. This leads us to the fact that S(i) is up-nilpotent hybrid or down-nilpotent hybrid.
We claim that if k\geq 2 and S(i) is up-nilpotent (resp. down-nilpotent) hybrid for some i, then S(j) is up-nilpotent (resp. down-nilpotent) hybrid for all 1\leq j \leq k. To prove this, fix an i\in \{1, \cdots , k\} and assume that S(i) is an up-nilpotent hybrid (the other case can be proven similarly). Suppose on the contrary that j\neq i and S(j) is a down-nilpotent hybrid subset. By Remark 3.13, there exists a functional f_i on \text{span}_\mathbb{R} S(i) with

and f_i (\delta) > 0. This implies that S(i)^+\cap S_{re}\subseteq S^{ln} and S(i)^-\cap S_{re}\subseteq S^{in}. So, by Lemma 3.14, there exist p\in \mathbb{Z} and \mu \in \text{supp}(M) such that

Let \beta \in S(j)\cap S_{re}. Since S(j) is down-nilpotent hybrid, we choose m > 0 such that

Assume that \mu+\beta -mp\delta \in \text{supp}(M). Then, as -\beta +2mp\delta \in S(j)^{in}, we have

which is a contradiction due to (Proof). Hence,

where \frak{L}(j) is the affine Lie algebra up to a central space corresponding to S(j). On the other hand, since \beta, \beta+2mp\delta \in S(j)\cap S_{re}, the root string property(We refer the reader to [Neher] or [Yousofzadeh:Extended, affine, 2016, Proposition 3.8] for more details.) for the affine root system S(j) implies that 2mp\delta \in S(j). So by (Proof), we have

Suppose that \beta=\dot\beta+k\delta for some k\in \Bbb Z and 0\neq x\in \frak{L}(j)^{\dot\beta}. Then, according to \beta, we can choose h\in\frak H in such a way that \beta(h)=1(To find an appropriate h concerning \beta, use Table ?, Table ?, and the definition of \frak H in Section 2.), and hence

This implies that [\frak{L}(j)^{\beta-mp\delta}, \frak{L}(j)^{2mp\delta}]\neq 0. By the fact that [\frak{L}(j)^{\beta-mp\delta}, \frak{L}(j)^{2mp\delta}]\subseteq \frak{L}(j)^{\beta+mp\delta} and [Yousofzadeh:Extended, affine, 2019, Proposition 2.5], we can conclude that [\frak{L}(j)^{\beta-mp\delta}, \frak{L}(j)^{2mp\delta}]= \frak{L}(j)^{\beta+mp\delta}. Therefore, by (Proof) and (Proof), we have

This contradicts the fact that \beta +mp\delta \in S(j)^{in}. Thus, S(j) is up-nilpotent hybrid, and consequently all real roots of S are up-nilpotent hybrid.

Repeat the proof of [Yousofzadeh:Finite, weight, 2020, Lemma 5.4], and use Theorem 3.8, Proposition 3.3(ii), Remark 3.10 (Theorem 3.9).

By Proposition 3.14, we can assume that S is up-nilpotent hybrid, and so all S(i) are up-nilpotent hybrid. According to Remark 3.13 (in fact, using affine Lie theory), we can suppose that \Pi_i is a base of S(i) as follows:

in which each

is the base of irreducible finite root system \dot S(i) with certain properties, and \theta_i is the highest root of \dot S(i) with respect to B_i. After a reindexing, we may assume that

Assume that

where each r_{j,i} is a positive integer, because each B_i is a base. We now proceed in the following sequence of cases to define a functional \zeta on {\rm span}_{\Bbb R}S (this gives us a triangular decomposition S=S^+\cup S^\circ\cup S^-), and at the end of these cases we show that

It is a straightforward check that \Pi:=\bigcup_{i=1}^k\Pi_i\setminus \{\delta-\theta_2,\cdots, \delta-\theta_k\} is a basis for {\rm span}_{\Bbb R}S. Thus, it is enough to define the effect of the functional \zeta on \Pi. Before defining \zeta, in each of the following cases, we note that \zeta(\delta)>0 by Remark 3.13:

Case 1. \delta-\theta_1, \cdots, \delta-\theta_k\in P\cap -P: Define

Now, for i\neq 1, we have

Case 2. \delta-\theta_1, \cdots, \delta-\theta_k\in P\setminus -P: Define

Now, for i\neq 1, we have

Case 3. \delta-\theta_{1}, \delta-\theta_{l_1}, \cdots, \delta-\theta_{l_p} \in P\setminus-P, and \delta-\theta_{i} \in P\cap -P for all i \in\{1, \cdots ,k\}\setminus\{1,l_1,\cdots, l_p\}: Define

Now, for i\in \{l_1,\cdots,l_p\}, we have

and for i\in \{1,\cdots,k\}\setminus\{l_1,\cdots, l_p\}

Case 4. \delta-\theta_{1}, \delta-\theta_{l_1}, \cdots, \delta-\theta_{l_p} \in P\cap -P, and \delta-\theta_{i} \in P\setminus -P for all i \in\{1, \cdots ,k\}\setminus\{1,l_1,\cdots, l_p\}: Define

Now, for i\in \{1,l_1,\cdots,l_p\}, we have

and for i\in \{1,\cdots,k\}\setminus\{l_1,\cdots, l_p\}

Now, we are ready to show that P=S^+\cup S^\circ. Let \alpha\in S^+\cup S^\circ. So, we can write \alpha by some elements of \Pi with positive integers coefficients, because \Pi is a base for S. However, elements of \Pi belong to P, and P is closed, which implies that \alpha\in P. Thus S^+\cup S^\circ\subseteq P.
Conversely, to show P \subseteq S^+\cup S^\circ, it is enough to prove that P\cap S^-=\emptyset. To see this, assume on the contrary that P\cap S^-\not=\emptyset. Then, we pick \alpha=\sum_{\epsilon \in \Pi}k_\epsilon \epsilon \in P\cap S^-. Since \zeta (\alpha)<0, we get that all k_\epsilon are negative integers. Thus, -\sum k_\epsilon is a positive integer. So, we may pick \alpha such that -\sum k_\epsilon is minimal, i.e. minimal height. Suppose that -\alpha\in \Pi. Then, -\alpha=\epsilon for some \epsilon\in \Pi. As \zeta(\epsilon)=\zeta(-\alpha)>0, we conclude from all the cases defined above for \zeta that \epsilon \in P\setminus -P. On the other hand, -\epsilon=\alpha\in P, and hence \epsilon\in -P. This is a contradiction; consequently, -\alpha \not \in \Pi. This allows us choosing \epsilon_0\in \Pi such that \alpha+\epsilon_0\in S. It is clear that \alpha+\epsilon_0\in S^-\cap P and has a smaller height than \alpha, a contradiction. Thus P\cap S^-=\emptyset and so

A straightforward check proves the rest statements.

Use (?) and apply Lemma 3.14.

Use (?) and apply Proposition 3.15.

Use (?) and apply Proposition 3.17.