Definition 2

A super Poisson bracket { , } fragments normal-{ normal-, normal-} \{,\} { , } on F u n ( W ) ) fragments F u n fragments normal-( W normal-) normal-) Fun(W)) italic_F italic_u italic_n ( italic_W ) ) is a bilinear operation assigning to every pair of functions f , g F u n ( W ) 𝑓 𝑔 𝐹 𝑢 𝑛 𝑊 f,g\in Fun(W) italic_f , italic_g ∈ italic_F italic_u italic_n ( italic_W ) a new function f , g F u n ( W ) 𝑓 𝑔 𝐹 𝑢 𝑛 𝑊 {f,g}\in Fun(W) italic_f , italic_g ∈ italic_F italic_u italic_n ( italic_W ) , such that for homogenous functions satisfies the following conditions:
i-Graded preserving

d e g ( { f , g } = d e g ( f ) + d e g ( g ) fragments d e g fragments ( fragments { f , g } d e g fragments ( f ) d e g fragments ( g ) deg(\{f,g\}=deg(f)+deg(g) italic_d italic_e italic_g ( { italic_f , italic_g } = italic_d italic_e italic_g ( italic_f ) + italic_d italic_e italic_g ( italic_g ) (3)

ii-Super skew-symmetry

{ f , g } = - ( - 1 ) ( d e g ( f ) ) ( d e g ( g ) ) { g , f } 𝑓 𝑔 superscript 1 𝑑 𝑒 𝑔 𝑓 𝑑 𝑒 𝑔 𝑔 𝑔 𝑓 \{f,g\}=-(-1)^{(deg(f))(deg(g))}\{g,f\} { italic_f , italic_g } = - ( - 1 ) start_POSTSUPERSCRIPT ( italic_d italic_e italic_g ( italic_f ) ) ( italic_d italic_e italic_g ( italic_g ) ) end_POSTSUPERSCRIPT { italic_g , italic_f } (4)

iii-Graded Leibniz rule

{ f , g h } = { f , g } h + ( - 1 ) d e g ( f ) d e g ( g ) g { f , h } 𝑓 𝑔 𝑓 𝑔 superscript 1 𝑑 𝑒 𝑔 𝑓 𝑑 𝑒 𝑔 𝑔 𝑔 𝑓 \{f,gh\}=\{f,g\}h+(-1)^{deg(f)deg(g)}g\{f,h\} { italic_f , italic_g italic_h } = { italic_f , italic_g } italic_h + ( - 1 ) start_POSTSUPERSCRIPT italic_d italic_e italic_g ( italic_f ) italic_d italic_e italic_g ( italic_g ) end_POSTSUPERSCRIPT italic_g { italic_f , italic_h } (5)

iv-Super Jocobi identity

( - 1 ) d e g ( f ) d e g ( h ) { f , { g , h } } + ( - 1 ) d e g ( g ) d e g ( f ) { g , { h , f } } + ( - 1 ) d e g ( h ) d e g ( g ) { h , { f , g } } = 0 superscript 1 𝑑 𝑒 𝑔 𝑓 𝑑 𝑒 𝑔 𝑓 𝑔 superscript 1 𝑑 𝑒 𝑔 𝑔 𝑑 𝑒 𝑔 𝑓 𝑔 𝑓 superscript 1 𝑑 𝑒 𝑔 𝑑 𝑒 𝑔 𝑔 𝑓 𝑔 0 (-1)^{deg(f)deg(h)}\{f,\{g,h\}\}+(-1)^{deg(g)deg(f)}\{g,\{h,f\}\}+(-1)^{deg(h)% deg(g)}\{h,\{f,g\}\}=0 ( - 1 ) start_POSTSUPERSCRIPT italic_d italic_e italic_g ( italic_f ) italic_d italic_e italic_g ( italic_h ) end_POSTSUPERSCRIPT { italic_f , { italic_g , italic_h } } + ( - 1 ) start_POSTSUPERSCRIPT italic_d italic_e italic_g ( italic_g ) italic_d italic_e italic_g ( italic_f ) end_POSTSUPERSCRIPT { italic_g , { italic_h , italic_f } } + ( - 1 ) start_POSTSUPERSCRIPT italic_d italic_e italic_g ( italic_h ) italic_d italic_e italic_g ( italic_g ) end_POSTSUPERSCRIPT { italic_h , { italic_f , italic_g } } = 0 (6)

