Definition 1.1 .

Let R 𝑅 R italic_R be a ring, σ 𝜎 \sigma italic_σ an endomorphism of R 𝑅 R italic_R and δ 𝛿 \delta italic_δ a σ 𝜎 \sigma italic_σ -derivation. The Ore extension R [ x ; σ , δ ] 𝑅 𝑥 𝜎 𝛿 R[x;\sigma,\delta] italic_R [ italic_x ; italic_σ , italic_δ ] is defined as the ring generated by R 𝑅 R italic_R and an element x R 𝑥 𝑅 x\notin R italic_x ∉ italic_R such that 1 , x , x 2 , 1 𝑥 superscript 𝑥 2 1,x,x^{2},\ldots 1 , italic_x , italic_x start_POSTSUPERSCRIPT 2 end_POSTSUPERSCRIPT , … form a basis for R [ x ; σ , δ ] 𝑅 𝑥 𝜎 𝛿 R[x;\sigma,\delta] italic_R [ italic_x ; italic_σ , italic_δ ] as a left R 𝑅 R italic_R -module and all r R 𝑟 𝑅 r\in R italic_r ∈ italic_R satisfy

x r = σ ( r ) x + δ ( r ) . 𝑥 𝑟 𝜎 𝑟 𝑥 𝛿 𝑟 xr=\sigma(r)x+\delta(r). italic_x italic_r = italic_σ ( italic_r ) italic_x + italic_δ ( italic_r ) . (1)

Definition 3

For a metric compatible connection normal-∇ \nabla , recalling the definition of the torsion operator 8 8 8 The torsion operator of a connection normal-∇ \nabla is the mapping τ : sec T M × sec T M sec T M normal-: 𝜏 normal-→ 𝑇 𝑀 𝑇 𝑀 𝑇 𝑀 \mathbf{\tau}:\sec TM\times\sec TM\rightarrow\sec TM italic_τ : roman_sec italic_T italic_M × roman_sec italic_T italic_M → roman_sec italic_T italic_M , ( 𝐮 , 𝐯 ) τ ( 𝐮 , 𝐯 ) = 𝐮 𝐯 - v 𝐯 - [ 𝐮 , 𝐯 ] maps-to 𝐮 𝐯 𝜏 𝐮 𝐯 subscript normal-∇ 𝐮 𝐯 subscript normal-∇ 𝑣 𝐯 𝐮 𝐯 (\boldsymbol{u,v})\mapsto\mathbf{\tau}(\boldsymbol{u,v})=\nabla_{\boldsymbol{u% }}\boldsymbol{v}-\nabla_{v}\boldsymbol{v}-[\boldsymbol{u,v}] ( bold_italic_u bold_, bold_italic_v ) ↦ italic_τ ( bold_italic_u bold_, bold_italic_v ) = ∇ start_POSTSUBSCRIPT bold_italic_u end_POSTSUBSCRIPT bold_italic_v - ∇ start_POSTSUBSCRIPT italic_v end_POSTSUBSCRIPT bold_italic_v - [ bold_italic_u bold_, bold_italic_v ] . we conveniently write

τ ( u , v ) = [[[ u , v ]]] , 𝜏 𝑢 𝑣 𝑢 𝑣 \mathbf{\tau(}u,v)=\mathopen{[\kern-2.2pt[\kern-2.3pt[}u,v\mathclose{]\kern-2.% 1pt]\kern-2.3pt]}, italic_τ ( italic_u , italic_v ) = italic_[[[ italic_u , italic_v italic_]]] , (15)

which we call the ( form ) torsion operator.


Definition 1

The operators ( a , b ) 𝑎 𝑏 (a,b) ( italic_a , italic_b ) are 𝒟 𝒟 {\mathcal{D}} caligraphic_D -pseudo bosonic ( 𝒟 𝒟 {\mathcal{D}} caligraphic_D -pb) if, for all f 𝒟 𝑓 𝒟 f\in{\mathcal{D}} italic_f ∈ caligraphic_D , we have

a b f - b a f = f . 𝑎 𝑏 𝑓 𝑏 𝑎 𝑓 𝑓 a\,b\,f-b\,a\,f=f. italic_a italic_b italic_f - italic_b italic_a italic_f = italic_f . (2.1)

Definition 1

The operators ( a , b ) 𝑎 𝑏 (a,b) ( italic_a , italic_b ) are 𝒟 𝒟 {\mathcal{D}} caligraphic_D -pseudo bosonic ( 𝒟 𝒟 {\mathcal{D}} caligraphic_D -pb) if, for all f 𝒟 𝑓 𝒟 f\in{\mathcal{D}} italic_f ∈ caligraphic_D , we have

a b f - b a f = f . 𝑎 𝑏 𝑓 𝑏 𝑎 𝑓 𝑓 a\,b\,f-b\,a\,f=f. italic_a italic_b italic_f - italic_b italic_a italic_f = italic_f . (2.2)

Definition 2.3 (Consistency) .

Consider two labeled graphs G 𝐺 G italic_G and H 𝐻 H italic_H in 𝒢 π , Σ subscript 𝒢 𝜋 normal-Σ {\cal G}_{\pi,\Sigma} caligraphic_G start_POSTSUBSCRIPT italic_π , roman_Σ end_POSTSUBSCRIPT , they are consistent if and only if for all vertices u , v , w 𝑢 𝑣 𝑤 u,v,w italic_u , italic_v , italic_w and ports k , l , p 𝑘 𝑙 𝑝 k,l,p italic_k , italic_l , italic_p we have:

{ u : k , v : l } E ( G ) { u : k , w : p } E ( H ) v = w l = p conditional-set 𝑢 : 𝑘 𝑣 𝑙 𝐸 𝐺 conditional-set 𝑢 : 𝑘 𝑤 𝑝 𝐸 𝐻 𝑣 𝑤 𝑙 𝑝 \{u:k,v:l\}\in E(G)\wedge\{u:k,w:p\}\in E(H)\Rightarrow v=w\wedge l=p { italic_u : italic_k , italic_v : italic_l } ∈ italic_E ( italic_G ) ∧ { italic_u : italic_k , italic_w : italic_p } ∈ italic_E ( italic_H ) ⇒ italic_v = italic_w ∧ italic_l = italic_p

and for all vertices u V ( G ) V ( H ) 𝑢 𝑉 𝐺 𝑉 𝐻 u\in V(G)\cap V(H) italic_u ∈ italic_V ( italic_G ) ∩ italic_V ( italic_H ) we have that σ G ( u ) = σ H ( u ) subscript 𝜎 𝐺 𝑢 subscript 𝜎 𝐻 𝑢 \sigma_{G}(u)=\sigma_{H}(u) italic_σ start_POSTSUBSCRIPT italic_G end_POSTSUBSCRIPT ( italic_u ) = italic_σ start_POSTSUBSCRIPT italic_H end_POSTSUBSCRIPT ( italic_u ) . Two graphs are trivially consistent if V ( G ) V ( H ) 𝑉 𝐺 𝑉 𝐻 V(G)\cap V(H) italic_V ( italic_G ) ∩ italic_V ( italic_H ) is empty.


Definition 1.2 .

We say that a function F : X * Y : 𝐹 superscript 𝑋 𝑌 F\colon X^{*}\to Y italic_F : italic_X start_POSTSUPERSCRIPT * end_POSTSUPERSCRIPT → italic_Y is preassociative if for every 𝐱𝐲𝐲 𝐳 X * superscript 𝐱𝐲𝐲 𝐳 superscript 𝑋 \mathbf{x}\mathbf{y}\mathbf{y}^{\prime}\mathbf{z}\in X^{*} bold_xyy start_POSTSUPERSCRIPT ′ end_POSTSUPERSCRIPT bold_z ∈ italic_X start_POSTSUPERSCRIPT * end_POSTSUPERSCRIPT we have

F ( 𝐲 ) = F ( 𝐲 ) F ( 𝐱𝐲𝐳 ) = F ( 𝐱𝐲 𝐳 ) . formulae-sequence 𝐹 𝐲 𝐹 superscript 𝐲 𝐹 𝐱𝐲𝐳 𝐹 superscript 𝐱𝐲 𝐳 F(\mathbf{y})~{}=~{}F(\mathbf{y}^{\prime})\quad\Rightarrow\quad F(\mathbf{x}% \mathbf{y}\mathbf{z})~{}=~{}F(\mathbf{x}\mathbf{y}^{\prime}\mathbf{z}). italic_F ( bold_y ) = italic_F ( bold_y start_POSTSUPERSCRIPT ′ end_POSTSUPERSCRIPT ) ⇒ italic_F ( bold_xyz ) = italic_F ( bold_xy start_POSTSUPERSCRIPT ′ end_POSTSUPERSCRIPT bold_z ) .
Definition 4.15 .

