# Reading roadmap: FI-modules and stability for representations of symmetric groups

The paper develops finite generation and representation stability for functors on finite sets and injections, adds partial injections for all-size formulas, and applies these structures to algebras, configurations and moduli.

Source SHA-256: `8292fa4281823f39db042337018993dc98ac87de537aa7cd9d540f378ee26f59`. Indexed 246,965 characters.

Edges point from prerequisite to consumer. Solid proof edges are attributed inferences, not formal certificates; references remain candidates.

## Internal claim index

### Definition 1.1 — `def:fimod`

Source lines 265–268.

{\bf FI-module}

An FI-module over a commutative ring $k$ is a functor $V$ from $\FI$ to the category of $k$-modules. We denote the $k$-module $V(\n)$ by $V_n$.

*Extractive statement; proof gist not independently reconstructed.*

### Definition 1.2 — `introdefinition:line:274:9276`

Source lines 274–276.

{\bf Finite generation}

An FI-module $V$ is finitely generated if there is a finite set $S$ of elements in $\coprod_iV_i$ so that no proper sub-FI-module of $V$ contains $S$; see Definition~\ref{def:finitegen}.

*Extractive statement; proof gist not independently reconstructed.*


Candidate references: `def:finitegen` (internal_reference).

### Theorem 1.3 — `th:noetherian`

Source lines 279–283.

{\bf Noetherian property}

Let $k$ be a Noetherian ring containing $\Q$. The category of FI-modules over $k$ is Noetherian: that is, any sub-FI-module of a finitely generated FI-module is finitely generated.

*Extractive statement; proof gist not independently reconstructed.*

Proof opening: Let $k$ be a Noetherian ring, let $V$ be a finitely generated FI-module over $k$ generated in degree $\leq a$, and let $W$ be a sub-FI-module of $V$; our goal is to prove that $W$ is finitely generated. Proposition~\ref{prop:charfg} implies that $V_n$ is a finitely generated $k$-module for each $n$. Since $k$ is Noetherian, its submodule $W_n$ is finitely generated as well. The functor $\Phi_a$ of Definition~\ref{def:Phia} is exact over rings containing $\Q$, so $\Phi_a(W)\subset \Phi_a(V)$. The graded $k[T]$-module $\Phi_a(V)$ is finitely generated by Proposition~\ref{pr:surjdeg}. Since $k[T]$ is a Noetherian ring, its submodule $\Phi_a(W)$ is finitely generated as a $k[T]$-module (equivale


Candidate references: `prop:charfg` (candidate_proof_dependency); `def:Phia` (candidate_proof_dependency); `pr:surjdeg` (candidate_proof_dependency); `pr:fgweight` (candidate_proof_dependency); `lem:coinvariants` (candidate_proof_dependency).

### Definition 1.4 — `introdefinition:line:347:15255`

Source lines 347–353.

{\bf Eventually polynomial characters}

A sequence $\chi_n$ of characters of $S_n$ is eventually polynomial if there exist integers $r$, $N$ and a character polynomial $P(X_1,\ldots,X_r)$ such that \[ \qquad\qquad\chi_n(\sigma)=P(X_1,\ldots ,X_r)(\sigma)\quad \ \ {\rm for \ all}\ n\geq N\text{ and all }\sigma\in S_n. \] The degree of the character polynomial $P(X_1,\ldots,X_r)$ is defined by setting $\deg(X_i)=i$.

*Extractive statement; proof gist not independently reconstructed.*

### Theorem 1.5 — `thm:intro:persinomial`

Source lines 357–362.

{\bf Polynomiality of characters}

Let $V$ be an FI-module over a field of characteristic~$0$. If $V$ is finitely generated then the sequence of characters $\chi_{V_n}$ of the $S_n$-representations $V_n$ is eventually polynomial. In particular, $\dim V_n$ is eventually polynomial.

*Extractive statement; proof gist not independently reconstructed.*

### Corollary 1.6 — `cor:polydim:app`

Source lines 373–376.

The characters $\chi_{V_n}$ of each of the sequences (1)--(9) of $S_n$-representations in Table~\ref{table:examples} are eventually polynomial. In particular $\dim(V_n)$ is eventually polynomial.

*Extractive statement; proof gist not independently reconstructed.*


Candidate references: `table:examples` (internal_reference).

### Theorem 1.7 — `thm:intro:FIsharp`

Source lines 394–401.

Let $V$ be an $\FIsharp$-module over any field $k$. The following are equivalent: \item $\dim(V_n)$ is bounded above by a polynomial in $n$. \item $\dim(V_n)$ is exactly equal to a polynomial in $n$ for all $n\geq 0$.

*Extractive statement; proof gist not independently reconstructed.*

### Theorem 1.8 — `thm:confncharpoly`

Source lines 421–426.

If $\dim M\geq 3$, there is a character polynomial $P_{M,i}$ of degree $\leq i$ so that \[\chi_{H^i(\Conf_n(M);\Q)}(\sigma)=P_{M,i}(\sigma)\quad\text{for all }n\geq 2i\text{ and all }\sigma\in S_n.\] In particular, the Betti number $b_i(\Conf_n(M))$ agrees with a polynomial of degree $i$ for all $n\geq 2i$.

*Extractive statement; proof gist not independently reconstructed.*

### Theorem 1.9 — `thm:intro:openbetti`

Source lines 430–439.

If $M$ is the interior of a compact manifold with nonempty boundary, each of the following invariants of $\Conf_n(M)$ is given by a polynomial in $n$ for all $n\geq 0$ (of degree $i$ if $\dim M\geq 3$, and of degree $2i$ if $\dim M=2$): \item the $i$-th rational Betti number $b_i(\Conf_n(M))$, i.e.\ $\dim_\Q\Conf_n(M;\Q)$; \item the $i$-th mod-$p$ Betti number of $\Conf_n(M)$, i.e.\ $\dim_{\F_p}\Conf_n(M;\F_p)$; \item the minimum number of generators of $H^i(\Conf_n(M);\Z)$; \item the minimum number of generators of the $p$-torsion part of $H^i

*Extractive statement; proof gist not independently reconstructed.*

- `thm:ConfMFIsharpfg` (proof_dependency; source_audit_annotation): Theorem~\ref{thm:intro:openbetti} follows immediately from Theorem~\ref{thm:ConfMFIsharpfg} by applying the classification of $\FIsharp$-modules from Theorem~\ref{th:charFIsharp}, just as in the proof of Theorem~\ref{thm:polyFIsharp}.
- `th:charFIsharp` (proof_dependency; source_audit_annotation): Theorem~\ref{thm:intro:openbetti} follows immediately from Theorem~\ref{thm:ConfMFIsharpfg} by applying the classification of $\FIsharp$-modules from Theorem~\ref{th:charFIsharp}, just as in the proof of Theorem~\ref{thm:polyFIsharp}.

### Problem 1.10 — `problem:megaprob`

Source lines 467–471.

For each $r\geq 2, n\geq 1$, and $J=(j_1,\ldots ,j_r)$, compute the character of each $R^{(r)}_J(n)$ as an $S_n$-representation, at least for $n$ sufficiently large. In particular, find a formula for $\dim(R^{(r)}_J(n))$.

*Extractive statement; proof gist not independently reconstructed.*

### Theorem 1.11 — `theorem:intro:polynomial:answer`

Source lines 488–492.

{\bf Characters of $R^{(r)}_J(n)$ are eventually polynomial in $n$}

For any fixed $r\geq 1$ and $J=(j_1,\ldots ,j_r)$, the characters $\chi_{R^{(r)}_J(n)}$ are eventually polynomial in $n$ of degree at most $|J|$. In particular there exists a polynomial $P_J^{(r)}(n)$ of degree at most $|J|$ such that \[\dim(R^{(r)}_J(n)) = P^{(r)}_J(n)\text{ for all }n\gg 0.\]

*Extractive statement; proof gist not independently reconstructed.*

Proof opening: For fixed $r\geq 1$, let $R^{(r)}$ be the graded co-FI-algebra $R^{(r)}\coloneq \coinv(\Q[M(1)^{\oplus r}])$ from Definition~\ref{def:generalcoinvariantalgebra}, with $M(1)^{\oplus r}$ concentrated in grading 1. To the set $\n$ this associates $R^{(r)}_n=\coinv(\Q[M(1)_n^{\oplus r}])$, which is naturally isomorphic to the diagonal coinvariant algebra \[ R^{(r)}(n)\coloneq \Q[x_1^{(1)},\ldots ,x_n^{(1)},\ldots ,x_1^{(r)},\ldots ,x_n^{(r)}]/\big(\Q[x_1^{(1)},\ldots ,x_n^{(1)},\ldots ,x_1^{(r)},\ldots ,x_n^{(r)}]_{>0}^{S_n}\big) \] defined in the introduction. By Proposition~\ref{pr:Mmuweight} we have $\wt(M(1)^{\oplus r})=1$, so Proposition~\ref{pr:cofisubalgebra} implies that $\Q[M(1)^{\oplus


Candidate references: `def:generalcoinvariantalgebra` (candidate_proof_dependency); `pr:Mmuweight` (candidate_proof_dependency); `pr:cofisubalgebra` (candidate_proof_dependency); `pr:coinvariantfinite` (candidate_proof_dependency); `thm:persinomial` (candidate_proof_dependency).

### Problem 1.12 — `introprob:line:497:29645`

Source lines 497–499.

{\bf Character polynomials for diagonal coinvariant algebras}

Describe the polynomials $P^{(r)}_{J}(n)$ whose existence is guaranteed by Theorem~\ref{theorem:intro:polynomial:answer}.

*Extractive statement; proof gist not independently reconstructed.*


Candidate references: `theorem:intro:polynomial:answer` (internal_reference).

### Theorem 1.13 — `theorem:intro:FIequiv`

Source lines 520–526.

{\bf Finite generation vs.\ representation stability}

An FI-module $V$ over a field of characteristic~0 is finitely generated if and only if the sequence $\{V_n\}$ of $S_n$-representations is uniformly representation stable in the sense of \cite{CF} and each $V_n$ is finite-dimensional. In particular, for any finitely generated FI-module $V$, we have for sufficiently large $n$ a decomposition: \[V_n\simeq \bigoplus c_\lambda V(\lambda)_n\] where the coefficients $c_\lambda$ do not depend on $n$.