Definition 2.3

Let σ : Σ × Σ Σ normal-: 𝜎 normal-→ superscript normal-Σ normal-∗ superscript normal-Σ normal-∗ superscript normal-Σ normal-∗ \sigma:\Sigma^{\ast}\times\Sigma^{\ast}\rightarrow\Sigma^{\ast} italic_σ : roman_Σ start_POSTSUPERSCRIPT ∗ end_POSTSUPERSCRIPT × roman_Σ start_POSTSUPERSCRIPT ∗ end_POSTSUPERSCRIPT → roman_Σ start_POSTSUPERSCRIPT ∗ end_POSTSUPERSCRIPT be any (possibly nontotal, possibly many-to-one) 2-ary function. We say σ 𝜎 \sigma italic_σ is overstrong if and only if for no f FP 𝑓 normal-FP f\in{\rm FP} italic_f ∈ roman_FP with f : { 1 , 2 } × Σ × Σ Σ × Σ normal-: 𝑓 normal-→ 1 2 superscript normal-Σ normal-∗ superscript normal-Σ normal-∗ superscript normal-Σ normal-∗ superscript normal-Σ normal-∗ f:\{1,2\}\times\Sigma^{\ast}\times\Sigma^{\ast}\rightarrow\Sigma^{\ast}\times% \Sigma^{\ast} italic_f : { 1 , 2 } × roman_Σ start_POSTSUPERSCRIPT ∗ end_POSTSUPERSCRIPT × roman_Σ start_POSTSUPERSCRIPT ∗ end_POSTSUPERSCRIPT → roman_Σ start_POSTSUPERSCRIPT ∗ end_POSTSUPERSCRIPT × roman_Σ start_POSTSUPERSCRIPT ∗ end_POSTSUPERSCRIPT does it hold that for each i { 1 , 2 } 𝑖 1 2 i\in\{1,2\} italic_i ∈ { 1 , 2 } and for all strings z , a Σ 𝑧 𝑎 superscript normal-Σ normal-∗ z,a\in\Sigma^{\ast} italic_z , italic_a ∈ roman_Σ start_POSTSUPERSCRIPT ∗ end_POSTSUPERSCRIPT :

( ( b Σ ) [ ( σ ( a , b ) = z i = 1 ) ( σ ( b , a ) = z i = 2 ) ] ) σ ( f ( i , z , a ) ) = z . fragments fragments ( fragments ( b superscript Σ ) fragments [ fragments ( σ fragments ( a , b ) z i 1 ) fragments ( σ fragments ( b , a ) z i 2 ) ] ) σ fragments ( f fragments ( i , z , a ) ) z . ((\exists b\in\Sigma^{\ast})[(\sigma(a,b)=z{\;\wedge\;}i=1){\;\vee\;}(\sigma(b% ,a)=z{\;\wedge\;}i=2)])\,\Longrightarrow\ \sigma(f(i,z,a))=z. ( ( ∃ italic_b ∈ roman_Σ start_POSTSUPERSCRIPT ∗ end_POSTSUPERSCRIPT ) [ ( italic_σ ( italic_a , italic_b ) = italic_z ∧ italic_i = 1 ) ∨ ( italic_σ ( italic_b , italic_a ) = italic_z ∧ italic_i = 2 ) ] ) ⟹ italic_σ ( italic_f ( italic_i , italic_z , italic_a ) ) = italic_z .

Definition 2.8 .