We say that a function F : X * Y : 𝐹 superscript 𝑋 𝑌 F\colon X^{*}\to Y italic_F : italic_X start_POSTSUPERSCRIPT * end_POSTSUPERSCRIPT → italic_Y is strongly preassociative if for every 𝐱𝐱 𝐲𝐳𝐳 X * superscript 𝐱𝐱 superscript 𝐲𝐳𝐳 superscript 𝑋 \mathbf{x}\mathbf{x}^{\prime}\mathbf{y}\mathbf{z}\mathbf{z}^{\prime}\in X^{*} bold_xx start_POSTSUPERSCRIPT ′ end_POSTSUPERSCRIPT bold_yzz start_POSTSUPERSCRIPT ′ end_POSTSUPERSCRIPT ∈ italic_X start_POSTSUPERSCRIPT * end_POSTSUPERSCRIPT we have

(5) F ( 𝐱𝐳 ) = F ( 𝐱 𝐳 ) F ( 𝐱𝐲𝐳 ) = F ( 𝐱 𝐲𝐳 ) . formulae-sequence 𝐹 𝐱𝐳 𝐹 superscript 𝐱 superscript 𝐳 𝐹 𝐱𝐲𝐳 𝐹 superscript 𝐱 superscript 𝐲𝐳 F(\mathbf{x}\mathbf{z})~{}=~{}F(\mathbf{x}^{\prime}\mathbf{z}^{\prime})\quad% \Rightarrow\quad F(\mathbf{x}\mathbf{y}\mathbf{z})~{}=~{}F(\mathbf{x}^{\prime}% \mathbf{y}\mathbf{z}^{\prime}). italic_F ( bold_xz ) = italic_F ( bold_x start_POSTSUPERSCRIPT ′ end_POSTSUPERSCRIPT bold_z start_POSTSUPERSCRIPT ′ end_POSTSUPERSCRIPT ) ⇒ italic_F ( bold_xyz ) = italic_F ( bold_x start_POSTSUPERSCRIPT ′ end_POSTSUPERSCRIPT bold_yz start_POSTSUPERSCRIPT ′ end_POSTSUPERSCRIPT ) .

Definition 3.1 .

Let σ : 𝒫 𝔪 𝒫 𝔪 : 𝜎 superscript 𝒫 𝔪 superscript 𝒫 𝔪 \sigma:\mathcal{P}^{\mathfrak{m}}\longrightarrow\mathcal{P}^{\mathfrak{m}} italic_σ : caligraphic_P start_POSTSUPERSCRIPT fraktur_m end_POSTSUPERSCRIPT ⟶ caligraphic_P start_POSTSUPERSCRIPT fraktur_m end_POSTSUPERSCRIPT be the involution

(3.1) σ = τ ( - 1 ) c 𝜎 𝜏 1 𝑐 \sigma=\tau(-1)\,c italic_σ = italic_τ ( - 1 ) italic_c

where c 𝑐 c italic_c is the real Frobenius of § 2.10 . For example, σ ( 𝕃 𝔪 ) = 𝕃 𝔪 𝜎 superscript 𝕃 𝔪 superscript 𝕃 𝔪 \sigma({\mathbb{L}}^{\mathfrak{m}})={\mathbb{L}}^{\mathfrak{m}} italic_σ ( blackboard_L start_POSTSUPERSCRIPT fraktur_m end_POSTSUPERSCRIPT ) = blackboard_L start_POSTSUPERSCRIPT fraktur_m end_POSTSUPERSCRIPT .

Definition 3.3 .

Define sv 𝔪 superscript sv 𝔪 \mathrm{sv}^{\mathfrak{m}} roman_sv start_POSTSUPERSCRIPT fraktur_m end_POSTSUPERSCRIPT to be the unique element of G d R ( 𝒫 𝔪 ) subscript 𝐺 𝑑 𝑅 superscript 𝒫 𝔪 G_{dR}(\mathcal{P}^{\mathfrak{m}}) italic_G start_POSTSUBSCRIPT italic_d italic_R end_POSTSUBSCRIPT ( caligraphic_P start_POSTSUPERSCRIPT fraktur_m end_POSTSUPERSCRIPT ) such that

(3.2) sv 𝔪 σ = id . superscript sv 𝔪 𝜎 id \mathrm{sv}^{\mathfrak{m}}\circ\sigma=\mathrm{id}\ . roman_sv start_POSTSUPERSCRIPT fraktur_m end_POSTSUPERSCRIPT ∘ italic_σ = roman_id .

Definition 4.3 ( Type1 , Type2 hexagon R 𝑅 R italic_R ) .

Let F M ( S , ϕ ) 𝐹 subscript 𝑀 𝑆 italic-ϕ F\subset M_{(S,\phi)} italic_F ⊂ italic_M start_POSTSUBSCRIPT ( italic_S , italic_ϕ ) end_POSTSUBSCRIPT be a surface admitting an open book foliation. Suppose that F 𝐹 F italic_F contains a hexagon region R 𝑅 R italic_R consisting of two bb-tiles of opposite signs meeting along a b-arc as in Sketch (1) of Figure 11 . We name the vertices (elliptic points) A , B , C , D , E , F 𝐴 𝐵 𝐶 𝐷 𝐸 𝐹 A,B,C,D,E,F italic_A , italic_B , italic_C , italic_D , italic_E , italic_F counterclockwise. We may assume that 𝚜𝚐𝚗 ( A ) = 𝚜𝚐𝚗 ( C ) = 𝚜𝚐𝚗 ( E ) = + 1 𝚜𝚐𝚗 𝐴 𝚜𝚐𝚗 𝐶 𝚜𝚐𝚗 𝐸 1 {\tt sgn}(A)={\tt sgn}(C)={\tt sgn}(E)=+1 typewriter_sgn ( italic_A ) = typewriter_sgn ( italic_C ) = typewriter_sgn ( italic_E ) = + 1 and 𝚜𝚐𝚗 ( B ) = 𝚜𝚐𝚗 ( D ) = 𝚜𝚐𝚗 ( F ) = - 1 𝚜𝚐𝚗 𝐵 𝚜𝚐𝚗 𝐷 𝚜𝚐𝚗 𝐹 1 {\tt sgn}(B)={\tt sgn}(D)={\tt sgn}(F)=-1 typewriter_sgn ( italic_B ) = typewriter_sgn ( italic_D ) = typewriter_sgn ( italic_F ) = - 1 . We require that the boundary b-arcs A B ¯ , C D ¯ , E F ¯ ¯ 𝐴 𝐵 ¯ 𝐶 𝐷 ¯ 𝐸 𝐹 \overline{AB},\overline{CD},\overline{EF} ¯ start_ARG italic_A italic_B end_ARG , ¯ start_ARG italic_C italic_D end_ARG , ¯ start_ARG italic_E italic_F end_ARG lie on the same page of the open book, and likewise B C ¯ , D E ¯ , F A ¯ ¯ 𝐵 𝐶 ¯ 𝐷 𝐸 ¯ 𝐹 𝐴 \overline{BC},\overline{DE},\overline{FA} ¯ start_ARG italic_B italic_C end_ARG , ¯ start_ARG italic_D italic_E end_ARG , ¯ start_ARG italic_F italic_A end_ARG lie on another same page.


(1) Original hexagon (1) hexagon R 𝑅 R italic_R A 𝐴 A italic_A B 𝐵 B italic_B 𝐩 𝐩 {\bf p} bold_p 𝐪 𝐪 {\bf q} bold_q C 𝐶 C italic_C D 𝐷 D italic_D E 𝐸 E italic_E F 𝐹 F italic_F retrograde prograde (2) A 𝐴 A italic_A B 𝐵 B italic_B C 𝐶 C italic_C D 𝐷 D italic_D E 𝐸 E italic_E F 𝐹 F italic_F (3) A 𝐴 A italic_A B 𝐵 B italic_B C 𝐶 C italic_C D 𝐷 D italic_D E 𝐸 E italic_E F 𝐹 F italic_F

Figure 11. (1) Original hexagon R 𝑅 R italic_R . (2) Hexagon after retrograde bypass move. (3) Hexagon after prograde bypass move. Dashed arcs are dividing sets. Shaded regions are negative regions and unshaded regions are positive regions.

