OctoInformBeta

OpenAI preprint, September 2026 · §8. Integral generators and the dimension squeeze

Lemma 8.1 [Depth addition]: With generators x,yx,y and m∈ℕm\in\mathbb N, brackets of at least two positive-depth inputs have zero coefficients at every xmyx^m y

The Ihara bracket respects ordinary depth.

Hover over a blue or green term in the text, or a node in the graph, to find it in the other. Select it to read its declaration both in Lean and in words.

The Setting

For iterated Ihara brackets of generators, the question concerns the coefficients at the words xmyx^m y in their associative expansions. The library states its depth estimates over an arbitrary commutative coefficient ring.

Let SS be a commutative ring and α\alpha an alphabet. Write LS(α)L_S(\alpha) for the ordinary free Lie algebra over SS on α\alpha, and LS=LS({x,y})L_S=L_S(\{x,y\}).✦ The free associative algebra AS(α)A_S(\alpha) is the monoid algebra of the free monoid α*\alpha^* of words.✦ For a word w∈α*w\in\alpha^*, the associative word aw∈AS(α)a_w\in A_S(\alpha) is the monomial with coefficient one at ww✦; for a letter a∈αa\in\alpha, the associative generator aa∈AS(α)a_a\in A_S(\alpha) is the monomial corresponding to the one-letter word aa.✦ The universal associative embedding ι\iota is the unique Lie homomorphism from LS(α)L_S(\alpha) to AS(α)A_S(\alpha) extending the generator assignment in (1).✦ AS(α)=⨁w∈α*Saw,auav=auv,ι(a)=aa.A_S(\alpha)=\bigoplus_{w\in\alpha^*} S a_w, \qquad a_u a_v=a_{uv}, \qquad \iota(a)=a_a.(1)

For a condition PP on words, the prescribed word span VS(P)V_S(P) is the submodule of AS(α)A_S(\alpha) spanned by the associative words satisfying PP.✦ A letter-weight function wt:α→ℕ\mathrm{wt}:\alpha\to\mathbb N gives a word its weight by summing the weights of its letters.✦ The associated filtration of AS(α)A_S(\alpha) consists of the submodules Fwt≥nF_{\mathrm{wt}}^{\ge n} spanned by words of weight at least nn, as in (2).✦ VS(P)=span⁡S{aw:P(w)},|w|wt=∑a in wwt(a),Fwt≥n=span⁡S{aw:|w|wt≥n}.\begin{aligned} V_S(P)&=\operatorname{span}_S\{a_w:P(w)\},\\ |w|_{\mathrm{wt}}&=\sum_{a\text{ in }w}\mathrm{wt}(a),\\ F_{\mathrm{wt}}^{\ge n} &=\operatorname{span}_S\{a_w:|w|_{\mathrm{wt}}\ge n\}. \end{aligned}(2) The depth letter weight dd on {x,y}\{x,y\} assigns weight zero to xx and weight one to yy✦, and the depth-one word wmw_m in {x,y}*\{x,y\}^* consists of mm copies of xx followed by yy, for m∈ℕm\in\mathbb N, as in (3).✦ Write Fd≥rF_d^{\ge r} for the resulting filtration, and [w]u[w]u for the coefficient of the associative word awa_w in u∈AS({x,y})u\in A_S(\{x,y\}). d(x)=0,d(y)=1,wm=xmy(m∈ℕ).d(x)=0,\qquad d(y)=1,\qquad w_m=x^m y\quad(m\in\mathbb N).(3)

A bracketed word tt on α\alpha is a single letter or the bracketing of two bracketed words.✦ The underlying word of a bracketed word tt on α\alpha is the word in α*\alpha^* that lists the letters of tt from left to right.✦ The bracket evaluation e(t)e(t) of a bracketed word tt on α\alpha is the element of LS(α)L_S(\alpha) obtained by replacing each letter by its free generator and each bracketing by the ordinary Lie bracket.✦ For letter weights wt\mathrm{wt}, the tree weight |t|wt|t|_{\mathrm{wt}} of a bracketed word tt on α\alpha is the sum of the weights of its letters✦, and the homogeneous Lie component Hnwt(α)H_n^{\mathrm{wt}}(\alpha) of LS(α)L_S(\alpha) is the submodule spanned by bracket evaluations of tree weight nn, as in (4).✦ |t|wt=∑a in the underlying word of twt(a),Hnwt(α)=span⁡S{e(t):|t|wt=n}.|t|_{\mathrm{wt}}=\sum_{a\text{ in the underlying word of }t}\mathrm{wt}(a), \qquad H_n^{\mathrm{wt}}(\alpha) =\operatorname{span}_S\{e(t):|t|_{\mathrm{wt}}=n\}.(4) With every letter assigned weight one, write Hn1(α)H_n^1(\alpha) for the component of bracket length nn.