*Extractive statement; proof gist not independently reconstructed.*

- `th:noetherian` (proof_dependency; source_audit_annotation): Assume that $V$ is finitely generated, say in degree $\leq g$. We have $\surjD(V)\leq g$ by Proposition~\ref{pr:surjdeg}, and we would like to bound $\injD(V)$ as well. By Lemma~\ref{lemma:Mspan} there is a free FI-module $M\coloneq \bigoplus_{i=0}^g M(i)^{\oplus b_i}$ with a surjection $M\onto V$; let $K$ be the kernel of this map. By Theorem~\ref{th:noetherian}, the submodule $K$ of $M$ is finitely generated, say in degree $\leq r$. Since $\Phi_a$ is exact over rings containing $\Q$, we have $\Phi_a(V)_n\simeq \Phi_a(M)_n / \Phi_a(K)_n$. Proposition~\ref{pr:stabdegprojective} states that $\injD(M)=0$, so the maps $T\colon \Phi_a(M)_n\to \Phi_a(M)_{n+1}$ are injective for all $n\geq 0$.  Proposition~\ref{pr:surjdeg} implies $\surjD(K)\leq r$, so for $n\geq r$ the maps $T\colon \Phi_a(K)_n\to \Phi_a(K)_{n+1}$ are surjective. We conclude that for $n\geq r$ the maps $T\colon \Phi_a(V)_n\to \Phi_a(V)_{n+1}$ are injective, or in other words that $\injD(V)\leq r$. Therefore $\stabD(V)\leq N\coloneq \max(g,r)$, so Proposition~\ref{pr:stabdegstabrange} implies that $\{V_n,\phi_n\}$ is uniformly representation stable in degrees $\geq N+g$.

Candidate references: `external:CF` (external_reference).

### Remark 1.A — `alphintroremark:line:569:40300`

Source lines 569–571.

The arXiv version of this paper is slightly expanded from the published version that will appear in Duke Math.\ J. All theorems, propositions, remarks, and so on are numbered consistently across both versions, so that for example Theorem~\ref{thm:persinomial} refers to the same theorem in both versions; those that appear only in the arXiv version, such as Remark~\ref{rem:projective}, are numbered separately.

*Extractive statement; proof gist not independently reconstructed.*


Candidate references: `thm:persinomial` (internal_reference); `rem:projective` (internal_reference).

### Definition 2.1.1 — `definition:line:616:49689`

Source lines 616–618.

The category of FI-modules is the category $\FIMod\coloneq [\FI,\kMod]$ of functors from $\FI$ to the category of $k$-modules, with natural transformations of functors as morphisms.

*Extractive statement; proof gist not independently reconstructed.*

### Remark 2.1.2 — `remark:line:622:50374`

Source lines 622–629.

$\FIMod$ inherits the structure of an abelian category from $\kMod$, with all notions such as kernel, cokernel, subobject, quotient object, injection, and surjection being defined ``pointwise'' from the corresponding notions in $\kMod$. This is true for any category of functors from any small category to an abelian category \cite[A.3.3]{We}. In particular, the same holds for $\FB$-modules, $\FIsharp$-modules, etc.; see Remark~\ref{remark:FInotation} below. For example, a map $F\colon V\to W$ of FI-modules is a surjection if and only if the maps

*Extractive statement; proof gist not independently reconstructed.*


Candidate references: `remark:FInotation` (internal_reference); `external:We` (external_reference).

### Remark 2.1.3 — `rem:nonsemisimple`

Source lines 631–636.

In some sense the category of FI-modules might be thought of as a ``limit'' of the category of representations of $S_n$ as $n \ra \infty$. But care is necessary. For example, the category $\FIMod$ is not semisimple, even when $k$ is a field of characteristic~$0$. Consider the FI-module $V$ with $V_S=k$ for all finite sets $S$ and $V_f=\id$ for all injections $f$. Let $W$ be the FI-module with $W_\emptyset=k$ and $W_S=0$ if $S\neq \emptyset$, with $W_f=0$ except for $f\colon \emptyset\to \emptyset$. There is an obvious surjection of FI-modules $

*Extractive statement; proof gist not independently reconstructed.*

### Remark 2.1.4 — `remark:FInotation`

Source lines 641–644.

We use the same notation for other categories: for example, an FB-module is a functor from $\FB$ to the category of $k$-modules; an FI-group is a functor from $\FI$ to the category of groups; an $\FIsharp$-space is a functor from $\FIsharp$ to the category of topological spaces, and so on. In all these cases, we take natural transformations as our morphisms, so $\FBMod\coloneq [\FB,\kMod]$ is the category of FB-modules, $\FIGroups\coloneq [\FI,\Groups]$ is the category of FI-groups, and so on.

*Extractive statement; proof gist not independently reconstructed.*

### Remark 2.1.5 — `remark:line:646:54043`

Source lines 646–648.

By a graded $k$-module we always mean an $\N$-graded $k$-module. In keeping with the approach of this paper, we can think of a graded $k$-module as a functor $A\colon \N\to \kMod$, where $\N$ is the discrete category with objects $\N$, so that $A_i=A(i)$ is the part of $A$ in grading $i$. A morphism $A\to B$ of graded $k$-modules is a natural transformation, i.e.\ a collection of homomorphisms $A_i\to B_i$ (so we only consider morphisms of ``degree 0''). To avoid phrases such as ``FI-(graded $k$-module)'', we will say instead ``graded FI-module

*Extractive statement; proof gist not independently reconstructed.*

### Remark 2.1.6 — `remark:FIalgebras`

Source lines 650–653.

A graded $k$-algebra is a graded $k$-module $A$ endowed with a unital multiplication $A\otimes A\to A$. We require that this multiplication be unital, but it need not be associative. An ideal $I\subset A$ is a graded submodule $I\subset A$ such that under multipliciation $A\otimes I$ and $I\otimes A$ map into $I$; in this case, the quotient $A/I$ is a graded $k$-algebra. In particular, a graded FI-algebra $V$ is a functor from $\FI$ to the category of graded $k$-algebras, and an FI-ideal $I\subset V$ yields for each finite set $S$ an ideal $I_S

*Extractive statement; proof gist not independently reconstructed.*

### Definition 2.2.1 — `definition:line:658:55707`

Source lines 658–660.

{\bf FB-modules}

Let $\FB$ denote the category of finite sets and bijections. An FB-module $W$ is a functor $W\colon \FB\to \kMod$.

*Extractive statement; proof gist not independently reconstructed.*

### Definition 2.2.2 — `def:MW`

Source lines 666–672.

{\bf The FI-module {\boldmath$M(W)$}}

We define the functor $M(-)\colon \FBMod \to \FIMod$ as the left adjoint of $\pi\colon \FIMod\to \FBMod$. Explicitly, if $W$ is an FB-module, then by \cite[(Eq.\ X.3.10)]{MacL} the FI-module $M(W)$ satisfies M(W)_S=\colim_{\substack{(T\in \FB,\\f\colon T\into S)}} W_T=\bigoplus_{T\subseteq S} W_T, with the map $f_*\colon M(W)_S\to M(W)_{S'}$ induced by $f\colon S\into S'$ being the sum of $(f|_T)_*\colon W_T\to W_{f(T)}$.

*Extractive statement; proof gist not independently reconstructed.*


Candidate references: `external:MacL` (external_reference).

### Definition 2.2.3 — `def:Mm`

Source lines 687–690.

The FI-module \boldmath$M(m)$

For any $m\geq 0$, we denote by $M(m)$ the ``representable'' FI-module $k[\Hom_{\FI}(\m,-)]\simeq M(k[S_m])$. For any finite set $S$, the $k$-module $M(m)_S$ has basis indexed by the injections $\m\into S$; in other words, by ordered sequences of $m$ distinct elements of $S$. A map of FI-modules $F\colon M(m)\to V$ is determined by the choice of an element $v=F(\id_\m)\in V_m$, via the adjunction defining $M(-)$.

*Extractive statement; proof gist not independently reconstructed.*

### Example 2.2.4 — `xample:line:692:58975`

Source lines 692–694.

Since there is precisely one injection $\emptyset =\mathbf{0}\into S$ for every finite set $S$, the FI-module $M(0)$ is the constant functor with value $k$; as an $S_n$-representation $M(0)_n\simeq k$ is the $1$-dimensional trivial representation. Since injections $\mathbf{1}\into S$ correpond simply to elements of $S$, the FI-module $M(1)$ assigns to a finite set $S$ the free $k$-module with basis $S$; as an $S_n$-representation $M(1)_n\simeq k^n$ is the usual permutation representation.

*Extractive statement; proof gist not independently reconstructed.*

### Definition 2.2.5 — `def:Vlambdan`

Source lines 709–716.

{\bf The irreducible representation $V(\lambda)_n$}

Given a partition $\lambda$, for any $n\geq \abs{\lambda}+\lambda_1$ we define the padded partition \[\lambda[n]\coloneq (n-\abs{\lambda},\lambda_1,\ldots,\lambda_\ell).\] For $n\geq \abs{\lambda}+\lambda_1$, we define $V(\lambda)_n$ to be the irreducible $S_n$--representation \[V(\lambda)_n\coloneq V_{\lambda[n]}.\]

*Extractive statement; proof gist not independently reconstructed.*

### Definition 2.2.6 — `definition:line:722:62198`

Source lines 722–724.

{\bf The FI-module {\boldmath$M(\lambda)$}}

When $k$ is a field of characteristic~0, given a partition $\lambda$ we write $M(\lambda)$ for the FI-module $M(\lambda)\coloneq M(V_\lambda)$.

*Extractive statement; proof gist not independently reconstructed.*

### Remark 2.2.A — `rem:projective`

Source lines 728–733.

{\bf Projective FI-modules}

A projective FI-module is a projective object in the abelian category $\FIMod$. It follows from general considerations (see \cite[Exercise 2.3.8]{We}) that the category $\FIMod$ has enough projectives. Indeed if $P_n$ is a projective $k[S_n]$-module, then $M(P_n)$ is projective, since $M(-)$ is the left adjoint to an exact functor and thus preserves projectives \cite[Proposition 2.3.10]{We}. In fact, it is not hard to show that every projective FI-module is a direct sum $\bigoplus_{n\geq 0} M(P_n)$ where $P_n$ is a projective $k[S_n]$-module; s

*Extractive statement; proof gist not independently reconstructed.*


Candidate references: `external:We` (external_reference); `external:Lue` (external_reference).

### Definition 2.3.1 — `definition:line:739:63746`

Source lines 739–741.

{\bf Span}

If $V$ is an FI-module and $\Sigma$ is a subset of the disjoint union $\coprod V_n$, the span $\spn_V(\Sigma)$ is the minimal sub-FI-module of $V$ containing each element of $\Sigma$. We say that $\spn_V(\Sigma)$ is the sub-FI-module of $V$ generated by $\Sigma$. We sometimes write $\spn(\Sigma)$ when there is no ambiguity. We write $\spn(V_{\leq m})$ for $\spn\big(\coprod_{k\leq m} V_k\big)$ and $\spn(V_{< m})$ for $\spn\big(\coprod_{k< m} V_k\big)$.