Let 𝐩 , 𝐪 𝐩 𝐪 {\bf p},{\bf q} bold_p , bold_q denote the two hyperbolic points of R 𝑅 R italic_R . From now on we assume that

𝚜𝚐𝚗 ( 𝐩 ) = + 1 , 𝚜𝚐𝚗 ( 𝐪 ) = - 1 . formulae-sequence 𝚜𝚐𝚗 𝐩 1 𝚜𝚐𝚗 𝐪 1 {\tt sgn}({\bf p})=+1,\qquad{\tt sgn}({\bf q})=-1. typewriter_sgn ( bold_p ) = + 1 , typewriter_sgn ( bold_q ) = - 1 .

(If 𝚜𝚐𝚗 ( 𝐩 ) = - 1 𝚜𝚐𝚗 𝐩 1 {\tt sgn}({\bf p})=-1 typewriter_sgn ( bold_p ) = - 1 , 𝚜𝚐𝚗 ( 𝐪 ) = + 1 𝚜𝚐𝚗 𝐪 1 {\tt sgn}({\bf q})=+1 typewriter_sgn ( bold_q ) = + 1 similar statements hold.) With this sign assumption there are two possible movie presentations realizing the open book foliation o b ( R ) subscript 𝑜 𝑏 𝑅 \mathcal{F}_{ob}(R) caligraphic_F start_POSTSUBSCRIPT italic_o italic_b end_POSTSUBSCRIPT ( italic_R ) on R 𝑅 R italic_R . See Figure 12 . We call them Type1 and Type2 .

Movie presentations of Type1 Type2 𝐩 𝐩 \bf p bold_p 𝐪 𝐪 \bf q bold_q 𝐪 𝐪 \bf q bold_q 𝐩 𝐩 \bf p bold_p

Figure 12. Movie presentations of Type1 and Type2 hexagon R 𝑅 R italic_R .

Definition 2.6 .

[ 28 ] If A 𝐴 A italic_A is an MV-algebra then a function s : A [ 0 , 1 ] normal-: 𝑠 normal-→ 𝐴 0 1 s:A\to[0,1] italic_s : italic_A → [ 0 , 1 ] is a state if the following properties are satisfied for any x 𝑥 x italic_x , y A 𝑦 𝐴 y\in A italic_y ∈ italic_A :

  1. (s1)

    if x y = 0 direct-product 𝑥 𝑦 0 x\odot y=0 italic_x ⊙ italic_y = 0 then s ( x y ) = s ( x ) + s ( y ) 𝑠 direct-sum 𝑥 𝑦 𝑠 𝑥 𝑠 𝑦 s(x\oplus y)=s(x)+s(y) italic_s ( italic_x ⊕ italic_y ) = italic_s ( italic_x ) + italic_s ( italic_y ) ,

  2. (s2)

    s ( 1 ) = 1 𝑠 1 1 s(1)=1 italic_s ( 1 ) = 1 .

Definition 3.1 .

A Riesz MV-algebra is a structure

( R , , , * , 0 ) fragments ( R , , direct-sum superscript , , 0 ) (R,\cdot,\oplus,^{*},0) ( italic_R , ⋅ , ⊕ , start_POSTSUPERSCRIPT * end_POSTSUPERSCRIPT , 0 ) ,

where ( R , , * , 0 ) fragments normal-( R normal-, direct-sum superscript normal-, normal-, 0 normal-) (R,\oplus,^{*},0) ( italic_R , ⊕ , start_POSTSUPERSCRIPT * end_POSTSUPERSCRIPT , 0 ) is an MV-algebra and the operation : [ 0 , 1 ] × R R fragments normal-⋅ normal-: fragments normal-[ 0 normal-, 1 normal-] R normal-→ R \cdot:[0,1]\times R\rightarrow R ⋅ : [ 0 , 1 ] × italic_R → italic_R satisfies the following identities for any r 𝑟 r italic_r , q [ 0 , 1 ] 𝑞 0 1 q\in[0,1] italic_q ∈ [ 0 , 1 ] and x 𝑥 x italic_x , y R 𝑦 𝑅 y\in R italic_y ∈ italic_R :

  1. (RMV1)

    r ( x y * ) = ( r x ) ( r y ) * 𝑟 direct-product 𝑥 superscript 𝑦 direct-product 𝑟 𝑥 superscript 𝑟 𝑦 r\cdot(x\odot y^{*})=(r\cdot x)\odot(r\cdot y)^{*} italic_r ⋅ ( italic_x ⊙ italic_y start_POSTSUPERSCRIPT * end_POSTSUPERSCRIPT ) = ( italic_r ⋅ italic_x ) ⊙ ( italic_r ⋅ italic_y ) start_POSTSUPERSCRIPT * end_POSTSUPERSCRIPT ,

  2. (RMV2)

    ( r q * ) x = ( r x ) ( q x ) * direct-product 𝑟 superscript 𝑞 𝑥 direct-product 𝑟 𝑥 superscript 𝑞 𝑥 (r\odot q^{*})\cdot x=(r\cdot x)\odot(q\cdot x)^{*} ( italic_r ⊙ italic_q start_POSTSUPERSCRIPT * end_POSTSUPERSCRIPT ) ⋅ italic_x = ( italic_r ⋅ italic_x ) ⊙ ( italic_q ⋅ italic_x ) start_POSTSUPERSCRIPT * end_POSTSUPERSCRIPT ,

  3. (RMV3)

    r ( q x ) = ( r q ) x 𝑟 𝑞 𝑥 𝑟 𝑞 𝑥 r\cdot(q\cdot x)=(rq)\cdot x italic_r ⋅ ( italic_q ⋅ italic_x ) = ( italic_r italic_q ) ⋅ italic_x ,

  4. (RMV4)

    1 x = x 1 𝑥 𝑥 1\cdot x=x 1 ⋅ italic_x = italic_x .

In the following we write r x 𝑟 𝑥 rx italic_r italic_x instead of r x normal-⋅ 𝑟 𝑥 r\cdot x italic_r ⋅ italic_x for r [ 0 , 1 ] 𝑟 0 1 r\in[0,1] italic_r ∈ [ 0 , 1 ] and x R 𝑥 𝑅 x\in R italic_x ∈ italic_R . Note that r q 𝑟 𝑞 rq italic_r italic_q is the real product for any r 𝑟 r italic_r , q [ 0 , 1 ] 𝑞 0 1 q\in[0,1] italic_q ∈ [ 0 , 1 ] .


Definition 4.1

An affine l subscript 𝑙 \mathbb{Z}_{l} blackboard_Z start_POSTSUBSCRIPT italic_l end_POSTSUBSCRIPT -module is a pair ( F , X ) 𝐹 𝑋 (F,X) ( italic_F , italic_X ) , where F 𝐹 F italic_F is a free finitely generated l subscript 𝑙 \mathbb{Z}_{l} blackboard_Z start_POSTSUBSCRIPT italic_l end_POSTSUBSCRIPT -module, and X 𝑋 X italic_X is a continuous simply transitive F 𝐹 F italic_F -space. Here F 𝐹 F italic_F is regarded as a topological abelian group, and the simple transitivity means that for every x X 𝑥 𝑋 x\in X italic_x ∈ italic_X , the natural map F X normal-→ 𝐹 𝑋 F\rightarrow X italic_F → italic_X given by f f x normal-→ 𝑓 normal-⋅ 𝑓 𝑥 f\rightarrow f\cdot x italic_f → italic_f ⋅ italic_x is a homeomorphism. A based affine l subscript 𝑙 \mathbb{Z}_{l} blackboard_Z start_POSTSUBSCRIPT italic_l end_POSTSUBSCRIPT -module is a triple ( F , X , B ) 𝐹 𝑋 𝐵 (F,X,B) ( italic_F , italic_X , italic_B ) so that the pair ( F , X ) 𝐹 𝑋 (F,X) ( italic_F , italic_X ) is an affine l subscript 𝑙 \mathbb{Z}_{l} blackboard_Z start_POSTSUBSCRIPT italic_l end_POSTSUBSCRIPT -module and B 𝐵 B italic_B is a basis for the l subscript 𝑙 \mathbb{Z}_{l} blackboard_Z start_POSTSUBSCRIPT italic_l end_POSTSUBSCRIPT -module F 𝐹 F italic_F . An isomorphism of affine l subscript 𝑙 \mathbb{Z}_{l} blackboard_Z start_POSTSUBSCRIPT italic_l end_POSTSUBSCRIPT -modules from ( F , X ) 𝐹 𝑋 (F,X) ( italic_F , italic_X ) to ( F , X ) superscript 𝐹 normal-′ superscript 𝑋 normal-′ (F^{\prime},X^{\prime}) ( italic_F start_POSTSUPERSCRIPT ′ end_POSTSUPERSCRIPT , italic_X start_POSTSUPERSCRIPT ′ end_POSTSUPERSCRIPT ) is a pair ( φ , χ ) 𝜑 𝜒 (\varphi,\chi) ( italic_φ , italic_χ ) , where φ : F F normal-: 𝜑 normal-→ 𝐹 superscript 𝐹 normal-′ \varphi:F\rightarrow F^{\prime} italic_φ : italic_F → italic_F start_POSTSUPERSCRIPT ′ end_POSTSUPERSCRIPT is a continuous isomorphism of l subscript 𝑙 \mathbb{Z}_{l} blackboard_Z start_POSTSUBSCRIPT italic_l end_POSTSUBSCRIPT -modules and χ : X X normal-: 𝜒 normal-→ 𝑋 superscript 𝑋 normal-′ \chi:X\rightarrow X^{\prime} italic_χ : italic_X → italic_X start_POSTSUPERSCRIPT ′ end_POSTSUPERSCRIPT is a homeomorphism, and where the relationship