Derivations and Ihara Evaluation

A first-order Lie jet in an SS-Lie algebra MM is a pair (u,v)(u,v) of elements of MM, its base point and velocity.✦ The Lie bracket on the algebra J1MJ^1M of first-order Lie jets in MM is given by (5). J1M=M⊕M,[(u,v),(u′,v′)]=([u,u′],[u,v′]−[u′,v]).J^1M=M\oplus M,\qquad [(u,v),(u',v')]=([u,u'],[u,v']-[u',v]).(5) For an assignment f:α→LS(α)f:\alpha\to L_S(\alpha), the jet lift jfj_f is the unique Lie homomorphism from LS(α)L_S(\alpha) to J1LS(α)J^1L_S(\alpha) taking each generator a∈αa\in\alpha to (a,f(a))(a,f(a)).✦ For an assignment f:α→LS(α)f:\alpha\to L_S(\alpha), the free derivation extension ∂f\partial_f is the unique SS-linear Lie derivation of LS(α)L_S(\alpha) with generator values ff; its value is the velocity in (6).✦ jf:LS(α)→J1LS(α),jf(q)=(q,∂f(q)),∂f(a)=f(a).j_f:L_S(\alpha)\longrightarrow J^1L_S(\alpha),\qquad j_f(q)=(q,\partial_f(q)),\qquad \partial_f(a)=f(a).(6)

For p∈LSp\in L_S, the Ihara derivation DpD_p is the free derivation extension with generator values in (7).✦ The Ihara bracket {p,q}\{p,q\} of elements p,q∈LSp,q\in L_S is the operation on LSL_S given by (7); [,][\ ,\ ] denotes the ordinary bracket.✦ Dp(x)=0,Dp(y)=[y,p],{p,q}=Dp(q)−Dq(p)+[p,q].D_p(x)=0,\qquad D_p(y)=[y,p],\qquad \{p,q\}=D_p(q)-D_q(p)+[p,q].(7) The ambient semidirect product 𝔞S\mathfrak a_S uses the natural action of SS-linear Lie derivations on LSL_S. The Ihara Lie algebra 𝔤S\mathfrak g_S is the derivation graph, with graph projection ϵ\epsilon as in (8). 𝔞S=LS⋊Der⁡S(LS),𝔤S={(p,Dp):p∈LS},ϵ(p,Dp)=p.\mathfrak a_S=L_S\rtimes\operatorname{Der}_S(L_S),\qquad \mathfrak g_S=\{(p,D_p):p\in L_S\},\qquad \epsilon(p,D_p)=p.(8) For an assignment f:α→LSf:\alpha\to L_S, the universal Ihara lift f̃:LS(α)→𝔤S\widetilde f:L_S(\alpha)\to\mathfrak g_S is the unique Lie homomorphism with the generator values in (9).✦ The Ihara evaluation ev⁡fI\operatorname{ev}^{\mathrm I}_f associated with an assignment f:α→LSf:\alpha\to L_S is the map from LS(α)L_S(\alpha) to LSL_S obtained by composing the universal Ihara lift f̃\widetilde f with the graph projection ϵ\epsilon.✦ f̃(a)=(f(a),Df(a)),ev⁡fI=ϵ∘f̃.\widetilde f(a)=(f(a),D_{f(a)}),\qquad \operatorname{ev}^{\mathrm I}_f=\epsilon\circ\widetilde f.(9)

The Claim

Theorem 1 (Depth addition). Let SS be a commutative ring. Let p,q∈LSp,q\in L_S and r,s∈ℕr,s\in\mathbb N satisfy ι(p)∈Fd≥r\iota(p)\in F_d^{\ge r} and ι(q)∈Fd≥s\iota(q)\in F_d^{\ge s}. Their Ihara bracket satisfies (10): ι({p,q})∈Fd≥r+s.\iota(\{p,q\})\in F_d^{\ge r+s}.(10)✦ Let α\alpha be an alphabet and f:α→LSf:\alpha\to L_S an assignment with ι(f(a))∈Fd≥1\iota(f(a))\in F_d^{\ge1} for every a∈αa\in\alpha. For n∈ℕn\in\mathbb N and P∈Hn1(α)P\in H_n^1(\alpha), Ihara evaluation satisfies (11): ι(ev⁡fI(P))∈Fd≥n.\iota\bigl(\operatorname{ev}^{\mathrm I}_f(P)\bigr)\in F_d^{\ge n}.(11)✦ Let p∈LSp\in L_S and r>1r>1 satisfy ι(p)∈Fd≥r\iota(p)\in F_d^{\ge r}. Its coefficients satisfy (12) for every m∈ℕm\in\mathbb N: [xmy]ι(p)=0.[x^m y]\iota(p)=0.(12)✦

What Is Checked

The estimates are stated over ordinary free Lie algebras over a commutative ring SS, with depth measured in their associative expansions and bracket length measured by Hn1(α)H_n^1(\alpha). The derivation graph 𝔤S\mathfrak g_S is a Lie subalgebra of 𝔞S\mathfrak a_S, and its projection ϵ\epsilon is an SS-linear isomorphism with inverse p↦(p,Dp)p\mapsto(p,D_p).✦