The algebra R ( SU ( 2 ) ) subscript 𝑅 normal-SU 2 \mathcal{F}_{R}(\mathrm{SU}(2)) caligraphic_F start_POSTSUBSCRIPT italic_R end_POSTSUBSCRIPT ( roman_SU ( 2 ) ) is the 𝔥 𝔥 \mathfrak{h} fraktur_h -Hopf algebroid R ( SL ( 2 ) ) subscript 𝑅 normal-SL 2 \mathcal{F}_{R}(\mathrm{SL}(2)) caligraphic_F start_POSTSUBSCRIPT italic_R end_POSTSUBSCRIPT ( roman_SL ( 2 ) ) equipped with the normal-∗ \ast -structure f ( λ ) = f ¯ ( λ ) 𝑓 superscript 𝜆 normal-∗ normal-¯ 𝑓 𝜆 f(\lambda)^{\ast}=\bar{f}(\lambda) italic_f ( italic_λ ) start_POSTSUPERSCRIPT ∗ end_POSTSUPERSCRIPT = ¯ start_ARG italic_f end_ARG ( italic_λ ) , f ( μ ) = f ¯ ( μ ) 𝑓 superscript 𝜇 normal-∗ normal-¯ 𝑓 𝜇 f(\mu)^{\ast}=\bar{f}(\mu) italic_f ( italic_μ ) start_POSTSUPERSCRIPT ∗ end_POSTSUPERSCRIPT = ¯ start_ARG italic_f end_ARG ( italic_μ ) ,

α = δ , β = - q γ , γ = - q - 1 β , δ = α . formulae-sequence superscript 𝛼 𝛿 formulae-sequence superscript 𝛽 𝑞 𝛾 formulae-sequence superscript 𝛾 superscript 𝑞 1 𝛽 superscript 𝛿 𝛼 \alpha^{\ast}=\delta,\ \ \ \beta^{\ast}=-q\gamma,\ \ \ \gamma^{\ast}=-q^{-1}% \beta,\ \ \ \delta^{\ast}=\alpha. italic_α start_POSTSUPERSCRIPT ∗ end_POSTSUPERSCRIPT = italic_δ , italic_β start_POSTSUPERSCRIPT ∗ end_POSTSUPERSCRIPT = - italic_q italic_γ , italic_γ start_POSTSUPERSCRIPT ∗ end_POSTSUPERSCRIPT = - italic_q start_POSTSUPERSCRIPT - 1 end_POSTSUPERSCRIPT italic_β , italic_δ start_POSTSUPERSCRIPT ∗ end_POSTSUPERSCRIPT = italic_α .

3.5 Definition .

In what follows, we shall consider cut–downs of operators in M 𝑀 M italic_M of the form

b = r b r , superscript 𝑏 𝑟 𝑏 𝑟 b^{\prime}=rbr, italic_b start_POSTSUPERSCRIPT ′ end_POSTSUPERSCRIPT = italic_r italic_b italic_r ,

where r 𝑟 r italic_r is a projection and b 𝑏 b italic_b is self-adjoint. Note that we may view b superscript 𝑏 b^{\prime} italic_b start_POSTSUPERSCRIPT ′ end_POSTSUPERSCRIPT as an element of M 𝑀 M italic_M or the von Neumann algebra r M r 𝑟 𝑀 𝑟 rMr italic_r italic_M italic_r . If s 𝑠 s italic_s is a real number then the cut down spectral interval projections determined by b , r 𝑏 𝑟 b,r italic_b , italic_r and s 𝑠 s italic_s are the spectral projections of b superscript 𝑏 b^{\prime} italic_b start_POSTSUPERSCRIPT ′ end_POSTSUPERSCRIPT determined by the intervals ( - , s ) 𝑠 (-\infty,s) ( - ∞ , italic_s ) and ( - , s ] 𝑠 (-\infty,s] ( - ∞ , italic_s ] . We compute the spectrum and the corresponding spectral projections in the algebra M r subscript 𝑀 𝑟 M_{r} italic_M start_POSTSUBSCRIPT italic_r end_POSTSUBSCRIPT where the projection r 𝑟 r italic_r is the identity.