χ ( f x ) = φ ( f ) χ ( x ) 𝜒 𝑓 𝑥 𝜑 𝑓 𝜒 𝑥 \chi(f\cdot x)=\varphi(f)\cdot\chi(x) italic_χ ( italic_f ⋅ italic_x ) = italic_φ ( italic_f ) ⋅ italic_χ ( italic_x )

holds for all f 𝑓 f italic_f and x 𝑥 x italic_x . An isomorphism φ 𝜑 \varphi italic_φ of based affine l subscript 𝑙 \mathbb{Z}_{l} blackboard_Z start_POSTSUBSCRIPT italic_l end_POSTSUBSCRIPT -modules is an isomorphism of the underlying affine l subscript 𝑙 \mathbb{Z}_{l} blackboard_Z start_POSTSUBSCRIPT italic_l end_POSTSUBSCRIPT -modules which preserves l subscript 𝑙 \mathbb{Z}_{l} blackboard_Z start_POSTSUBSCRIPT italic_l end_POSTSUBSCRIPT -multiples of basis elements, i.e. for all z l 𝑧 subscript 𝑙 z\in\mathbb{Z}_{l} italic_z ∈ blackboard_Z start_POSTSUBSCRIPT italic_l end_POSTSUBSCRIPT and b B 𝑏 𝐵 b\in B italic_b ∈ italic_B there exist z 𝕝 superscript 𝑧 normal-′ subscript 𝕝 z^{\prime}\in\mathbb{Z_{l}} italic_z start_POSTSUPERSCRIPT ′ end_POSTSUPERSCRIPT ∈ blackboard_Z start_POSTSUBSCRIPT blackboard_l end_POSTSUBSCRIPT and b B superscript 𝑏 normal-′ 𝐵 b^{\prime}\in B italic_b start_POSTSUPERSCRIPT ′ end_POSTSUPERSCRIPT ∈ italic_B so that φ ( z b ) = z b 𝜑 𝑧 𝑏 superscript 𝑧 normal-′ superscript 𝑏 normal-′ \varphi({zb})=z^{\prime}b^{\prime} italic_φ ( italic_z italic_b ) = italic_z start_POSTSUPERSCRIPT ′ end_POSTSUPERSCRIPT italic_b start_POSTSUPERSCRIPT ′ end_POSTSUPERSCRIPT . If for all z 𝑧 z italic_z and b 𝑏 b italic_b , there is an element z l superscript 𝑧 normal-′ subscript 𝑙 z^{\prime}\in\mathbb{Z}_{l} italic_z start_POSTSUPERSCRIPT ′ end_POSTSUPERSCRIPT ∈ blackboard_Z start_POSTSUBSCRIPT italic_l end_POSTSUBSCRIPT so that φ ( z b ) = z b 𝜑 𝑧 𝑏 superscript 𝑧 normal-′ 𝑏 \varphi(zb)=z^{\prime}b italic_φ ( italic_z italic_b ) = italic_z start_POSTSUPERSCRIPT ′ end_POSTSUPERSCRIPT italic_b , then the isomorphism is said to be a translation .


Definition 1.4 (Discriminant) .

There is another invariant of a binary form f 𝑓 f italic_f called the discriminant of f 𝑓 f italic_f . If f = ( a , b , c ) 𝑓 𝑎 𝑏 𝑐 f=(a,b,c) italic_f = ( italic_a , italic_b , italic_c ) , then the discriminant d ( f ) 𝑑 𝑓 d(f) italic_d ( italic_f ) of f 𝑓 f italic_f is defined by the formula

(1.2) d ( f ) = b 2 - 4 a c . 𝑑 𝑓 superscript 𝑏 2 4 𝑎 𝑐 d(f)=b^{2}-4ac. italic_d ( italic_f ) = italic_b start_POSTSUPERSCRIPT 2 end_POSTSUPERSCRIPT - 4 italic_a italic_c .

The discriminant is also an invariant, although this is not immediately obvious from the definition. Later on, we will give a more invariant definition of the discriminant, from which the invariance will be clear.


Definition 5.3 (Support set partition for IHT)

Suppose δ ( 0 , 1 ] 𝛿 0 1 \delta\in(0,1] italic_δ ∈ ( 0 , 1 ] , ρ ( 0 , 1 / 2 ] 𝜌 0 1 2 \rho\in(0,1/2] italic_ρ ∈ ( 0 , 1 / 2 ] and α > 0 𝛼 0 \alpha>0 italic_α > 0 . Given ζ > 0 𝜁 0 \zeta>0 italic_ζ > 0 , let us write

a ( δ , ρ ; ζ ) = def a ( δ , ρ ) + ζ , superscript def superscript 𝑎 𝛿 𝜌 𝜁 𝑎 𝛿 𝜌 𝜁 a^{\ast}(\delta,\rho;\zeta)\stackrel{{\scriptstyle\rm def}}{{=}}a(\delta,\rho)% +\zeta, italic_a start_POSTSUPERSCRIPT ∗ end_POSTSUPERSCRIPT ( italic_δ , italic_ρ ; italic_ζ ) start_RELOP SUPERSCRIPTOP start_ARG = end_ARG start_ARG roman_def end_ARG end_RELOP italic_a ( italic_δ , italic_ρ ) + italic_ζ , (5.6)

where a ( δ , ρ ) 𝑎 𝛿 𝜌 a(\delta,\rho) italic_a ( italic_δ , italic_ρ ) is defined in ( 5.2 ), let us write { Γ i : i S n } conditional-set subscript normal-Γ 𝑖 𝑖 subscript 𝑆 𝑛 \{\Gamma_{i}:i\in S_{n}\} { roman_Γ start_POSTSUBSCRIPT italic_i end_POSTSUBSCRIPT : italic_i ∈ italic_S start_POSTSUBSCRIPT italic_n end_POSTSUBSCRIPT } for the set of all possible support sets of cardinality k 𝑘 k italic_k , and let us disjointly partition S n = def Θ n 1 Θ n 2 superscript normal-def subscript 𝑆 𝑛 subscript superscript normal-Θ 1 𝑛 subscript superscript normal-Θ 2 𝑛 S_{n}\stackrel{{\scriptstyle\rm def}}{{=}}\Theta^{1}_{n}\cup\Theta^{2}_{n} italic_S start_POSTSUBSCRIPT italic_n end_POSTSUBSCRIPT start_RELOP SUPERSCRIPTOP start_ARG = end_ARG start_ARG roman_def end_ARG end_RELOP roman_Θ start_POSTSUPERSCRIPT 1 end_POSTSUPERSCRIPT start_POSTSUBSCRIPT italic_n end_POSTSUBSCRIPT ∪ roman_Θ start_POSTSUPERSCRIPT 2 end_POSTSUPERSCRIPT start_POSTSUBSCRIPT italic_n end_POSTSUBSCRIPT such that