*Extractive statement; proof gist not independently reconstructed.*

### Lemma 2.3.2 — `lemma:Mspan`

Source lines 743–746.

Given an FI-module $V$, a choice of element $v\in V_m$ determines a map $M(m)\to V$. The image of this map is $\spn_V(v)$. More generally, if $\Sigma$ is the disjoint union of $\Sigma_n\subset V_n$, the image of the natural map $\bigoplus_{n\geq 0} M(n)^{\oplus \Sigma_n}\to V$ is $\spn_V(\Sigma)$.

*Extractive statement; proof gist not independently reconstructed.*

Proof opening: Let $W$ be the image of the map $M(m)\to V$ sending $\id_\m\in M(m)_m$ to $v\in V_m$. Certainly $v\in W_m$, so $\spn_V(v)\subset W$. A basis for $M(m)_S$ consists of the morphisms $f\colon \m\into S$, and $f\in M(m)_S$ is sent to $f_*(v)\in V_S$ (as it must be for $M(m)\to V$ to be a map of FI-modules). Thus $W_S$ is the submodule of $V_S$ spanned by the elements $f_*(v)$, where $f$ ranges over injections $f\colon \m\into S$. Since any sub-FI-module of $V$ containing $v$ must contain all these elements $f_*(v)$, we have $W\subset \spn_V(v)$ as well. This completes the proof for $\spn_V(v)$; the argument for general $\Sigma$ is identical.

### Definition 2.3.3 — `definition:line:754:65611`

Source lines 754–756.

{\bf Generation in degree {\boldmath $\leq m$}}

We say that an FI-module $V$ is generated in degree $\leq m$ if $V$ is generated by elements of $V_k$ for $k\leq m$, i.e.\ if $\spn(V_{\leq m})=V$.

*Extractive statement; proof gist not independently reconstructed.*

### Definition 2.3.4 — `def:finitegen`

Source lines 761–764.

{\bf Finite generation}

We say that an FI-module $V$ is finitely generated if there is a finite set of elements $v_1,\ldots,v_k$ with $v_i\in V_{m_i}$ which generates $V$, meaning that $\spn(v_1,\ldots,v_k)=V$.

*Extractive statement; proof gist not independently reconstructed.*

### Proposition 2.3.5 — `prop:charfg`

Source lines 767–769.

{\bf Finite generation in terms of \boldmath$M(m)$}

An FI-module $V$ is finitely generated if and only if it admits a surjection $\bigoplus_i M(m_i)\onto V$ for some finite sequence of integers $\{m_i\}$.

*Extractive statement; proof gist not independently reconstructed.*

### Proposition 2.3.6 — `pr:tensor`

Source lines 773–776.

{\bf Tensor products of f.g.\ FI-modules}

If $V$ and $W$ are finitely generated FI-modules, so is $V \tensor W$. If $V$ is generated in degree $\leq m_1$ and $W$ is generated in degree $\leq m_2$, then $V \tensor W$ is generated in degree $\leq m_1+m_2$.

*Extractive statement; proof gist not independently reconstructed.*

Proof opening: By Proposition~\ref{prop:charfg}, it suffices to show that $U\coloneq M(m_1) \tensor M(m_2)$ is finitely generated in degree $\leq m_1 + m_2$. Each $k$-module $U_n$ is certainly finitely generated (of rank $\binom{n}{m_1}\cdot\binom{n}{m_2}$), so it suffices to prove that $U=\spn(U_{\leq m_1+m_2})$. A basis for $U_S$ is given by pairs $(f_1\colon \m_1\into S, f_2\colon \m_2\into S)$. The basis element $(f_1,f_2)$ lies in the image of $U_T$, where $T=\im(f_1)\cup \im(f_2)$. Since $\abs{T}$ is at most $m_1+m_2$, this shows that $\spn(U_{\leq m_1+m_2})$ is all of $U$, as desired.


Candidate references: `prop:charfg` (candidate_proof_dependency).

### Definition 2.3.7 — `def:H0`

Source lines 785–792.

{\bf The functor ${H_0}$}

We define the functor $H_0\colon \FIMod\to \FBMod$ as the left adjoint of this inclusion $\FBMod\into \FIMod$. Explicitly, given an FI-module $V$, the FB-module $H_0(V)$ satisfies H_0(V)_S\coloneq V_S/\spn(V_{<\abs{S}})_S; given a bijection $f\colon S\to S'$, the map $f_*\colon H_0(V)_S\to H_0(V)_{S'}$ is that induced by $f_*\colon V_S\to V_{S'}$.

*Extractive statement; proof gist not independently reconstructed.*

### Remark 2.3.8 — `remark:H0`

Source lines 793–798.

An FI-module $V$ is generated in degree $\leq m$ if and only if $H_0(V)_n$ vanishes for $n>m$. Similarly, $V$ is finitely generated if and only if the $k$-module $\bigoplus_{n\geq 0} H_0(V)_n$ is finitely generated. If $W_a$ is a $k[S_a]$-representation, the FI-module $M(W_a)$ is generated by $M(W_a)_a\simeq W_a$, so $H_0(M(W_a))_n$ vanishes for all $n\neq a$, and we also have $H_0(M(W_a))_a\simeq W_a$. More generally, for any FB-module $W$ we have a natural isomorphism $H_0(M(W))\simeq W$.

*Extractive statement; proof gist not independently reconstructed.*

### Remark 2.3.9 — `remark:line:800:69509`

Source lines 800–803.

The functor $H_0\colon \FIMod\to \FBMod$ is right-exact but not exact. In the paper \cite{CE} we investigate the higher FI-homology $H_i(V)$ of an FI-module $V$, namely the left-derived functors of $H_0$. In particular, we show that the vanishing of higher FI-homology is connected with the existence of inductive presentations for $V$, and apply this to homological stability for FI-groups.

*Extractive statement; proof gist not independently reconstructed.*


Candidate references: `external:CE` (external_reference).

### Definition 3.1.1 — `definition:line:821:71862`

Source lines 821–823.

A graded $k[T]$-module $U$ consists of a collection of $k$-modules $U_i$ for each $i\in \N$, endowed with a map $T\colon U_i\to U_{i+1}$ for each $i\in \N$ (i.e.\ $T$ acts by a map of grading 1).

*Extractive statement; proof gist not independently reconstructed.*

### Definition 3.1.2 — `def:Phia`

Source lines 827–833.

{\bf The graded $k[T]$-module $\Phi_a(V)$}

Given $a\geq 0$ and an FI-module $V$, we define the graded $k[T]$-module $\Phi_a(V)$ as follows. For $n\in \N$, let $\Phi_a(V)_n$ be the coinvariant quotient \[\Phi_a (V)_n\coloneq (V_{\na\disjoint \n})_{S_n}.\] To define $T\colon \Phi_a(V)_n\to \Phi_a(V)_{n+1}$, choose any injection $f\colon \n\into \npone$. This determines an injection $\id\disjoint f\colon \na\disjoint \n\into \na\disjoint \npone$, and thus a map $(\id\disjoint f)_*\colon V_{\na\disjoint \n}\to V_{\na\disjoint \npone}$. The map $T\colon \Phi_a(V)_n\to \Phi_a(V)_{n+1}$ is def

*Extractive statement; proof gist not independently reconstructed.*

### Definition 3.1.3 — `def:stabilitydegree`

Source lines 837–841.

{\bf Stability degree}

The stability degree $\stabD(V)$ of an FI-module $V$ is the smallest $s\geq 0$ such that for all $a\geq 0$, the map $T\colon \Phi_a(V)_n\to \Phi_a(V)_{n+1}$ is an isomorphism for all $n \geq s$. (If no such $s$ exists, we take $\stabD(V)=\infty$.)

*Extractive statement; proof gist not independently reconstructed.*

### Remark 3.1.4 — `remark:line:845:74473`

Source lines 845–847.

The permutations of $\na$ act on the functor $\Phi_a$, so $\Phi_a(V)_n$ is an $S_a$-representation; explicitly, $\Phi_a(V)_n\simeq (\Res^{S_{a+n}}_{S_a\times S_n}V_{a+n})_{S_n}$ as an $S_a$-representation. The map $T\colon \Phi_a(V)_n\to \Phi_a(V)_{n+1}$ is $S_a$-equivariant, so it is equivalent to ask in Definition~\ref{def:stabilitydegree} that $T$ be an isomorphism of $k$-modules or of $S_a$-representations. We make use of this equivalence in Proposition~\ref{pr:stabdegstabrange} and Corollary~\ref{co:vakilwood}~below.

*Extractive statement; proof gist not independently reconstructed.*


Candidate references: `def:stabilitydegree` (internal_reference); `pr:stabdegstabrange` (internal_reference); `co:vakilwood` (internal_reference).

### Definition 3.1.5 — `def:injectivitydegree`

Source lines 853–857.

We say that the injectivity degree $\injD(V)$ (resp.\ surjectivity degree $\surjD(V)$) of an FI-module $V$ is the smallest $s\geq 0$ such that for all $a\geq 0$, the map $\Phi_a(V)_n\to \Phi_a(V)_{n+1}$ induced by multiplication by $T$ is injective (resp.\ surjective) for all $n\geq s$. By definition, $\stabD(V)=\max(\injD(V),\surjD(V))$.

*Extractive statement; proof gist not independently reconstructed.*

### Lemma 3.1.6 — `lemma:injsurjdescend`

Source lines 860–863.

Let $V$ be an FI-module. Any quotient $V\onto W$ satisfies $\surjD(W)\leq \surjD(V)$. When $k$ contains $\Q$, any submodule $W\subset V$ satisfies $\injD(W)\leq \injD(V)$.

*Extractive statement; proof gist not independently reconstructed.*

Proof opening: Given $V\onto W$, the maps $\Phi_a(V)_n\onto \Phi_a(W)_n$ are surjective since $\Phi_a$ is right-exact. Thus if $T\colon \Phi_a(V)_n\to \Phi_a(V)_{n+1}$ is surjective, the same is true of $T\colon \Phi_a(W)_n\to \Phi_a(W)_{n+1}$. The proof for $W\subset V$ is identical, using that $\Phi_a$ is left-exact when $k$ contains $\Q$.

### Proposition 3.1.7 — `pr:stabdegprojective`

Source lines 870–873.

For any FB-module $W$, the FI-module $M(W)$ has $\injD(M(W))=0$; if $W_i=0$ for $i>m$, then $\surjD(M(W))\leq m$. The FI-module $M(m)$ has $\injD(M(m))=0$ and $\surjD(M(m))=\stabD(M(m))=m$.