If ( s 1 , 𝐭 1 ) , ( s 2 , 𝐭 2 ) , , ( s k , 𝐭 k ) subscript 𝑠 1 subscript 𝐭 1 subscript 𝑠 2 subscript 𝐭 2 subscript 𝑠 𝑘 subscript 𝐭 𝑘 (s_{1},\mathbf{t}_{1}),(s_{2},\mathbf{t}_{2}),\dots,(s_{k},\mathbf{t}_{k}) ( italic_s start_POSTSUBSCRIPT 1 end_POSTSUBSCRIPT , bold_t start_POSTSUBSCRIPT 1 end_POSTSUBSCRIPT ) , ( italic_s start_POSTSUBSCRIPT 2 end_POSTSUBSCRIPT , bold_t start_POSTSUBSCRIPT 2 end_POSTSUBSCRIPT ) , … , ( italic_s start_POSTSUBSCRIPT italic_k end_POSTSUBSCRIPT , bold_t start_POSTSUBSCRIPT italic_k end_POSTSUBSCRIPT ) is a sequence of spectral pairs, then its associated facial complex is defined inductively as follows. Write

r 1 = p s 1 , 𝐭 1 + - p s 1 , 𝐭 1 - and b 𝐭 2 = r 1 b 𝐭 2 r 1 . subscript 𝑟 1 superscript subscript 𝑝 subscript 𝑠 1 subscript 𝐭 1 superscript subscript 𝑝 subscript 𝑠 1 subscript 𝐭 1 and superscript subscript 𝑏 subscript 𝐭 2 subscript 𝑟 1 subscript 𝑏 subscript 𝐭 2 subscript 𝑟 1 r_{1}=p_{s_{1},\mathbf{t}_{1}}^{+}-p_{s_{1},\mathbf{t}_{1}}^{-}\text{ and }b_{% \mathbf{t}_{2}}^{\prime}=r_{1}b_{\mathbf{t}_{2}}r_{1}. italic_r start_POSTSUBSCRIPT 1 end_POSTSUBSCRIPT = italic_p start_POSTSUBSCRIPT italic_s start_POSTSUBSCRIPT 1 end_POSTSUBSCRIPT , bold_t start_POSTSUBSCRIPT 1 end_POSTSUBSCRIPT end_POSTSUBSCRIPT start_POSTSUPERSCRIPT + end_POSTSUPERSCRIPT - italic_p start_POSTSUBSCRIPT italic_s start_POSTSUBSCRIPT 1 end_POSTSUBSCRIPT , bold_t start_POSTSUBSCRIPT 1 end_POSTSUBSCRIPT end_POSTSUBSCRIPT start_POSTSUPERSCRIPT - end_POSTSUPERSCRIPT and italic_b start_POSTSUBSCRIPT bold_t start_POSTSUBSCRIPT 2 end_POSTSUBSCRIPT end_POSTSUBSCRIPT start_POSTSUPERSCRIPT ′ end_POSTSUPERSCRIPT = italic_r start_POSTSUBSCRIPT 1 end_POSTSUBSCRIPT italic_b start_POSTSUBSCRIPT bold_t start_POSTSUBSCRIPT 2 end_POSTSUBSCRIPT end_POSTSUBSCRIPT italic_r start_POSTSUBSCRIPT 1 end_POSTSUBSCRIPT .