Θ n 1 = def { i S n : x Λ Γ i > σ a ( δ , ρ ; ζ ) } ; Θ n 2 = def { i S n : x Λ Γ i σ a ( δ , ρ ; ζ ) } . formulae-sequence superscript def subscript superscript Θ 1 𝑛 conditional-set 𝑖 subscript 𝑆 𝑛 norm subscript superscript 𝑥 Λ subscript Γ 𝑖 𝜎 superscript 𝑎 𝛿 𝜌 𝜁 superscript def subscript superscript Θ 2 𝑛 conditional-set 𝑖 subscript 𝑆 𝑛 norm subscript superscript 𝑥 Λ subscript Γ 𝑖 𝜎 superscript 𝑎 𝛿 𝜌 𝜁 \Theta^{1}_{n}\stackrel{{\scriptstyle\rm def}}{{=}}\left\{i\in S_{n}\;\;:\;\;% \|x^{\ast}_{\Lambda\setminus\Gamma_{i}}\|>\sigma\cdot a^{\ast}(\delta,\rho;% \zeta)\right\};\;\;\;\;\Theta^{2}_{n}\stackrel{{\scriptstyle\rm def}}{{=}}% \left\{i\in S_{n}\;\;:\;\;\|x^{\ast}_{\Lambda\setminus\Gamma_{i}}\|\leq\sigma% \cdot a^{\ast}(\delta,\rho;\zeta)\right\}. roman_Θ start_POSTSUPERSCRIPT 1 end_POSTSUPERSCRIPT start_POSTSUBSCRIPT italic_n end_POSTSUBSCRIPT start_RELOP SUPERSCRIPTOP start_ARG = end_ARG start_ARG roman_def end_ARG end_RELOP { italic_i ∈ italic_S start_POSTSUBSCRIPT italic_n end_POSTSUBSCRIPT : ∥ italic_x start_POSTSUPERSCRIPT ∗ end_POSTSUPERSCRIPT start_POSTSUBSCRIPT roman_Λ ∖ roman_Γ start_POSTSUBSCRIPT italic_i end_POSTSUBSCRIPT end_POSTSUBSCRIPT ∥ > italic_σ ⋅ italic_a start_POSTSUPERSCRIPT ∗ end_POSTSUPERSCRIPT ( italic_δ , italic_ρ ; italic_ζ ) } ; roman_Θ start_POSTSUPERSCRIPT 2 end_POSTSUPERSCRIPT start_POSTSUBSCRIPT italic_n end_POSTSUBSCRIPT start_RELOP SUPERSCRIPTOP start_ARG = end_ARG start_ARG roman_def end_ARG end_RELOP { italic_i ∈ italic_S start_POSTSUBSCRIPT italic_n end_POSTSUBSCRIPT : ∥ italic_x start_POSTSUPERSCRIPT ∗ end_POSTSUPERSCRIPT start_POSTSUBSCRIPT roman_Λ ∖ roman_Γ start_POSTSUBSCRIPT italic_i end_POSTSUBSCRIPT end_POSTSUBSCRIPT ∥ ≤ italic_σ ⋅ italic_a start_POSTSUPERSCRIPT ∗ end_POSTSUPERSCRIPT ( italic_δ , italic_ρ ; italic_ζ ) } . (5.7)

Definition 4.4 ( [ KLS11 ] ) .

Given a cyclic rank matrix corresponding to a positroid on [ n ] delimited-[] 𝑛 [n] [ italic_n ] , we can form an affine permutation . This will be a bijection π : : 𝜋 \pi:\mathbb{Z}\to\mathbb{Z} italic_π : blackboard_Z → blackboard_Z such that i π ( i ) i + n 𝑖 𝜋 𝑖 𝑖 𝑛 i\leq\pi(i)\leq i+n italic_i ≤ italic_π ( italic_i ) ≤ italic_i + italic_n and π ( i + n ) = π ( i ) + n 𝜋 𝑖 𝑛 𝜋 𝑖 𝑛 \pi(i+n)=\pi(i)+n italic_π ( italic_i + italic_n ) = italic_π ( italic_i ) + italic_n for all i 𝑖 i italic_i . We define π 𝜋 \pi italic_π as a matrix by putting a 1 in position ( i , j ) 𝑖 𝑗 (i,j) ( italic_i , italic_j ) if r i j = r i , j - 1 = r i + 1 , j r i + 1 , j - 1 subscript 𝑟 𝑖 𝑗 subscript 𝑟 𝑖 𝑗 1 subscript 𝑟 𝑖 1 𝑗 subscript 𝑟 𝑖 1 𝑗 1 r_{ij}=r_{i,j-1}=r_{i+1,j}\neq r_{i+1,j-1} italic_r start_POSTSUBSCRIPT italic_i italic_j end_POSTSUBSCRIPT = italic_r start_POSTSUBSCRIPT italic_i , italic_j - 1 end_POSTSUBSCRIPT = italic_r start_POSTSUBSCRIPT italic_i + 1 , italic_j end_POSTSUBSCRIPT ≠ italic_r start_POSTSUBSCRIPT italic_i + 1 , italic_j - 1 end_POSTSUBSCRIPT and putting a 0 there otherwise. One can check that each row and each column will have exactly one 1. Note that to describe π 𝜋 \pi italic_π it’s enough to describe the images of the elements of [ n ] delimited-[] 𝑛 [n] [ italic_n ] .

One can reverse this process to read off the rank matrix from the affine permutation matrix. Given an interval [ i , j ] 𝑖 𝑗 [i,j] [ italic_i , italic_j ] , consider the entries of the matrix weakly southwest of position ( i , j ) 𝑖 𝑗 (i,j) ( italic_i , italic_j ) in the affine permutation matrix. (That is, positions ( k , l ) 𝑘 𝑙 (k,l) ( italic_k , italic_l ) for which [ k , l ] [ i , j ] 𝑘 𝑙 𝑖 𝑗 [k,l]\subseteq[i,j] [ italic_k , italic_l ] ⊆ [ italic_i , italic_j ] .) The rank of [ i , j ] 𝑖 𝑗 [i,j] [ italic_i , italic_j ] then works out to be # [ i , j ] - d # 𝑖 𝑗 𝑑 \#[i,j]-d # [ italic_i , italic_j ] - italic_d , where d 𝑑 d italic_d is the number of 1’s among the these entries. Equivalently, thinking of the affine permutation as a function, we can say

d = # { k : i , k , π ( k ) , π ( i ) occur cyclically consecutively } . 𝑑 # conditional-set 𝑘 𝑖 𝑘 𝜋 𝑘 𝜋 𝑖 occur cyclically consecutively d=\#\{k:i,k,\pi(k),\pi(i)\mbox{ occur cyclically consecutively}\}. italic_d = # { italic_k : italic_i , italic_k , italic_π ( italic_k ) , italic_π ( italic_i ) occur cyclically consecutively } .

It’s also possible to determine the codimension of a positroid variety from an affine permutation. For an affine permutation π 𝜋 \pi italic_π , define the length of π 𝜋 \pi italic_π , written l ( π ) 𝑙 𝜋 l(\pi) italic_l ( italic_π ) , as the number of inversions , that is, the number of pairs i , k 𝑖 𝑘 i,k italic_i , italic_k with i , k , π ( k ) , π ( i ) 𝑖 𝑘 𝜋 𝑘 𝜋 𝑖 i,k,\pi(k),\pi(i) italic_i , italic_k , italic_π ( italic_k ) , italic_π ( italic_i ) occurring cyclically consecutively. Each of these pairs will correspond to a pair of 1’s in the affine permutation matrix arranged southwest-to-northeast. Then l ( π ) 𝑙 𝜋 l(\pi) italic_l ( italic_π ) is the codimension of the positroid variety corresponding to π 𝜋 \pi italic_π . (This is proved in [ KLS11 , 5.9] .)


Definition 2.3 .

Let SS SS \SS roman_SS be a semiring. A map τ : SS SS : 𝜏 SS SS \tau:\SS\to\SS italic_τ : roman_SS → roman_SS is a symmetry if

(2.1a) τ ( a + b ) = τ ( a ) + τ ( b ) 𝜏 𝑎 𝑏 𝜏 𝑎 𝜏 𝑏 \displaystyle\tau(a+b)=\tau(a)+\tau(b) italic_τ ( italic_a + italic_b ) = italic_τ ( italic_a ) + italic_τ ( italic_b )
(2.1b) τ ( 0 ) = 0 𝜏 0 0 \displaystyle\tau(0)=0 italic_τ ( 0 ) = 0
(2.1c) τ ( a b ) = a τ ( b ) = τ ( a ) b 𝜏 𝑎 𝑏 𝑎 𝜏 𝑏 𝜏 𝑎 𝑏 \displaystyle\tau(a\cdot b)=a\cdot\tau(b)=\tau(a)\cdot b italic_τ ( italic_a ⋅ italic_b ) = italic_a ⋅ italic_τ ( italic_b ) = italic_τ ( italic_a ) ⋅ italic_b
(2.1d) τ ( τ ( a ) ) = a . 𝜏 𝜏 𝑎 𝑎 \displaystyle\tau(\tau(a))=a. italic_τ ( italic_τ ( italic_a ) ) = italic_a .