*Extractive statement; proof gist not independently reconstructed.*

Proof opening: We will directly compute the graded $k[T]$-module $\Phi_a M(W)$. For any FB-module $W$, by \eqref{eq:MWdecomp} we have \[M(W)_{\na\disjoint\n}=\bigoplus_{T\subset\, \na\disjoint \n} W_T.\] Since $\Phi_a M(W)_n$ is by definition the $S_n$-coinvariants of $M(W)_{\na\disjoint\n}$, it splits as a direct sum over the orbits of $S_n$ acting on $\{T\subset \na\disjoint \n\}$. Such an orbit is determined by the intersection $T\cap \na$ (which can be any subset $U\subset \na$) together with the cardinality $\abs{T\cap \n}$ (which can be any integer $0\leq k\leq n$). As a representative for the orbit corresponding to $(U,k)$ we can take $T=U\disjoint \kk$. The stabilizer in $S_n$ of this set consists


Candidate references: `eq:MWdecomp` (candidate_proof_dependency); `def:H0` (candidate_proof_dependency); `eq:PhiaMW` (candidate_proof_dependency).

### Proposition 3.1.8 — `pr:surjdeg`

Source lines 894–897.

If $V$ is generated in degree $\leq d$ then $\surjD(V)\leq d$.

*Extractive statement; proof gist not independently reconstructed.*

Proof opening: If $V$ is generated in degree $\leq d$, Lemma~\ref{lemma:Mspan} shows that $V$ is a quotient of $\bigoplus_{m\leq d} M(V_m)$. The latter has surjectivity degree $\leq d$ by Proposition~\ref{pr:stabdegprojective}, so $\surjD(V)\leq d$ by Lemma~\ref{lemma:injsurjdescend}.


Candidate references: `lemma:Mspan` (candidate_proof_dependency); `pr:stabdegprojective` (candidate_proof_dependency); `lemma:injsurjdescend` (candidate_proof_dependency).

### Definition 3.2.1 — `de:weight`

Source lines 910–913.

{\bf Weight of an FI-module}

Let $V$ be an FI-module over a field of characteristic~0. The weight $\wt(V)$ of $V$ is the maximum of $\abs{\lambda}$ over all irreducible constituents $V(\lambda)_n$ occurring in the $S_n$-representations $V_n$. (If $V=0$ we set $\wt(V)=0$; if $\abs{\lambda}$ is unbounded then $\wt(V)=\infty$.)

*Extractive statement; proof gist not independently reconstructed.*

### Proposition 3.2.2 — `pr:tensorwt`

Source lines 915–918.

If $V$ and $W$ are FI-modules over a field of characteristic~0, the tensor product $V\otimes W$ satisfies $\wt(V\otimes W)\leq \wt(V)+\wt(W)$.

*Extractive statement; proof gist not independently reconstructed.*

### Lemma 3.2.3 — `lemma:branching`

Source lines 921–929.

{\bf Branching rule for $S_n$-representations}

Let $\lambda\vdash n$ be a partition, and $V_\lambda$ the corresponding irreducible $S_n$-representation over $\Q$. \item[\rm{(i)}] As $S_{n+k}$-representations, we have the decomposition \[\Ind_{S_n\times S_k}^{S_{n+k}} V_\lambda\boxtimes \Q\simeq \bigoplus_{\mu} V_\mu\] over those partitions $\mu\vdash n+k$ obtained from $\lambda$ by adding one box to $k$ different columns. \item[\rm{(ii)}] As $S_{n-k}$-representations, we have the decomposition \[(\Res^{S_n}_{S_{n-k}\times S_k}V_\lambda)_{S_k}\simeq\bigoplus_\nu V_\nu\] over those partitions

*Extractive statement; proof gist not independently reconstructed.*

### Proposition 3.2.4 — `pr:Mmuweight`

Source lines 931–934.

For any partition $\mu\vdash m$, the FI-module $M(\mu)$ over a field of characteristic~0 has $\wt(M(\mu))=m$.

*Extractive statement; proof gist not independently reconstructed.*

Proof opening: The identification \eqref{eq:MWinduced} shows that $M(\mu)_n=M(V_\mu)_n$ is isomorphic as an $S_n$-representation to $\Ind_{S_m\times S_{n-m}}^{S_n} V_\mu\boxtimes k$. Thus by Lemma~\ref{lemma:branching}(i), those $\nu\vdash n$ for which $V_\nu$ occurs in $M(\mu)_n$ are those obtained from $\mu$ by adding one box to $n-m$ different columns. Since $\nu_1$ is the number of columns of $\nu$, we must have $\nu_1\geq n-m$. When writing $\nu=\lambda[n]$ we have $\abs{\lambda}=n-\nu_1$, so we can rephrase this as saying that $\abs{\lambda}\leq m$ for any constituent $V(\lambda)_n$ of $M(\mu)_n$. This shows that $\wt(M(\mu))\leq m$. For any $n\geq m+\mu_1$, adding one box to the first $n-m$ columns


Candidate references: `eq:MWinduced` (candidate_proof_dependency); `lemma:branching` (candidate_proof_dependency).

### Proposition 3.2.5 — `pr:fgweight`

Source lines 941–944.

Let $V$ be an FI-module over a field of characteristic~0. If $V$ is generated in degree $\leq d$, then $\wt(V)\leq d$.

*Extractive statement; proof gist not independently reconstructed.*

### Proposition 3.2.6 — `pr:stabdegMlambda`

Source lines 949–953.

For any partition $\lambda$ the FI-module $M(\lambda)$ over a field of characteristic~0 has $\stabD(M(\lambda))=\lambda_1$.

*Extractive statement; proof gist not independently reconstructed.*

Proof opening: From our earlier computation of $\Phi_a M(V_\lambda)$ in \eqref{eq:PhiaMW} we have: \Phi_a M(V_\lambda)_n\simeq\bigoplus_{\substack{k\leq n\\U\subset \na}}\Phi_{\abs{U}}(V_\lambda)_k=\bigoplus_{\substack{k\leq n\\\abs{U}+k=\abs{\lambda}}}(V_\lambda)_{S_k} We saw in the proof of Proposition~\ref{pr:stabdegprojective} that $T\colon \Phi_a M(V_\lambda)_n\to \Phi_a M(V_\lambda)_{n+1}$ is the identity on each factor of this decomposition, so our goal is to show that no new factors occur in $\Phi_a M(V_\lambda)_n$ for $n>\lambda_1$. By Lemma~\ref{lem:coinvariants}(i), $(V_\lambda)_{S_k} = 0$ when $k>\lambda_1$. Thus we can add the condition $k\leq \lambda_1$ to \eqref{eq:PhiaMVlambda}; the conditi


Candidate references: `eq:PhiaMW` (candidate_proof_dependency); `pr:stabdegprojective` (candidate_proof_dependency); `lem:coinvariants` (candidate_proof_dependency); `eq:PhiaMVlambda` (candidate_proof_dependency).

### Lemma 3.2.7 — `lem:coinvariants`

Source lines 956–969.

Let $\mu$ be a partition of $n$, and write $\mu=\lambda[n]$ as in Definition~\ref{def:Vlambdan}. Given $a\leq n$, set $k\coloneq n-a$, and consider the $S_a$-representation $(V_\mu)_{S_k}=(V(\lambda)_n)_{S_{n-a}}$ over $\Q$. \item[\rm{(i)}] $(V_\mu)_{S_k}=0\iff k>\mu_1$; equivalently, $(V(\lambda)_n)_{S_{n-a}}=0 \iff a<\abs{\lambda}$. \item[\rm{(ii)}] If $k=\mu_1$ (i.e.\ if $a=\abs{\lambda}$), we have $(V_\mu)_{S_k}=(V(\lambda)_n)_{S_{n-a}}\simeq V_\lambda$. \item[\rm{(iii)}] For fixed $\lambda$ and $a\geq \abs{\lambda}$, the $S_a$-representati

*Extractive statement; proof gist not independently reconstructed.*

Proof opening: Lemma~\ref{lemma:branching}(ii) states that $(V_\mu)_{S_k}$ is the sum of $V_\nu$ over partitions $\nu\vdash a$ obtained from $\mu$ by removing boxes from $k$ different columns. Since $\mu$ has only $\mu_1$ columns, this is impossible when $k>\mu_1$. This demonstrates (i), from which (iv) follows immediately. When $k=\mu_1$, this can be done only by removing one box from each of the $\mu_1$ columns of $\mu$; by definition, to say that $\mu=\lambda[n]$ means that the resulting partition is $\lambda$, demonstrating (ii). To prove (iii), let $r=n-a$ and $c=a-\abs{\lambda}$; our assumption is that $r\geq \lambda_1$. By Lemma~\ref{lemma:branching}(i), the right side of \eqref{eq:coinvariants} con


Candidate references: `def:Vlambdan` (internal_reference); `lemma:branching` (candidate_proof_dependency); `eq:coinvariants` (candidate_proof_dependency).

### Proposition 3.2.8 — `pr:narrow`

Source lines 992–995.

Let $V$ be an FI-module over a field of characteristic~0. For all $n\geq 0$, every irreducible constituent $V(\lambda)_n$ of the $S_n$-representation $V_n$ satisfies $\lambda_1\leq \stabD(V)$.

*Extractive statement; proof gist not independently reconstructed.*

Proof opening: Let $s\coloneq \stabD(V)$. Consider the filtration $F^iV\coloneq \spn(V_{\leq i})$, which satisfies $V=\bigcup F^i V$. We will prove the inequality $\lambda_1\leq s$ for the irreducible constituents of each quotient $F^mV/F^{m-1}V$, and this implies the same inequality for $V$. We first prove that for all $m\geq 0$, every irreducible constituent $V_\mu$ of $H_0(V)_m$ satisfies $\mu_1\leq s$. Fix $m\geq 0$, and let $\overline{V}\coloneq V/F^{m-1}V$. We have $\overline{V}_m=V_m/\spn(V_{&lt;m})_m=H_0(V)_m$ as $S_m$-representations, and of course $\overline{V}_n=0$ for $n&lt;m$. Being a quotient of $V$, we have $\surjD(\overline{V})\leq s$ by Lemma~\ref{lemma:injsurjdescend}, so the map $T\colon \Phi_


Candidate references: `lemma:injsurjdescend` (candidate_proof_dependency); `lem:coinvariants` (candidate_proof_dependency); `pr:Mmuweight` (candidate_proof_dependency); `lemma:Mspan` (candidate_proof_dependency).

### Remark 3.3.1 — `eq:consistentseqcondition`

Source lines 1047–1055.