Let q s 2 , 𝐭 2 ± superscript subscript 𝑞 subscript 𝑠 2 subscript 𝐭 2 plus-or-minus q_{s_{2},\mathbf{t}_{2}}^{\pm} italic_q start_POSTSUBSCRIPT italic_s start_POSTSUBSCRIPT 2 end_POSTSUBSCRIPT , bold_t start_POSTSUBSCRIPT 2 end_POSTSUBSCRIPT end_POSTSUBSCRIPT start_POSTSUPERSCRIPT ± end_POSTSUPERSCRIPT denote the associated cut down spectral interval projections determined by b 𝐭 2 , r 1 subscript 𝑏 subscript 𝐭 2 subscript 𝑟 1 b_{\mathbf{t}_{2}},r_{1} italic_b start_POSTSUBSCRIPT bold_t start_POSTSUBSCRIPT 2 end_POSTSUBSCRIPT end_POSTSUBSCRIPT , italic_r start_POSTSUBSCRIPT 1 end_POSTSUBSCRIPT and s 2 subscript 𝑠 2 s_{2} italic_s start_POSTSUBSCRIPT 2 end_POSTSUBSCRIPT and write

r 2 = q s 2 , 𝐭 2 + - q s 2 , 𝐭 2 - . subscript 𝑟 2 superscript subscript 𝑞 subscript 𝑠 2 subscript 𝐭 2 superscript subscript 𝑞 subscript 𝑠 2 subscript 𝐭 2 r_{2}=q_{s_{2},\mathbf{t}_{2}}^{+}-q_{s_{2},\mathbf{t}_{2}}^{-}. italic_r start_POSTSUBSCRIPT 2 end_POSTSUBSCRIPT = italic_q start_POSTSUBSCRIPT italic_s start_POSTSUBSCRIPT 2 end_POSTSUBSCRIPT , bold_t start_POSTSUBSCRIPT 2 end_POSTSUBSCRIPT end_POSTSUBSCRIPT start_POSTSUPERSCRIPT + end_POSTSUPERSCRIPT - italic_q start_POSTSUBSCRIPT italic_s start_POSTSUBSCRIPT 2 end_POSTSUBSCRIPT , bold_t start_POSTSUBSCRIPT 2 end_POSTSUBSCRIPT end_POSTSUBSCRIPT start_POSTSUPERSCRIPT - end_POSTSUPERSCRIPT .

Now suppose that for some 1 < i < k 1 𝑖 𝑘 1<i<k 1 < italic_i < italic_k we have defined sequences

r 1 > r 2 > > r i , subscript 𝑟 1 subscript 𝑟 2 subscript 𝑟 𝑖 \displaystyle r_{1}>r_{2}>\cdots>r_{i}, italic_r start_POSTSUBSCRIPT 1 end_POSTSUBSCRIPT > italic_r start_POSTSUBSCRIPT 2 end_POSTSUBSCRIPT > ⋯ > italic_r start_POSTSUBSCRIPT italic_i end_POSTSUBSCRIPT ,
b 𝐭 2 , , b 𝐭 i , and superscript subscript 𝑏 subscript 𝐭 2 superscript subscript 𝑏 subscript 𝐭 𝑖 and \displaystyle b_{\mathbf{t}_{2}}^{\prime},\dots,b_{\mathbf{t}_{i}}^{\prime},% \text{ and } italic_b start_POSTSUBSCRIPT bold_t start_POSTSUBSCRIPT 2 end_POSTSUBSCRIPT end_POSTSUBSCRIPT start_POSTSUPERSCRIPT ′ end_POSTSUPERSCRIPT , … , italic_b start_POSTSUBSCRIPT bold_t start_POSTSUBSCRIPT italic_i end_POSTSUBSCRIPT end_POSTSUBSCRIPT start_POSTSUPERSCRIPT ′ end_POSTSUPERSCRIPT , and
q s 2 , 𝐭 2 ± , , q s i , 𝐭 i ± superscript subscript 𝑞 subscript 𝑠 2 subscript 𝐭 2 plus-or-minus superscript subscript 𝑞 subscript 𝑠 𝑖 subscript 𝐭 𝑖 plus-or-minus \displaystyle q_{s_{2},\mathbf{t}_{2}}^{\pm},\dots,q_{s_{i},\mathbf{t}_{i}}^{\pm} italic_q start_POSTSUBSCRIPT italic_s start_POSTSUBSCRIPT 2 end_POSTSUBSCRIPT , bold_t start_POSTSUBSCRIPT 2 end_POSTSUBSCRIPT end_POSTSUBSCRIPT start_POSTSUPERSCRIPT ± end_POSTSUPERSCRIPT , … , italic_q start_POSTSUBSCRIPT italic_s start_POSTSUBSCRIPT italic_i end_POSTSUBSCRIPT , bold_t start_POSTSUBSCRIPT italic_i end_POSTSUBSCRIPT end_POSTSUBSCRIPT start_POSTSUPERSCRIPT ± end_POSTSUPERSCRIPT