Definition 2 (MKR mixing measure) .

Let ρ : [ 0 , 1 ) d { ± 1 } normal-: 𝜌 normal-→ superscript 0 1 𝑑 plus-or-minus 1 \rho:[0,1)^{d}\to\{\pm 1\} italic_ρ : [ 0 , 1 ) start_POSTSUPERSCRIPT italic_d end_POSTSUPERSCRIPT → { ± 1 } be a periodic function with

ρ 𝑑 x = 0 . 𝜌 differential-d 𝑥 0 \int\rho\,dx=0. ∫ italic_ρ italic_d italic_x = 0 .

Let ρ + subscript 𝜌 \rho_{+} italic_ρ start_POSTSUBSCRIPT + end_POSTSUBSCRIPT and ρ - subscript 𝜌 \rho_{-} italic_ρ start_POSTSUBSCRIPT - end_POSTSUBSCRIPT be the probability densities defined by

ρ + = 2 max { ρ , 0 } 𝑎𝑛𝑑 ρ - = - 2 min { ρ , 0 } , formulae-sequence subscript 𝜌 2 𝜌 0 𝑎𝑛𝑑 subscript 𝜌 2 𝜌 0 \rho_{+}=2\max\{\rho,0\}\quad\mbox{and}\quad\rho_{-}=-2\min\{\rho,0\}, italic_ρ start_POSTSUBSCRIPT + end_POSTSUBSCRIPT = 2 roman_max { italic_ρ , 0 } and italic_ρ start_POSTSUBSCRIPT - end_POSTSUBSCRIPT = - 2 roman_min { italic_ρ , 0 } ,

so that ρ = 1 2 ( ρ + - ρ - ) 𝜌 1 2 subscript 𝜌 subscript 𝜌 \rho=\frac{1}{2}(\rho_{+}-\rho_{-}) italic_ρ = divide start_ARG 1 end_ARG start_ARG 2 end_ARG ( italic_ρ start_POSTSUBSCRIPT + end_POSTSUBSCRIPT - italic_ρ start_POSTSUBSCRIPT - end_POSTSUBSCRIPT ) . We then define the MKR mixing measure as

D ( ρ ) := exp ( d ln ( ρ + , ρ - ) ) . assign 𝐷 𝜌 subscript 𝑑 subscript 𝜌 subscript 𝜌 D(\rho):=\exp\left(d_{\ln}(\rho_{+},\rho_{-})\right). italic_D ( italic_ρ ) := roman_exp ( italic_d start_POSTSUBSCRIPT roman_ln end_POSTSUBSCRIPT ( italic_ρ start_POSTSUBSCRIPT + end_POSTSUBSCRIPT , italic_ρ start_POSTSUBSCRIPT - end_POSTSUBSCRIPT ) ) .

Definition 8 (Geodesic completeness for generalized metrics) .

Let g 𝒢 2 0 ( M ) 𝑔 subscript superscript 𝒢 0 2 𝑀 g\in\mathcal{G}^{0}_{2}(M) italic_g ∈ caligraphic_G start_POSTSUPERSCRIPT 0 end_POSTSUPERSCRIPT start_POSTSUBSCRIPT 2 end_POSTSUBSCRIPT ( italic_M ) be a generalized metric of Lorentzian signature. Then the generalized space-time ( M , g ) 𝑀 𝑔 (M,g) ( italic_M , italic_g ) is said to be geodesically complete if every geodesic can be defined on \mathbb{R} blackboard_R , i.e., every solution of the geodesic equation

c ′′ = 0 superscript 𝑐 ′′ 0 c^{\prime\prime}=0 italic_c start_POSTSUPERSCRIPT ′ ′ end_POSTSUPERSCRIPT = 0

is in 𝒢 [ , M ] 𝒢 𝑀 \mathcal{G}[\mathbb{R},M] caligraphic_G [ blackboard_R , italic_M ] .


Definition 2.1 .

[ 14 ] A vector space T 𝑇 T italic_T together with a trilinear map ( x , y , z ) [ x y z ] maps-to 𝑥 𝑦 𝑧 delimited-[] 𝑥 𝑦 𝑧 (x,y,z)\mapsto[xyz] ( italic_x , italic_y , italic_z ) ↦ [ italic_x italic_y italic_z ] is called a Lie triple system (LTS) if

  1. (1)

    [ x x z ] = 0 delimited-[] 𝑥 𝑥 𝑧 0 [xxz]=0 [ italic_x italic_x italic_z ] = 0 ,

  2. (2)

    [ x y z ] + [ y z x ] + [ z x y ] = 0 delimited-[] 𝑥 𝑦 𝑧 delimited-[] 𝑦 𝑧 𝑥 delimited-[] 𝑧 𝑥 𝑦 0 [xyz]+[yzx]+[zxy]=0 [ italic_x italic_y italic_z ] + [ italic_y italic_z italic_x ] + [ italic_z italic_x italic_y ] = 0 ,

  3. (3)

    [ u v [ x y z ] ] = [ [ u v x ] y z ] + [ x [ u v y ] z ] + [ x y [ u v z ] ] delimited-[] 𝑢 𝑣 delimited-[] 𝑥 𝑦 𝑧 delimited-[] delimited-[] 𝑢 𝑣 𝑥 𝑦 𝑧 delimited-[] 𝑥 delimited-[] 𝑢 𝑣 𝑦 𝑧 delimited-[] 𝑥 𝑦 delimited-[] 𝑢 𝑣 𝑧 [uv[xyz]]=[[uvx]yz]+[x[uvy]z]+[xy[uvz]] [ italic_u italic_v [ italic_x italic_y italic_z ] ] = [ [ italic_u italic_v italic_x ] italic_y italic_z ] + [ italic_x [ italic_u italic_v italic_y ] italic_z ] + [ italic_x italic_y [ italic_u italic_v italic_z ] ] ,

for all x , y , z , u , v T 𝑥 𝑦 𝑧 𝑢 𝑣 𝑇 x,y,z,u,v\in T italic_x , italic_y , italic_z , italic_u , italic_v ∈ italic_T .


Definition 1.1 .

A quadratic form q 𝑞 q italic_q on V 𝑉 V italic_V is a function q : V R normal-: 𝑞 normal-→ 𝑉 𝑅 q:V\to R italic_q : italic_V → italic_R such that

q ( a x ) = a 2 q ( x ) 𝑞 𝑎 𝑥 superscript 𝑎 2 𝑞 𝑥 q(ax)=a^{2}q(x) italic_q ( italic_a italic_x ) = italic_a start_POSTSUPERSCRIPT 2 end_POSTSUPERSCRIPT italic_q ( italic_x ) (1.1)

for any a R , 𝑎 𝑅 a\in R, italic_a ∈ italic_R , x V , 𝑥 𝑉 x\in V, italic_x ∈ italic_V , and there exists a symmetric bilinear form b : V × V R normal-: 𝑏 normal-→ 𝑉 𝑉 𝑅 b:V\times V\to R italic_b : italic_V × italic_V → italic_R (not necessarily uniquely determined by q 𝑞 q italic_q ) such that for any x , y V 𝑥 𝑦 𝑉 x,y\in V italic_x , italic_y ∈ italic_V

q ( x + y ) = q ( x ) + q ( y ) + b ( x , y ) . 𝑞 𝑥 𝑦 𝑞 𝑥 𝑞 𝑦 𝑏 𝑥 𝑦 q(x+y)=q(x)+q(y)+b(x,y). italic_q ( italic_x + italic_y ) = italic_q ( italic_x ) + italic_q ( italic_y ) + italic_b ( italic_x , italic_y ) . (1.2)

We then also say that ( q , b ) 𝑞 𝑏 (q,b) ( italic_q , italic_b ) is a quadratic pair on V 𝑉 V italic_V .

Definition 1.2 .