Not every consistent sequence can arise from an FI-module. For any $\sigma\in S_{n+k}$ with $\sigma|_\n=\id$, we have the identity $f_{n,n+k}=\sigma \circ f_{n,n+k}$ in $\Hom_{\FI}(\n,\mathbf{n+k})$. Therefore on any consistent sequence arising from an FI-module, \text{all }\sigma\in S_{n+k}\text{ with }\sigma|_\n=\id\text{ must act trivially on }\im(V_n\to\cdots\to V_{n+k})\subset V_{n+k}. For example, the consistent sequence of regular representations $k\to k \to k[S_2] \to k[S_3]\to \cdots$ induced by the standard inclusions $S_m\into S_{m+1

*Extractive statement; proof gist not independently reconstructed.*

### Definition 3.3.2 — `definition:repstab`

Source lines 1057–1073.

{\bf Representation stability \cite{CF}}

Let $\{V_n,\phi_n\}$ be a consistent sequence of $S_n$--representations over a field of characteristic~0. The sequence $\{V_n,\phi_n\}$ is uniformly representation stable with stable range $n\geq N$ if each of the following conditions holds. [{\bf I.}] \item Injectivity: The map $\phi_n\colon V_n\to V_{n+1}$ is injective for all $n\geq N$. \item Surjectivity: The span of the $S_{n+1}$--orbit of $\phi_n(V_n)$ equals all of $V_{n+1}$ for all $n\geq N$. \item Multiplicities: Decompose $V_n$ into irreducible representations as \[V_n=\bigoplus_\lamb

*Extractive statement; proof gist not independently reconstructed.*


Candidate references: `external:CF` (external_reference).

### Proposition 3.3.3 — `pr:stabdegstabrange`

Source lines 1079–1082.

{\bf Stability degree and\ representation stability}

Let $V$ be an FI-module over a field of characteristic~0. The sequence of $S_n$-representations $\{V_n,\phi_n\}$ is uniformly representation stable with stable range ${n\geq \wt(V)+\stabD(V)}$.

*Extractive statement; proof gist not independently reconstructed.*

Proof opening: Set $s\coloneq \stabD(V)$ and $d\coloneq \wt(V)$. Let $K_n$ be the kernel of the $S_n$-equivariant map $\phi_n\colon V_n\to V_{n+1}$, and let $C_{n+1}$ be the cokernel of the $S_{n+1}$-equivariant map $\phi'_n\coloneq \Ind_{S_n}^{S_{n+1}}\phi_n \colon \Ind_{S_n}^{S_{n+1}} V_n\to V_{n+1}$ induced by $\phi_n$. To prove Conditions I and II of Definition~\ref{definition:repstab}, we must show that $K_n = 0$ and $C_{n+1}=0$ for all $n \geq s+d$. Since $K_n$ is a subrepresentation of $V_n$ and $C_{n+1}$ is a quotient of $V_{n+1}$, Lemma~\ref{lem:coinvariants}(iv) implies for any $n\geq d$ that $K_n=0\iff (K_n)_{S_{n-d}}=0$ and $C_{n+1}=0\iff (C_{n+1})_{S_{n+1-d}}=0$. Since taking $S_a$-coinvariant


Candidate references: `definition:repstab` (candidate_proof_dependency); `lem:coinvariants` (candidate_proof_dependency); `def:stabilitydegree` (candidate_proof_dependency); `pr:narrow` (candidate_proof_dependency); `eq:repstabcoinvariants` (candidate_proof_dependency).

### Theorem 3.3.4 — `thm:persinomial`

Source lines 1129–1134.

{\bf Polynomiality of characters}

Let $V$ be a finitely generated FI-module over a field of characteristic~0. There is a unique polynomial $P_V \in \Q[X_1, X_2, \ldots]$ with $\deg P_V\leq \wt(V)$ such that for all $n\geq \stabD(V)+\wt(V)$ and all $\sigma\in S_n$, \[\chi_{V_n}(\sigma)=P_V(\sigma).\]

*Extractive statement; proof gist not independently reconstructed.*

Proof opening: Classically, the interest in character polynomials is driven by the following fact: for each partition $\lambda$, there exists a polynomial $P_\lambda\in\Q[X_1, X_2, \ldots]$ of degree $\abs{\lambda}$ such that $\chi_{V(\lambda)_n}(\sigma) = P_\lambda(\sigma)$ for all $n \geq \abs{\lambda} + \lambda_1$ and all $\sigma\in S_n$ \cite[Example I.7.14]{Mac}. This seems to have been known, at least implicitly, as far back as Murnaghan; Macdonald traces it back to work of Frobenius in 1904. For a more recent reference, see Garsia--Goupil~\cite{GG}. Proposition~\ref{pr:stabdegstabrange} implies that for every finitely generated FI-module $V$, there exist constants $c_\lambda$ so that $ V_n \simeq \b

- `pr:stabdegstabrange` (proof_dependency; source_audit_annotation): Proposition~\ref{pr:stabdegstabrange} implies that for every finitely generated FI-module $V$, there exist constants $c_\lambda$ so that $ V_n \simeq \bigoplus_\lambda c_\lambda V(\lambda)_n$ for all $n \geq  \stabD(V)+\wt(V)$. If $\lambda$ is a partition with $c_\lambda \neq 0$, the definition of weight implies that $\abs{\lambda} \leq  \wt(V)$ and Proposition~\ref{pr:narrow} states that that $\lambda_1 \leq \stabD(V)$. The polynomial $P_V\coloneq \sum_\lambda c_\lambda P_\lambda$ thus satisfies the desired claim.

Candidate references: `pr:stabdegstabrange` (candidate_proof_dependency); `pr:narrow` (candidate_proof_dependency); `external:Mac` (external_reference); `external:GG` (external_reference).

### Proposition 3.4.1 — `pr:Vlambda`

Source lines 1155–1163.

{\bf The FI-module $V(\lambda)$}

Let $k$ be a field of characteristic~0. For any partition $\lambda$, there is an FI-module $V(\lambda)$ satisfying V(\lambda)_\n\simeq V_{\lambda[n]}&\text{if }n\geq \abs{\lambda}+\lambda_1\\0&\text{otherwise} that is finitely generated in degree $\abs{\lambda}+\lambda_1$. It has $\stabD(V(\lambda))=\lambda_1$ and $\wt(V(\lambda))=\abs{\lambda}$.

*Extractive statement; proof gist not independently reconstructed.*

Proof opening: We will define $V(\lambda)$ as a sub-FI-module of $M(\lambda)$. Let $d\coloneq \abs{\lambda}+\lambda_1$. Since every finite set is isomorphic to $\n$ for some $n\geq 0$, it suffices to define the subspace $V(\lambda)_\n\subset M(\lambda)_\n$ for $n\geq d$. Lemma~\ref{lemma:branching}(i) states that $V_\mu$ occurs in $M(\lambda)_\n$ if $\mu\vdash n$ is obtained from $\lambda$ by adding one box to $n-\abs{\lambda}$ different columns. When $n-\abs{\lambda}\geq \lambda_1$, adding boxes to the first $n-\abs{\lambda}$ columns yields the partition $\mu=\lambda[n]$. (For future reference, we note that $\lambda[n]$ has $\abs{\lambda}$ boxes below the first row, while all other partitions $\mu\neq \la


Candidate references: `lemma:branching` (candidate_proof_dependency); `eq:Vlambdan` (candidate_proof_dependency); `lemma:injsurjdescend` (candidate_proof_dependency); `pr:stabdegMlambda` (candidate_proof_dependency); `lem:coinvariants` (candidate_proof_dependency); `external:FH` (external_reference); `external:Ch` (external_reference).

### Theorem 3.4.2 — `thm:Murnaghan`

Source lines 1188–1195.

{\bf Murnaghan's Theorem}

For each pair of partitions $\lambda$ and $\mu$ there exists a finite set $S$ of partitions $\nu$ and a set of nonnegative integers $g_{\lambda,\mu}^\nu$ such that for all sufficiently large $n$: V(\lambda)_n \tensor V(\mu)_n = \bigoplus_{\nu \in S} g_{\lambda,\mu}^\nu V(\nu)_n.

*Extractive statement; proof gist not independently reconstructed.*

### Proposition 3.4.3 — `pr:schurfunctor`

Source lines 1212–1215.

{\bf Finite generation of Schur functors}

Let $V$ be an FI-module over a field of characteristic 0. Assume that $V$ is finitely generated (resp.\ generated in degree $\leq m$, resp.\ of weight $\leq m$). Then for any partition $\lambda$, the FI-module $\Schur_\lambda(V)$ is finitely generated (resp.\ generated in degree $\leq m\cdot \abs{\lambda}$, resp.\ of weight $\leq m\cdot \abs{\lambda}$).

*Extractive statement; proof gist not independently reconstructed.*

Proof opening: Let $d=\abs{\lambda}$, and let $c_\lambda\in k[S_d]$ be the Young symmetrizer associated to the partition $\lambda$. The permutation group $S_d$ acts on the FI-module $V^{\otimes d}$ by permuting the factors, so $c_\lambda$ defines an idempotent endomorphism of $V^{\otimes d}$ whose image is isomorphic to $\Schur_\lambda(V)$. Propositions~\ref{pr:tensor} and \ref{pr:tensorwt} imply that the FI-module $V^{\otimes d}$ has the desired properties, so the same is true of its quotient $\Schur_\lambda(V)$.


Candidate references: `pr:tensor` (candidate_proof_dependency); `pr:tensorwt` (candidate_proof_dependency).

### Corollary 3.4.4 — `corollary:line:1223:116053`

Source lines 1223–1225.

{\bf Stability of Schur functors}

The sequence of $S_n$--representations $\{\Schur_\lambda(\Q^n)\}$ is monotone in the sense of \cite{Ch} and is uniformly representation stable.

*Extractive statement; proof gist not independently reconstructed.*


Candidate references: `external:Ch` (external_reference).

### Proposition 3.4.5 — `prop:line:1229:116516`

Source lines 1229–1234.

{\bf Schur functors of irreducibles}

Let $\lambda, \mu$ be partitions. There exists a finite set $S$ of partitions $\nu$ and a set of nonnegative integers $\beta_{\lambda,\mu}^\nu$ such that for all sufficiently large $n$: \beq \Schur_\lambda(V(\mu)_n) = \bigoplus_{\nu \in S} \beta_{\lambda,\mu}^\nu V(\nu)_n \eeq

*Extractive statement; proof gist not independently reconstructed.*

### Definition 4.1.1 — `definition:line:1249:117459`

Source lines 1249–1254.