such that if 2 j i 2 𝑗 𝑖 2\leq j\leq i 2 ≤ italic_j ≤ italic_i , then

b 𝐭 j = r j - 1 b 𝐭 j r j - 1 , r j = q s j , 𝐭 j + - q s j , 𝐭 j - formulae-sequence superscript subscript 𝑏 subscript 𝐭 𝑗 subscript 𝑟 𝑗 1 subscript 𝑏 subscript 𝐭 𝑗 subscript 𝑟 𝑗 1 subscript 𝑟 𝑗 superscript subscript 𝑞 subscript 𝑠 𝑗 subscript 𝐭 𝑗 superscript subscript 𝑞 subscript 𝑠 𝑗 subscript 𝐭 𝑗 b_{\mathbf{t}_{j}}^{\prime}=r_{j-1}b_{\mathbf{t}_{j}}r_{j-1},\quad r_{j}=q_{s_% {j},\mathbf{t}_{j}}^{+}-q_{s_{j},\mathbf{t}_{j}}^{-} italic_b start_POSTSUBSCRIPT bold_t start_POSTSUBSCRIPT italic_j end_POSTSUBSCRIPT end_POSTSUBSCRIPT start_POSTSUPERSCRIPT ′ end_POSTSUPERSCRIPT = italic_r start_POSTSUBSCRIPT italic_j - 1 end_POSTSUBSCRIPT italic_b start_POSTSUBSCRIPT bold_t start_POSTSUBSCRIPT italic_j end_POSTSUBSCRIPT end_POSTSUBSCRIPT italic_r start_POSTSUBSCRIPT italic_j - 1 end_POSTSUBSCRIPT , italic_r start_POSTSUBSCRIPT italic_j end_POSTSUBSCRIPT = italic_q start_POSTSUBSCRIPT italic_s start_POSTSUBSCRIPT italic_j end_POSTSUBSCRIPT , bold_t start_POSTSUBSCRIPT italic_j end_POSTSUBSCRIPT end_POSTSUBSCRIPT start_POSTSUPERSCRIPT + end_POSTSUPERSCRIPT - italic_q start_POSTSUBSCRIPT italic_s start_POSTSUBSCRIPT italic_j end_POSTSUBSCRIPT , bold_t start_POSTSUBSCRIPT italic_j end_POSTSUBSCRIPT end_POSTSUBSCRIPT start_POSTSUPERSCRIPT - end_POSTSUPERSCRIPT

where q s j , 𝐭 j ± superscript subscript 𝑞 subscript 𝑠 𝑗 subscript 𝐭 𝑗 plus-or-minus q_{s_{j},\mathbf{t}_{j}}^{\pm} italic_q start_POSTSUBSCRIPT italic_s start_POSTSUBSCRIPT italic_j end_POSTSUBSCRIPT , bold_t start_POSTSUBSCRIPT italic_j end_POSTSUBSCRIPT end_POSTSUBSCRIPT start_POSTSUPERSCRIPT ± end_POSTSUPERSCRIPT are the cut down spectral interval projections determined by b 𝐭 j , r j - 1 subscript 𝑏 subscript 𝐭 𝑗 subscript 𝑟 𝑗 1 b_{\mathbf{t}_{j}},r_{j-1} italic_b start_POSTSUBSCRIPT bold_t start_POSTSUBSCRIPT italic_j end_POSTSUBSCRIPT end_POSTSUBSCRIPT , italic_r start_POSTSUBSCRIPT italic_j - 1 end_POSTSUBSCRIPT and s j subscript 𝑠 𝑗 s_{j} italic_s start_POSTSUBSCRIPT italic_j end_POSTSUBSCRIPT .