We call a quadratic pair ( q , b ) 𝑞 𝑏 (q,b) ( italic_q , italic_b ) on V 𝑉 V italic_V balanced if for any x V 𝑥 𝑉 x\in V italic_x ∈ italic_V

b ( x , x ) = 2 q ( x ) . 𝑏 𝑥 𝑥 2 𝑞 𝑥 b(x,x)=2q(x). italic_b ( italic_x , italic_x ) = 2 italic_q ( italic_x ) . (1.4)
Definition 1.5 .

A quadratic form q : V R normal-: 𝑞 normal-→ 𝑉 𝑅 q:V\to R italic_q : italic_V → italic_R is called quasilinear 3 3 3 In the case that R normal-R R italic_R is a valuation domain, cf. [ 16 , I,§6] . In [ 2 ] these forms are called “totally singular”. if q 𝑞 q italic_q , together with the null form b : V × V R , normal-: 𝑏 normal-→ 𝑉 𝑉 𝑅 b:V\times V\to R, italic_b : italic_V × italic_V → italic_R , b ( x , y ) = 0 𝑏 𝑥 𝑦 0 b(x,y)=0 italic_b ( italic_x , italic_y ) = 0 for all x , y V , 𝑥 𝑦 𝑉 x,y\in V, italic_x , italic_y ∈ italic_V , is a quadratic pair. This means that for any x , y V , 𝑥 𝑦 𝑉 x,y\in V, italic_x , italic_y ∈ italic_V , a R 𝑎 𝑅 a\in R italic_a ∈ italic_R ,

q ( a x ) = a 2 q ( x ) , q ( x + y ) = q ( x ) + q ( y ) . formulae-sequence 𝑞 𝑎 𝑥 superscript 𝑎 2 𝑞 𝑥 𝑞 𝑥 𝑦 𝑞 𝑥 𝑞 𝑦 q(ax)=a^{2}q(x),\qquad q(x+y)=q(x)+q(y). italic_q ( italic_a italic_x ) = italic_a start_POSTSUPERSCRIPT 2 end_POSTSUPERSCRIPT italic_q ( italic_x ) , italic_q ( italic_x + italic_y ) = italic_q ( italic_x ) + italic_q ( italic_y ) . (1.9)
Definition 2.1 .

Let b : V × V R normal-: 𝑏 normal-→ 𝑉 𝑉 𝑅 b:V\times V\to R italic_b : italic_V × italic_V → italic_R be a symmetric bilinear form. We say that b 𝑏 b italic_b is a companion of q 𝑞 q italic_q at a point ( x , y ) V × V 𝑥 𝑦 𝑉 𝑉 (x,y)\in V\times V ( italic_x , italic_y ) ∈ italic_V × italic_V (or, that b 𝑏 b italic_b accompanies q 𝑞 q italic_q at ( x , y ) 𝑥 𝑦 (x,y) ( italic_x , italic_y ) ), if

q ( x + y ) = q ( x ) + q ( y ) + b ( x , y ) , 𝑞 𝑥 𝑦 𝑞 𝑥 𝑞 𝑦 𝑏 𝑥 𝑦 q(x+y)=q(x)+q(y)+b(x,y), italic_q ( italic_x + italic_y ) = italic_q ( italic_x ) + italic_q ( italic_y ) + italic_b ( italic_x , italic_y ) , (2.1)

which can be rephrased as

q ( x ) + q ( y ) + b ( x , y ) = q ( x ) + q ( y ) + b 0 ( x , y ) . 𝑞 𝑥 𝑞 𝑦 𝑏 𝑥 𝑦 𝑞 𝑥 𝑞 𝑦 subscript 𝑏 0 𝑥 𝑦 q(x)+q(y)+b(x,y)=q(x)+q(y)+b_{0}(x,y). italic_q ( italic_x ) + italic_q ( italic_y ) + italic_b ( italic_x , italic_y ) = italic_q ( italic_x ) + italic_q ( italic_y ) + italic_b start_POSTSUBSCRIPT 0 end_POSTSUBSCRIPT ( italic_x , italic_y ) . (2.2)

If this happens to be true for every point ( x , y ) 𝑥 𝑦 (x,y) ( italic_x , italic_y ) of a set T V × V , 𝑇 𝑉 𝑉 T\subset V\times V, italic_T ⊂ italic_V × italic_V , we call b 𝑏 b italic_b a companion of q 𝑞 q italic_q on T ; 𝑇 T; italic_T ; and in the subcase T = S × S , 𝑇 𝑆 𝑆 T=S\times S, italic_T = italic_S × italic_S , where S V , 𝑆 𝑉 S\subset V, italic_S ⊂ italic_V , we say more briefly that b 𝑏 b italic_b is a companion of q 𝑞 q italic_q on S . 𝑆 S. italic_S . We then also say that b 𝑏 b italic_b accompanies q 𝑞 q italic_q on T , 𝑇 T, italic_T , resp. on S . 𝑆 S. italic_S .

Definition 2.3 .

We say that q 𝑞 q italic_q is quasilinear on a set T V × V 𝑇 𝑉 𝑉 T\subset V\times V italic_T ⊂ italic_V × italic_V , if b = 0 𝑏 0 b=0 italic_b = 0 accompanies q 𝑞 q italic_q on T , 𝑇 T, italic_T , i.e., for any ( x , y ) T 𝑥 𝑦 𝑇 (x,y)\in T ( italic_x , italic_y ) ∈ italic_T

q ( x + y ) = q ( x ) + q ( y ) ; 𝑞 𝑥 𝑦 𝑞 𝑥 𝑞 𝑦 q(x+y)=q(x)+q(y); italic_q ( italic_x + italic_y ) = italic_q ( italic_x ) + italic_q ( italic_y ) ;

and we say that q 𝑞 q italic_q is quasilinear on S V 𝑆 𝑉 S\subset V italic_S ⊂ italic_V if this happens on T = S × S . 𝑇 𝑆 𝑆 T=S\times S. italic_T = italic_S × italic_S .


Definition 4.7 .

We define the exponential tensor _ _ : E n d ( ) × E n d ( ) E n d ( ) normal-: direct-product normal-_ normal-_ normal-→ 𝐸 𝑛 𝑑 𝐸 𝑛 𝑑 𝐸 𝑛 𝑑 \_\odot\_:End(\mathbb{N})\times End(\mathbb{N})\rightarrow End(\mathbb{N}) _ ⊙ _ : italic_E italic_n italic_d ( blackboard_N ) × italic_E italic_n italic_d ( blackboard_N ) → italic_E italic_n italic_d ( blackboard_N ) in terms of the exponential bijection of Definition 4.5 above, as follows:

f g = ψ ( f × g ) ψ - 1 direct-product 𝑓 𝑔 𝜓 𝑓 𝑔 superscript 𝜓 1 f\odot g\ =\ \psi(f\times g)\psi^{-1} italic_f ⊙ italic_g = italic_ψ ( italic_f × italic_g ) italic_ψ start_POSTSUPERSCRIPT - 1 end_POSTSUPERSCRIPT

Diagramatically,

\xymatrix × \ar [ r r ] f × g & & × \ar [ d ] ψ \ar [ u ] ψ - 1 \ar [ r r ] f g & & \xymatrix \ar superscript delimited-[] 𝑟 𝑟 𝑓 𝑔 & & \ar superscript delimited-[] 𝑑 𝜓 \ar superscript delimited-[] 𝑢 superscript 𝜓 1 \ar subscript delimited-[] 𝑟 𝑟 direct-product 𝑓 𝑔 & & \xymatrix{\mathbb{N}\times\mathbb{N}\ar[rr]^{f\times g}&&\mathbb{N}\times% \mathbb{N}\ar[d]^{\psi}\\ \mathbb{N}\ar[u]^{\psi^{-1}}\ar[rr]_{f\odot g}&&\mathbb{N}} blackboard_N × blackboard_N [ italic_r italic_r ] start_POSTSUPERSCRIPT italic_f × italic_g end_POSTSUPERSCRIPT & & blackboard_N × blackboard_N [ italic_d ] start_POSTSUPERSCRIPT italic_ψ end_POSTSUPERSCRIPT blackboard_N [ italic_u ] start_POSTSUPERSCRIPT italic_ψ start_POSTSUPERSCRIPT - 1 end_POSTSUPERSCRIPT end_POSTSUPERSCRIPT [ italic_r italic_r ] start_POSTSUBSCRIPT italic_f ⊙ italic_g end_POSTSUBSCRIPT & & blackboard_N

Definition 25 .