{\bf \boldmath$\FIsharp$-modules}

Let $\FIsharp$ be the category whose objects are finite sets, with morphisms being partially-defined injections $S\supset A\xrightarrow{\phi}B\subset T$. That is, $\Hom_{\FIsharp}(S,T)$ consists of triples $(A,B,\phi)$ with $A\subseteq S$, $B\subseteq T$ and $\phi\colon A \ra B$ an isomorphism, with the composition of \[S\supset \phi^{-1}(B)\xrightarrow{\phi}B\subset T\quad\text{and}\quad T\supset C\xrightarrow{\psi}\psi(C)\subset U\quad\text{being}\quad S\supset \phi^{-1}(B\cap C)\xrightarrow{\psi\circ \phi}\psi(B\cap C)\subset U. \] An $\FIsh

*Extractive statement; proof gist not independently reconstructed.*

### Example 4.1.2 — `ex:standardFIsharp`

Source lines 1258–1261.

The most basic example is the $\FIsharp$-module $M(1)$ taking a finite set $S$ to the free $k$-module $M(1)_S=\langle e_s|s\in S\rangle$ with basis $S$. A morphism $f\colon S\supset A\xrightarrow{\phi}B\subset T$ induces the map $f_*\colon M(1)_S\to M(1)_T$ defined by $f_*(e_s)=e_{\phi(s)}$ if $s\in A$, and $f_*(e_s)=0$ otherwise.

*Extractive statement; proof gist not independently reconstructed.*

### Remark 4.1.3 — `remark:FIsharpremarks`

Source lines 1263–1270.

The category $\FI$ embeds in $\FIsharp$ by taking only those morphisms of the form $S=S\xrightarrow{\approx} B\subset T$, so every $\FIsharp$-module can be considered as an FI-module. The category $\coFI$ also embeds in $\FIsharp$ by taking only those morphisms of the form $S\supset A\xrightarrow{\approx} T=T$, so every $\FIsharp$-module can be considered as a co-FI-module as well. The relations in $\FIsharp$ impose conditions on how these FI-module and co-FI-module structures interact. In particular, the endomorphisms $\End_{\FIsharp}(\n)$ for

*Extractive statement; proof gist not independently reconstructed.*


Candidate references: `ex:standardFIsharp` (internal_reference); `external:So` (external_reference).

### Example 4.1.4 — `ex:MWFIsharp`

Source lines 1277–1281.

{\bf \boldmath$M(W)$ is an \boldmath$\FIsharp$-module}

Given an FB-module $W$, we defined the FI-module $M(W)$ in Definition~\ref{def:MW}. It turns out this extends to a natural $\FIsharp$-module structure on $M(W)$, as follows. To a finite set $S$, the $\FIsharp$-module $M(W)$ assigns \[M(W)_S\coloneq \bigoplus_{T\subseteq S}W_T.\] To a morphism $f\colon S\supset A\xrightarrow{\phi}B\subset S'$, it assigns the map $f_*\colon M(W)_S\to M(W)_{S'}$ which on the factor $W_T$ is $(\phi|_T)_*\colon W_T\to W_{\phi(T)}$ if $T\subseteq A$, and is $0$ if $T\not\subset A$.

*Extractive statement; proof gist not independently reconstructed.*


Candidate references: `def:MW` (internal_reference).

### Theorem 4.1.5 — `th:charFIsharp`

Source lines 1288–1294.

{\bf Classification of \boldmath$\FIsharp$-modules}

The category of $\FIsharp$-modules is equivalent to the category of FB-modules, via the equivalence of categories \[M(-)\colon \FBMod\leftrightarrows \FIsharpMod\cocolon H_0(-).\] Thus every $\FIsharp$-module $V$ is of the form $V=\bigoplus_{i=0}^\infty M(W_i)$, where $W_i$ is the $S_i$-representation $H_0(V)_i$.

*Extractive statement; proof gist not independently reconstructed.*

Proof opening: We have already seen the canonical isomorphism $H_0(M(W))\simeq W$ for any FB-module $W$ in Remark~\ref{remark:H0}. It remains to prove that for any $\FIsharp$-module, we have a canonical isomorphism $V\simeq M(H_0(V))$. Fix $n\geq 0$. We will analyze the category of $\FIsharp$-modules $V$ satisfying the condition V_S=0\quad\text{ for all }S\text{ with }\abs{S}&lt;n. The assumption \eqref{eq:FIsharpcond} implies that the quotient \eqref{eq:defH0} defining $H_0(V)_n$ is vacuous, so $H_0(V)_n\simeq V_n$ as $S_n$-representations. Our goal will be to prove that we have a natural isomorphism V\simeq M(V_n)\oplus V'\simeq M(H_0(V)_n)\oplus V', where $V'$ is an $\FIsharp$-module for which $V_S=0$ when


Candidate references: `remark:H0` (candidate_proof_dependency); `eq:FIsharpcond` (candidate_proof_dependency); `eq:defH0` (candidate_proof_dependency); `eq:FIsharpE` (candidate_proof_dependency); `eq:FIsharpgoal` (candidate_proof_dependency); `ex:MWFIsharp` (candidate_proof_dependency).

### Remark 4.1.6 — `remark:line:1296:122151`

Source lines 1296–1300.

Pirashvili established in \cite[Theorem~3.1]{Pi} an equivalence between the category of $\Gamma$-modules and FS-modules; here $\Gamma$ is the category of finite based sets and all based maps, and FS is the category of finite sets and surjections. The classification of $\FIsharp$-modules in Theorem~\ref{th:charFIsharp} is closely related to Pirashvili's result, although technically neither follows from the other. Since FB is a full subcategory of FS, an FS-module can be thought of as an FB-module with additional morphisms. Moreover, the full sub

*Extractive statement; proof gist not independently reconstructed.*


Candidate references: `th:charFIsharp` (internal_reference); `external:Pi` (external_reference).

### Theorem 4.1.7 — `thm:polyFIsharp`

Source lines 1364–1382.

Let $V$ be an $\FIsharp$-module over $k$. The following conditions are equivalent: \item[(i)] $V$ is finitely generated as an $\FIsharp$-module. \item[(ii)] $V$ is finitely generated as an FI-module. \item[(iii)] $\bigoplus_{n\geq 0} H_0(V)_n$ is finitely generated as a $k$-module. \item[(iv)] There exists $d\geq 0$ such that $V_n$ is generated as a $k$-module by $O(n^d)$ elements. When $k$ is a field, these are also equivalent to the following conditions: \item[(v)] There exists $d\geq 0$ such that $\dim_k V_n=O(n^d)$. \item[(vi)] There exists

*Extractive statement; proof gist not independently reconstructed.*

Proof opening: It is immediate that $(ii) \implies (i)$, and we have already seen in Remark~\ref{remark:H0} that $(ii)\iff (iii)$. Consider the FB-module $W=H_0(V)$, so that $V\simeq M(W)$. Theorem~\ref{th:charFIsharp} states that $(i)$ holds if and only if $H_0(V)$ is finitely generated as an FB-module. Since $k[S_n]$ is finite over $k$, this implies that $\bigoplus_{n\geq 0}H_0(V)_n$ is finitely generated as a $k$-module, so $(i)\implies (iii)$. Let $c_i$ be the smallest number of generators for the $k$-module $H_0(V)_i=W_i$. Condition $(iii)$ states that there exists $d$ such that $c_i<\infty$ for all $i\geq d$ and $c_i=0$ for all $i>d$. From \eqref{eq:MWdecomp} we have $M(W)_n\simeq \bigoplus_{T\subset

- `th:charFIsharp` (proof_dependency; source_audit_annotation): \begin{proof} It is immediate that $(ii) \implies (i)$, and we have already seen in Remark~\ref{remark:H0} that $(ii)\iff (iii)$. Consider the FB-module $W=H_0(V)$, so that $V\simeq M(W)$.  Theorem~\ref{th:charFIsharp} states that $(i)$ holds if and only if $H_0(V)$ is finitely generated as an FB-module. Since $k[S_n]$ is finite over $k$, this implies that $\bigoplus_{n\geq 0}H_0(V)_n$ is finitely generated as a $k$-module, so $(i)\implies (iii)$. Let $c_i$ be the smallest number of generators for the $k$-module $H_0(V)_i=W_i$.  Condition $(iii)$ states that there exists $d$ such that $c_i<\infty$ for all $i\geq d$ and $c_i=0$ for all $i>d$. From \eqref{eq:MWdecomp} we have $M(W)_n\simeq \bigoplus_{T\subset \n} W_T$, so as $k$-modules we have $V_n=M(W)_n\simeq \bigoplus_{i\geq 0}W_i^{\oplus \binom{n}{i}}$. This is generated by $\sum_{i\geq 0} c_i\cdot \binom{n}{i}$ elements, so $(iii)\implies (iv)$. Conversely, assume that $V_n$ is generated by $O(n^d)$ elements, so $V_n$ admits a surjection from $k^{c n^d}$ for some constant $c$.  Suppose furthermore that $W_i$ is nonzero for some $i > 0$, and let $\mathfrak{m}$ be a maximal ideal of $k$ with quotient field $\F\coloneq k/\mathfrak{m}$ such that $W_i \otimes_k \F\neq 0$.  The hypothesized surjection shows that the  $\F$-dimension of $k^{c n^d} \otimes_k \F\simeq \F^{cn^d}$ is at least that of $W_i^{\oplus \binom{n}{i}} \otimes_k \F\simeq (W_i\otimes_k \F)^{\oplus\binom{n}{i}}$ for all $n$, which  implies $i \leq d$.  Therefore $W_i$ is zero for all $i > d$. Since $V_i$ surjects to $W_i$, the $k$-module $W_i$ must be finitely generated for $i \leq d$. This shows that $(iv)\implies (iii)$.

Candidate references: `remark:H0` (candidate_proof_dependency); `th:charFIsharp` (candidate_proof_dependency); `eq:MWdecomp` (candidate_proof_dependency).

### Corollary 4.1.8 — `co:fisharprange`

Source lines 1401–1404.

Let $V$ be an $\FIsharp$-module over a field of characteristic~0. The sequence of $S_n$-representations $\{V_n,\phi_n\}$ is uniformly representation stable with stable range $n\geq 2\cdot\wt(V)$.

*Extractive statement; proof gist not independently reconstructed.*

Proof opening: We first show that $\stabD(V)\leq \wt(V)$. By Theorem~\ref{th:charFIsharp}, we have an isomorphism $V\simeq \bigoplus_\lambda M(\lambda)^{\oplus c_\lambda}$. Applying Propositions~\ref{pr:stabdegMlambda} and \ref{pr:Mmuweight}, each factor $M(\lambda)$ occurring in this sum satisfies $\stabD(M(\lambda))=\lambda_1\leq \abs{\lambda}=\wt(M(\lambda))$, so the same inequality holds for $V$. The corollary now follows from Proposition~\ref{pr:stabdegstabrange}.