We may then set

b 𝐭 i + 1 = r i b 𝐭 i + 1 r i , superscript subscript 𝑏 subscript 𝐭 𝑖 1 subscript 𝑟 𝑖 subscript 𝑏 subscript 𝐭 𝑖 1 subscript 𝑟 𝑖 b_{\mathbf{t}_{i+1}}^{\prime}=r_{i}b_{\mathbf{t}_{i+1}}r_{i}, italic_b start_POSTSUBSCRIPT bold_t start_POSTSUBSCRIPT italic_i + 1 end_POSTSUBSCRIPT end_POSTSUBSCRIPT start_POSTSUPERSCRIPT ′ end_POSTSUPERSCRIPT = italic_r start_POSTSUBSCRIPT italic_i end_POSTSUBSCRIPT italic_b start_POSTSUBSCRIPT bold_t start_POSTSUBSCRIPT italic_i + 1 end_POSTSUBSCRIPT end_POSTSUBSCRIPT italic_r start_POSTSUBSCRIPT italic_i end_POSTSUBSCRIPT ,

let q s i + 1 , 𝐭 i + 1 ± superscript subscript 𝑞 subscript 𝑠 𝑖 1 subscript 𝐭 𝑖 1 plus-or-minus q_{s_{i+1},\mathbf{t}_{i+1}}^{\pm} italic_q start_POSTSUBSCRIPT italic_s start_POSTSUBSCRIPT italic_i + 1 end_POSTSUBSCRIPT , bold_t start_POSTSUBSCRIPT italic_i + 1 end_POSTSUBSCRIPT end_POSTSUBSCRIPT start_POSTSUPERSCRIPT ± end_POSTSUPERSCRIPT denote the spectral interval projections determined by b 𝐭 i + 1 , r i subscript 𝑏 subscript 𝐭 𝑖 1 subscript 𝑟 𝑖 b_{\mathbf{t}_{i+1}},r_{i} italic_b start_POSTSUBSCRIPT bold_t start_POSTSUBSCRIPT italic_i + 1 end_POSTSUBSCRIPT end_POSTSUBSCRIPT , italic_r start_POSTSUBSCRIPT italic_i end_POSTSUBSCRIPT and s i + 1 subscript 𝑠 𝑖 1 s_{i+1} italic_s start_POSTSUBSCRIPT italic_i + 1 end_POSTSUBSCRIPT and write

r i + 1 = q s i + 1 , 𝐭 i + 1 + - q s i + 1 , 𝐭 i + 1 - . subscript 𝑟 𝑖 1 superscript subscript 𝑞 subscript 𝑠 𝑖 1 subscript 𝐭 𝑖 1 superscript subscript 𝑞 subscript 𝑠 𝑖 1 subscript 𝐭 𝑖 1 r_{i+1}=q_{s_{i+1},\mathbf{t}_{i+1}}^{+}-q_{s_{i+1},\mathbf{t}_{i+1}}^{-}. italic_r start_POSTSUBSCRIPT italic_i + 1 end_POSTSUBSCRIPT = italic_q start_POSTSUBSCRIPT italic_s start_POSTSUBSCRIPT italic_i + 1 end_POSTSUBSCRIPT , bold_t start_POSTSUBSCRIPT italic_i + 1 end_POSTSUBSCRIPT end_POSTSUBSCRIPT start_POSTSUPERSCRIPT + end_POSTSUPERSCRIPT - italic_q start_POSTSUBSCRIPT italic_s start_POSTSUBSCRIPT italic_i + 1 end_POSTSUBSCRIPT , bold_t start_POSTSUBSCRIPT italic_i + 1 end_POSTSUBSCRIPT end_POSTSUBSCRIPT start_POSTSUPERSCRIPT - end_POSTSUPERSCRIPT .