Given two functions f , g 𝑓 𝑔 f,g italic_f , italic_g with suitable regularity, we define the convolution f g 𝑓 𝑔 f\ast g italic_f ∗ italic_g as

( f g ) ( ξ , η ) = f ( ξ , ζ ) g ( ζ , η ) 𝑑 ζ . 𝑓 𝑔 𝜉 𝜂 𝑓 𝜉 𝜁 𝑔 𝜁 𝜂 differential-d 𝜁 (f\ast g)(\xi,\eta)=\int f(\xi,\zeta)g(\zeta,\eta)\,d\zeta. ( italic_f ∗ italic_g ) ( italic_ξ , italic_η ) = ∫ italic_f ( italic_ξ , italic_ζ ) italic_g ( italic_ζ , italic_η ) italic_d italic_ζ . (127)

Definition 19 :

A measure 𝗑 𝒳 𝗑 𝒳 \mathsf{x}\in\mathcal{X} sansserif_x ∈ caligraphic_X is a DE fixed point for the LDPC( λ , ρ 𝜆 𝜌 \lambda,\rho italic_λ , italic_ρ ) ensemble if

𝗑 = 𝖼 λ ( ρ ( 𝗑 ) ) . 𝗑 𝖼 superscript 𝜆 superscript 𝜌 𝗑 \displaystyle\mathsf{x}=\mathsf{c}\varoast\lambda^{\varoast}\left(\rho^{% \boxast}(\mathsf{x})\right). sansserif_x = sansserif_c ⊛ italic_λ start_POSTSUPERSCRIPT ⊛ end_POSTSUPERSCRIPT ( italic_ρ start_POSTSUPERSCRIPT ⧆ end_POSTSUPERSCRIPT ( sansserif_x ) ) .

Definition 9.9 .

Consider the following terms.

𝗌𝗎𝖼 ( x ) = 1 ; ( p ˘ ; x ; q ˘ ) 𝗌𝗎𝖼 𝑥 1 ˘ 𝑝 𝑥 ˘ 𝑞 {\sf suc}(x)=1;(\breve{p};x;\breve{q}) sansserif_suc ( italic_x ) = 1 ; ( ˘ start_ARG italic_p end_ARG ; italic_x ; ˘ start_ARG italic_q end_ARG )

and

𝗉𝗋𝖾𝖽 ( x ) = p ˘ ; 𝗋𝗇𝗀 x ; q . 𝗉𝗋𝖾𝖽 𝑥 ˘ 𝑝 𝗋𝗇𝗀 𝑥 𝑞 {\sf pred}(x)=\breve{p};{\sf rng}x;q. sansserif_pred ( italic_x ) = ˘ start_ARG italic_p end_ARG ; sansserif_rng italic_x ; italic_q .

Definition 1 .

A bijective, anti-linear operator θ : 𝗁 𝗁 normal-: 𝜃 normal-⟶ 𝗁 𝗁 \theta:\mathsf{h}\longrightarrow\mathsf{h} italic_θ : sansserif_h ⟶ sansserif_h is called anti-unitary if

θ u , θ v = v , u , for all x , y 𝗁 . formulae-sequence 𝜃 𝑢 𝜃 𝑣 𝑣 𝑢 for all 𝑥 𝑦 𝗁 \displaystyle\langle\theta u,\theta v\rangle=\langle v,u\rangle,\ \ \ \text{ % for all }x,y\in\mathsf{h}. ⟨ italic_θ italic_u , italic_θ italic_v ⟩ = ⟨ italic_v , italic_u ⟩ , for all italic_x , italic_y ∈ sansserif_h .

Definition 2.8

Let φ 𝜑 \varphi italic_φ be a formula. The necessary input length for φ 𝜑 \varphi italic_φ , l ( φ ) 𝑙 𝜑 l(\varphi) italic_l ( italic_φ ) is defined inductively:

l ( ) = l ( μ ) 𝑙 top 𝑙 𝜇 \displaystyle l(\top)=l(\mu) italic_l ( ⊤ ) = italic_l ( italic_μ ) = 0 absent 0 \displaystyle=0 = 0
l ( ¬ φ ) = l ( i φ ) 𝑙 𝜑 𝑙 subscript 𝑖 𝜑 \displaystyle l(\neg\varphi)=l(\operatorname{\ast}_{i}\varphi) italic_l ( ¬ italic_φ ) = italic_l ( ∗ start_POSTSUBSCRIPT italic_i end_POSTSUBSCRIPT italic_φ ) = l ( φ ) absent 𝑙 𝜑 \displaystyle=l(\varphi) = italic_l ( italic_φ )
l ( φ 1 φ 2 ) 𝑙 subscript 𝜑 1 subscript 𝜑 2 \displaystyle l(\varphi_{1}\lor\varphi_{2}) italic_l ( italic_φ start_POSTSUBSCRIPT 1 end_POSTSUBSCRIPT ∨ italic_φ start_POSTSUBSCRIPT 2 end_POSTSUBSCRIPT ) = max ( l ( φ 1 ) , l ( φ 2 ) ) absent 𝑙 subscript 𝜑 1 𝑙 subscript 𝜑 2 \displaystyle=\max(l(\varphi_{1}),l(\varphi_{2})) = roman_max ( italic_l ( italic_φ start_POSTSUBSCRIPT 1 end_POSTSUBSCRIPT ) , italic_l ( italic_φ start_POSTSUBSCRIPT 2 end_POSTSUBSCRIPT ) )
l ( φ 1 𝑼 I φ 2 ) 𝑙 subscript 𝜑 1 subscript 𝑼 𝐼 subscript 𝜑 2 \displaystyle l(\varphi_{1}\operatorname{\textbf{U}}_{I}\varphi_{2}) italic_l ( italic_φ start_POSTSUBSCRIPT 1 end_POSTSUBSCRIPT U start_POSTSUBSCRIPT italic_I end_POSTSUBSCRIPT italic_φ start_POSTSUBSCRIPT 2 end_POSTSUBSCRIPT ) = max ( l ( φ 1 ) , l ( φ 2 ) ) + sup I absent 𝑙 subscript 𝜑 1 𝑙 subscript 𝜑 2 supremum 𝐼 \displaystyle=\max(l(\varphi_{1}),l(\varphi_{2}))+\sup I = roman_max ( italic_l ( italic_φ start_POSTSUBSCRIPT 1 end_POSTSUBSCRIPT ) , italic_l ( italic_φ start_POSTSUBSCRIPT 2 end_POSTSUBSCRIPT ) ) + roman_sup italic_I

Definition 6 (Bounds of Type 1)

A time bound β 𝛽 \beta italic_β is said to be of Type 1 iff

β ( n ) = n α ( n ) 𝛽 𝑛 𝑛 𝛼 𝑛 \beta(n)=n\cdot\alpha(n) italic_β ( italic_n ) = italic_n ⋅ italic_α ( italic_n )

with a nondecreasing function α 𝛼 \alpha italic_α such that 2 a α ( n ) a n subscript a 2 𝛼 𝑛 subscript a 𝑛 2\mbox{$\leq_{\mbox{\footnotesize\rm a}}$}\alpha(n)\mbox{$\leq_{\mbox{% \footnotesize\rm a}}$}n 2 ≤ start_POSTSUBSCRIPT a end_POSTSUBSCRIPT italic_α ( italic_n ) ≤ start_POSTSUBSCRIPT a end_POSTSUBSCRIPT italic_n and α 𝛼 \alpha italic_α is f-consistent and constructible in linear time.

Definition 8 (Bounds of Type 2)

A time bound β 𝛽 \beta italic_β is said to be of Type 2  iff

β ( n ) = n α ( n ) 𝛽 𝑛 superscript 𝑛 𝛼 𝑛 \beta(n)=n^{\alpha(n)} italic_β ( italic_n ) = italic_n start_POSTSUPERSCRIPT italic_α ( italic_n ) end_POSTSUPERSCRIPT

with a nondecreasing function α 𝛼 \alpha italic_α such that 2 a α ( n ) a log ( n ) subscript a 2 𝛼 𝑛 subscript a 𝑛 2\,\mbox{$\leq_{\mbox{\footnotesize\rm a}}$}\,\alpha(n)\,\mbox{$\leq_{\mbox{% \footnotesize\rm a}}$}\,\log(n) 2 ≤ start_POSTSUBSCRIPT a end_POSTSUBSCRIPT italic_α ( italic_n ) ≤ start_POSTSUBSCRIPT a end_POSTSUBSCRIPT roman_log ( italic_n ) and α 𝛼 \alpha italic_α is e-consistent and constructible in linear time.