Candidate references: `th:charFIsharp` (candidate_proof_dependency); `pr:stabdegMlambda` (candidate_proof_dependency); `pr:Mmuweight` (candidate_proof_dependency); `pr:stabdegstabrange` (candidate_proof_dependency).

### Remark 4.1.A — `rem:finiteprojresolution`

Source lines 1411–1440.

{\bf Finite projective resolutions}

Over a field of characteristic~0, Remark~\ref{rem:projective} implies that every projective FI-module extends to an $\FIsharp$-module. So Theorem~\ref{thm:polyFIsharp} implies that the characters of any finitely generated projective FI-module are polynomial for all $n\geq 0$. Since characters are additive in exact sequences, the same is true for any finitely generated FI-module admitting a finite projective resolution: the character $\chi_{V_n}$ is given by the character polynomial $P_V$ for all $n\geq 0$. So if this is not the case, then $V$ d

*Extractive statement; proof gist not independently reconstructed.*


Candidate references: `rem:projective` (internal_reference); `thm:polyFIsharp` (internal_reference); `rem:nonsemisimple` (internal_reference).

### Problem 4.1.B — `alphprob:line:1476:145118`

Source lines 1476–1478.

{\bf $\FIsharp$-module tensor coefficients}

Give a geometric interpretation of the structural coefficients $d_{\lambda,\mu}^\nu$ in \eqref{eq:Kronecker}. Give a method for determining which of these coefficients are nonzero.

*Extractive statement; proof gist not independently reconstructed.*


Candidate references: `eq:Kronecker` (internal_reference).

### Definition 4.2.1 — `definition:line:1490:146415`

Source lines 1490–1492.

{\bf Graded FI-modules of finite type}

Let $V$ be a graded FI-module. We say that $V$ is of finite type if each FI-module $V^i$ is a finitely generated FI-module. If $k$ is a field of characteristic~0, we say that $V$ has slope $\leq m$ if $\wt(V^i)\leq m\cdot i$ for all $i\in \N$. (For instance, this occurs when $V^i$ is generated in degree $\leq m\cdot i$, via Proposition~\ref{pr:fgweight}.)

*Extractive statement; proof gist not independently reconstructed.*


Candidate references: `pr:fgweight` (internal_reference).

### Proposition 4.2.2 — `pr:gradedtensor`

Source lines 1500–1503.

Let $U$ and $W$ be graded FI-modules. If $U$ and $W$ are of finite type, the tensor product $U\tensor W$ is of finite type. If $U$ and $W$ have slope $\leq m$, then $U\tensor W$ has slope $\leq m$.

*Extractive statement; proof gist not independently reconstructed.*

Proof opening: By Proposition~\ref{pr:tensor}, $U^i\otimes W^j$ is a finitely generated FI-module for each fixed $i$ and $j$. For fixed $k$ the graded piece $(U\otimes W)^k$ is the direct sum of $k+1$ such summands, and thus is finitely generated. The second claim follows from the inequality $\wt(U^i\otimes W^j)\leq \wt(U^i)+\wt(W^j)\leq mi+mj=mk$ from Proposition~\ref{pr:tensorwt}.


Candidate references: `pr:tensor` (candidate_proof_dependency); `pr:tensorwt` (candidate_proof_dependency).

### Theorem 4.2.3 — `thm:fisubalgebra`

Source lines 1518–1521.

{\bf Algebras generated by finite type FI-modules}

Let $A$ be a graded FI-algebra generated by a graded sub-FI-module $V\subset A$ with $V^0=0$. If $V$ is of finite type then $A$ is of finite type. If $V$ has slope $\leq m$ then $A$ has slope $\leq m$.

*Extractive statement; proof gist not independently reconstructed.*

Proof opening: We first consider $k\{V\}$, which decomposes as a graded FI-module as \[k\{V\} \simeq \bigoplus_{j=0}^\infty (V^{\otimes j})^{\oplus P_j}\] where $P_j$ is the set of parenthesizations of a $j$-fold product (so that $\abs{P_j}$ is the $j$-th Catalan number). If $V$ is of finite type (resp.\ has slope $\leq m$), the same is true of each summand $V^{\otimes j}$ by Proposition~\ref{pr:gradedtensor}. Moreover, since $V^i\neq 0$ only for $i\geq 1$, we have $(V^{\otimes j})^i\neq 0$ only for $i\geq j$. For fixed $i\in \N$ we thus have a finite sum $k\{V\}^i=\bigoplus_{j=0}^i\big((V^{\otimes j})^i\big)^{\oplus P_j}$. We conclude that the FI-module $k\{V \}^i$ is finitely generated (resp.\ of weight


Candidate references: `pr:gradedtensor` (candidate_proof_dependency).

### Proposition 4.2.4 — `pr:cofisubalgebra`

Source lines 1546–1548.

Let $A$ be a graded co-FI-algebra over a field of characteristic~0, generated by a graded co-FI-submodule $W\subset A$ with $W^0=0$. If $W$ is of finite type, $A$ is of finite type. If $V$ has slope $\leq m$, then $A$ has slope $\leq m$.

*Extractive statement; proof gist not independently reconstructed.*

### Corollary 4.2.A — `cor:subalgebraslope`

Source lines 1550–1553.

If a graded FI-algebra (resp.\ co-FI-algebra) $A$ is generated by a sub-FI-module (resp.\ co-FI-module) $V$ concentrated in grading $1$ (i.e.\ $V^i=0$ for $i\neq 1$), then $A$ has slope $\leq \wt(V)$.

*Extractive statement; proof gist not independently reconstructed.*

### Definition 4.2.5 — `def:tensorbullet`

Source lines 1559–1578.

Given an $\FIsharp$-module $V$ equipped with a splitting $M(0)\into V\onto M(0)$, define the $\FIsharp$-module $V^{\tensor\bullet}$ by $(V^{\tensor\bullet})_S\coloneq (V_S)^{\otimes S}$. Given $f\colon S\supset A\xrightarrow{\phi}B\subset T$, let $f_*\colon (V^{\tensor\bullet})_S\to (V^{\tensor\bullet})_T$ be the composition \bigotimes_{s\in S}V_S\onto \bigotimes_{s\in A}V_S\xrightarrow{(f_*)^{\otimes \phi}}\bigotimes_{t\in B}V_T\into \bigotimes_{t\in T}V_T Here the first map applies the surjection $V_S\onto M(0)_S\simeq k$ to each factor with

*Extractive statement; proof gist not independently reconstructed.*


Candidate references: `eq:tensorbullet` (internal_reference).

### Remark 4.2.6 — `remark:line:1580:156140`

Source lines 1580–1582.

If $U$ is a graded $k$-module with splitting $k\into U^0\onto k$, we can consider $U$ as a ``constant'' graded $\FIsharp$-module, sending every finite set $S$ to $U$ and every morphism to the identity. Definition~\ref{def:tensorbullet} thus defines a graded $\FIsharp$-module $U^{\otimes \bullet}$, which satisfies $(U^{\tensor\bullet})_n\simeq U^{\otimes n}$.

*Extractive statement; proof gist not independently reconstructed.*


Candidate references: `def:tensorbullet` (internal_reference).

### Proposition 4.2.7 — `pr:Vtensorbullet`

Source lines 1584–1587.

Let $V$ be a graded FI-module endowed with an isomorphism $V^0\simeq M(0)$. If $V$ is of finite type then the FI-module $V^{\tensor\bullet}$ is of finite type.

*Extractive statement; proof gist not independently reconstructed.*

Proof opening: Consider a graded FI-module $M$ that is free in each grading, meaning that there exists an index set $L$, and for each $\ell\in L$ numbers $m_\ell\in \N$ and $i_\ell\in \N$, so that $M^i\simeq \bigoplus_{i_\ell=i}M(m_\ell)$. Every graded FI-module $V$ admits a surjection $M\onto V$ from such a graded FI-module $M$. If $V$ is of finite type, then we can take $M$ to be of finite type, meaning that $\{\ell \in L\,|\, i_\ell=j\}$ is finite for each $j$. If $V^0\simeq M(0)$, we can assume that there is a unique $\ell_0\in L$ with $i_{\ell_0}=0$, and it satisfies $m_{\ell_0}=0$. Since a surjection $M\onto V$ induces a surjection $M^{\tensor\bullet}\onto V^{\tensor \bullet}$, it suffices to prove t


Candidate references: `eq:Mtensorbulletbasis` (candidate_proof_dependency).

### Definition 5.1.1 — `definition:line:1657:164202`

Source lines 1657–1659.

{\bf Symmetrically presented FI-algebra}

Given a collection $\PP=\{P_j(x_1,\ldots,x_{m_j})\}$ of homogeneous polynomials, the symmetrically presented FI-algebra $R(\PP)$ is the quotient $k[M(1)]/((\PP))$.

*Extractive statement; proof gist not independently reconstructed.*

### Example 5.1.2 — `xample:line:1663:164760`

Source lines 1663–1670.

{\bf Nilpotent rings}

For $d\geq 2$, let $\PP_d=\{x_1^d\}$. The symmetrically presented FI-algebra $R(\PP_d)$ over $\Q$ corresponds to the sequence of nilpotent rings \[ R(\PP_d)_n\simeq \Q[x_1,\ldots ,x_n]/(x_1^d,\ldots ,x_n^d) \] on which $S_n$ acts by permutations. Since $R(\PP_d)$ is of finite type, for each $i\geq 0$ the sequence of $S_n$-representations $R(\PP_d)_n^i$ is uniformly representation stable. When $d=2$, this fact was first proved by Ashraf--Azam--Berceneau in \cite[Corollary 5.2]{AAB}.

*Extractive statement; proof gist not independently reconstructed.*


Candidate references: `external:AAB` (external_reference).

### Example 5.1.A — `example:arnoldalgebra`

Source lines 1674–1713.

{\bf The Arnol'd algebra}

Braid groups were first studied by Artin and Hurwitz over a century ago. In this section we focus on the pure braid group $P_n$, which is the fundamental group of the configuration space of $n$ distinct complex numbers $(z_1,\ldots,z_n)$ with $z_i\in \C$ and $z_i\neq z_j$. This space is aspherical, so its cohomology coincides with $H^*(P_n;\Q)$. These cohomology algebras $H^*(P_n;\Q)$ fit together into a graded $\FIsharp$-algebra $H^*(P_\bullet;\Q)$; this follows from a general result that we will prove below in Proposition~\ref{pr:CMhTop}. In

*Extractive statement; proof gist not independently reconstructed.*


Candidate references: `pr:CMhTop` (internal_reference); `thm:fisubalgebra` (internal_reference); `cor:subalgebraslope` (internal_reference); `co:fisharprange` (internal_reference); `eq:arnoldideal` (internal_reference); `eq:dlmn` (internal_reference); `external:Ar` (external_reference); `external:CF` (external_reference).