This completes the induction.

Thus the facial complex determined by the spectral pairs ( s 1 , 𝐭 1 ) , , ( s k , 𝐭 k ) subscript 𝑠 1 subscript 𝐭 1 subscript 𝑠 𝑘 subscript 𝐭 𝑘 (s_{1},\mathbf{t}_{1}),\dots,(s_{k},\mathbf{t}_{k}) ( italic_s start_POSTSUBSCRIPT 1 end_POSTSUBSCRIPT , bold_t start_POSTSUBSCRIPT 1 end_POSTSUBSCRIPT ) , … , ( italic_s start_POSTSUBSCRIPT italic_k end_POSTSUBSCRIPT , bold_t start_POSTSUBSCRIPT italic_k end_POSTSUBSCRIPT ) consists of the projections p s 1 , 𝐭 1 ± superscript subscript 𝑝 subscript 𝑠 1 subscript 𝐭 1 plus-or-minus p_{s_{1},\mathbf{t}_{1}}^{\pm} italic_p start_POSTSUBSCRIPT italic_s start_POSTSUBSCRIPT 1 end_POSTSUBSCRIPT , bold_t start_POSTSUBSCRIPT 1 end_POSTSUBSCRIPT end_POSTSUBSCRIPT start_POSTSUPERSCRIPT ± end_POSTSUPERSCRIPT together with the sequences

r 1 > > r k - 1 , subscript 𝑟 1 subscript 𝑟 𝑘 1 \displaystyle r_{1}>\cdots>r_{k-1}, italic_r start_POSTSUBSCRIPT 1 end_POSTSUBSCRIPT > ⋯ > italic_r start_POSTSUBSCRIPT italic_k - 1 end_POSTSUBSCRIPT ,
b 𝐭 2 , , b 𝐭 k , and superscript subscript 𝑏 subscript 𝐭 2 superscript subscript 𝑏 subscript 𝐭 𝑘 and \displaystyle b_{\mathbf{t}_{2}}^{\prime},\dots,b_{\mathbf{t}_{k}}^{\prime},% \text{ and } italic_b start_POSTSUBSCRIPT bold_t start_POSTSUBSCRIPT 2 end_POSTSUBSCRIPT end_POSTSUBSCRIPT start_POSTSUPERSCRIPT ′ end_POSTSUPERSCRIPT , … , italic_b start_POSTSUBSCRIPT bold_t start_POSTSUBSCRIPT italic_k end_POSTSUBSCRIPT end_POSTSUBSCRIPT start_POSTSUPERSCRIPT ′ end_POSTSUPERSCRIPT , and
q s 2 , 𝐭 2 ± , , q s k , 𝐭 k ± superscript subscript 𝑞 subscript 𝑠 2 subscript 𝐭 2 plus-or-minus superscript subscript 𝑞 subscript 𝑠 𝑘 subscript 𝐭 𝑘 plus-or-minus \displaystyle q_{s_{2},\mathbf{t}_{2}}^{\pm},\dots,q_{s_{k},\mathbf{t}_{k}}^{\pm} italic_q start_POSTSUBSCRIPT italic_s start_POSTSUBSCRIPT 2 end_POSTSUBSCRIPT , bold_t start_POSTSUBSCRIPT 2 end_POSTSUBSCRIPT end_POSTSUBSCRIPT start_POSTSUPERSCRIPT ± end_POSTSUPERSCRIPT , … , italic_q start_POSTSUBSCRIPT italic_s start_POSTSUBSCRIPT italic_k end_POSTSUBSCRIPT , bold_t start_POSTSUBSCRIPT italic_k end_POSTSUBSCRIPT end_POSTSUBSCRIPT start_POSTSUPERSCRIPT ± end_POSTSUPERSCRIPT

as defined above.