### Definition 5.1.3 — `def:generalcoinvariantalgebra`

Source lines 1727–1730.

Given a graded co-FI-algebra $A$, the coinvariant algebra $\coinv(A)$ is the graded co-FI-algebra $\coinv(A)\coloneq A/((A^{\inv}_{>0}))$, which assigns to $T$ the algebra $\coinv(A)_T=A_T/((A_T)^{S_T}_{>0})$.

*Extractive statement; proof gist not independently reconstructed.*

### Proposition 5.1.4 — `pr:coinvariantfinite`

Source lines 1731–1734.

Let $A$ be a graded co-FI-algebra over a field of characteristic 0. If $A$ is of finite type, $\coinv(A)$ is of finite type. If $A$ has slope $\leq m$, then $\coinv(A)$ has slope $\leq m$.

*Extractive statement; proof gist not independently reconstructed.*

### Example 5.1.5 — `xample:line:1736:172087`

Source lines 1736–1738.

Consider the graded co-FI-algebra $R\coloneq \coinv(\Q[M(1)])$, which is of finite type with slope $\leq 1$ by Proposition~\ref{pr:coinvariantfinite}. As $S_n$-representations, $R_n\simeq \Q[x_1,\ldots,x_n]/(\Q[x_1,\ldots,x_n])^{S_n}_{>0}\simeq R(n)$. In other words, we have combined the classical coinvariant algebras $R(n)$ into a single co-FI-algebra $R$. Theorem~\ref{theorem:intro:FIequiv} implies that for each $i\geq 0$, the graded pieces $R^i_n\simeq R(n)^i$ of the coinvariant algebra form a uniformly representation stable sequence of $S_n

*Extractive statement; proof gist not independently reconstructed.*


Candidate references: `pr:coinvariantfinite` (internal_reference); `theorem:intro:FIequiv` (internal_reference); `external:CF` (external_reference).

### Proposition 5.2.1 — `proposition:line:1760:176167`

Source lines 1760–1763.

Assume that $k$ is a field of characteristic~0. There is a graded co-FI-algebra $BS$ of finite type satisfying $BS_n\simeq G(R_n/k)$ as graded $S_n$-representations, where the $S_n$-action on $G(R_n/k)$ is via the diagonal subgroup $S_n \subset S_n\times S_{n+1}$. In particular, the $S_n$-representations $G(R_n/k)^i$ are uniformly representation stable.

*Extractive statement; proof gist not independently reconstructed.*

Proof opening: The Galois closure $G(R_n/k)$ is defined as the quotient of $R_n^{\otimes n+1}$ by certain relations, and Bhargava--Satriano show \cite[\S11.2]{BS} in this case that the ideal $J_n$ of relations in $R_n^{\otimes n+1}$ is generated by the elements: \[\gamma_n(x_i)\coloneq x_i\otimes 1\otimes\cdots \otimes 1\ +\ 1\otimes x_i\otimes\cdots \otimes 1\ \ +\ \cdots\ +\ \ 1\otimes\cdots \otimes 1\otimes x_i\] Let $R$ be the graded co-FI-algebra $R$ with $R^0\simeq M(0)$, $R^1\simeq M(1)^*$, and $R^i=0$ for $i>1$; the multiplication is uniquely determined, since $R^1\otimes R^1\to R^2=0$. To the set $\n$ this assigns the ring $R_n=k[x_1,\ldots,x_n]/(x_ix_j)$ defined above. Recall the graded co-FI-alg


Candidate references: `def:tensorbullet` (candidate_proof_dependency); `pr:Vtensorbullet` (candidate_proof_dependency); `pr:gradedtensor` (candidate_proof_dependency); `th:FIequiv` (candidate_proof_dependency); `external:BS` (external_reference).

### Question 5.3.1 — `question:rankvar`

Source lines 1804–1807.

What is the dimension of the space of degree $d\geq 1$ polynomials on $X_{P,r}(n)$; that is, what is the dimension of the degree $d\geq 1$ component of $\O(X_{P,r}(n))\simeq k[\{x_{ij}\}]/J_n$?

*Extractive statement; proof gist not independently reconstructed.*

### Theorem 5.3.2 — `theorem:sections:polygrowth`

Source lines 1811–1818.

{\bf Polynomiality for rank varieties}

Fix $P$, $r$, and $d$. Then the dimension of the space $\O(X_{P,r}(n))^d$ of degree-$d$ polynomials on the rank variety $X_{P,r}(n)$ is polynomial in $n$ of degree at most $2d$ for all $n\geq 0$. If $\car k=0$ then the same is true of the characters $\chi_{\O(X_{P,r}(n))^d}$.

*Extractive statement; proof gist not independently reconstructed.*

Proof opening: The matrix varieties $\Mat_{n\times n}$ fit together into an $\FIsharp$-scheme $\Mat$; the affine scheme $\Mat_{S\times S}$ has coordinate ring $\O(\Mat_{S\times S})=k[x_{s,s'}\,|\,s,s'\in S]$. Given $f\colon S\supset A\xrightarrow{\phi}\phi(A)\subset T$, the map $f_*\colon \Mat_{S\times S}\to \Mat_{T\times T}$ is given in coordinates by $x_{\phi(s),\phi(s')}\mapsto x_{s,s'}$ and $x_{t,t'}\mapsto 0$ if $t\not\in \phi(A)$ or $t'\not\in \phi(A)$. The rank varieties $X_{P,r}(S)\into \Mat_{S\times S}$ are preserved by these maps, and thus form an $\FIsharp$-subscheme $X_{P,r}\into \Mat$. The coordinate rings $\O(\Mat)$ and $\O(X_{P,r})$ are naturally $\FIsharp^{\op}$-algebras, but via the isomor

- `thm:fisubalgebra` (proof_dependency; source_audit_annotation): $M(1)\tensor M(1)$ is finitely generated in degree $2$, so in particular $\wt(M(1)\otimes M(1))=2$. Thus Theorem~\ref{thm:fisubalgebra} implies that $\O(X_{P,r})$ is a graded $\FIsharp$-algebra of finite type and slope $\leq 2$.
- `thm:persinomial` (proof_dependency; source_audit_annotation): The theorem now follows from Theorems~\ref{thm:persinomial} and \ref{thm:polyFIsharp}.
- `thm:polyFIsharp` (proof_dependency; source_audit_annotation): The theorem now follows from Theorems~\ref{thm:persinomial} and \ref{thm:polyFIsharp}.

Candidate references: `remark:FIsharpremarks` (candidate_proof_dependency); `thm:fisubalgebra` (candidate_proof_dependency); `thm:persinomial` (candidate_proof_dependency); `thm:polyFIsharp` (candidate_proof_dependency).

### Remark 5.3.3 — `remark:line:1829:183489`

Source lines 1829–1830.

The rank variety $X_{P,r}(n)$ of Theorem~\ref{theorem:sections:polygrowth} may be non-reduced as a scheme. However, the same theorem holds for the underlying reduced variety. The proof amounts to replacing the ideal $J_n$ by its radical and applying the same argument.

*Extractive statement; proof gist not independently reconstructed.*

- `theorem:sections:polygrowth` (proof_dependency; source_audit_annotation): \begin{remark} The rank variety $X_{P,r}(n)$ of Theorem~\ref{theorem:sections:polygrowth} may be non-reduced as a scheme.  However, the same theorem holds for the underlying reduced variety. The proof amounts to replacing the ideal $J_n$ by its radical and applying the same argument. \end{remark}

Candidate references: `theorem:sections:polygrowth` (internal_reference).

### Definition 6.1.1 — `def:coFIMbullet`

Source lines 1849–1852.

Given a topological space $M$, we define the co-FI-space $M^\bullet$ as follows. For any finite set $S$, we have the topological space $M^S$ of functions from $S$ to $M$. An injection $f\colon T\into S$ induces a projection $f_*\colon M^S\onto M^T$ by restriction, given explicitly by $f_*(\varphi)\coloneq \varphi\circ f\colon T\to S\to M$.

*Extractive statement; proof gist not independently reconstructed.*

### Proposition 6.1.2 — `pr:cohoMbullet`

Source lines 1854–1857.

Let $M$ be a connected topological space, and let $k$ be a field. If $H^i(M;k)$ is finite-dimensional for all $i\geq 0$, the graded FI-algebra $H^*(M^\bullet;k)$ is of finite type.

*Extractive statement; proof gist not independently reconstructed.*

Proof opening: The graded FI-algebra $H^*(M^\bullet;k)$ associates to a finite set $S$ the cohomology $H^*(M^S;k)$. Since $k$ is a field, the K\"{u}nneth theorem implies that $H^*(M^S;k)\simeq H^*(M;k)^{\otimes S}$. It thus seems that $H^*(M^\bullet;k)$ should be isomorphic to the FI-module $H^*(M;k)^{\otimes \bullet}$ defined in Definition~\ref{def:tensorbullet}. This is almost true; the only difference is that the K\"{u}nneth isomorphism $H^*(M^S;k)\simeq H^*(M;k)^{\otimes S}$ depends on an ordering of $S$, and when these factors are permuted, a sign is sometimes introduced depending on the grading. We can nevertheless deduce the proposition from Proposition~\ref{pr:Vtensorbullet}, since it is easy to se


Candidate references: `def:tensorbullet` (candidate_proof_dependency); `pr:Vtensorbullet` (candidate_proof_dependency).

### Remark 6.1.3 — `remark:FIspaces`

Source lines 1865–1877.

Let $M$ be a connected well-based topological space (i.e.\ with specified basepoint $\ast\in M$). We can extend the co-FI-space structure of Definition~\ref{def:coFIMbullet} to consider $M^\bullet$ as an $\FIsharp$-space: given $f\colon S\supset A\xrightarrow{\phi}B\subset T$ and $\varphi\colon S\to M$, we define $f_*\colon M^S\to M^T$ by f_*(\varphi)(t)=\varphi(\phi^{-1}(t))&\text{ if }t\in B\\\ast&\text{ if }t\in T-B \pagebreak Considering $M^\bullet$ as an FI-space, given a map defined on $S$, an inclusion $f\colon S\into T$ simply extends i

*Extractive statement; proof gist not independently reconstructed.*


Candidate references: `def:coFIMbullet` (internal_reference); `external:SagaveSch` (external_reference).

### Theorem 6.2.1 — `theorem:coho:config2`

Source lines 1888–1891.

{\bf Rational cohomology of configuration spaces}

Let $M$ be a connected, oriented manifold of dimension at least 2 with $\dim_\Q(H^*(M;\Q))<\infty$. Then the graded FI-algebra $H^*(\Conf(M);\Q)$ is of finite type.

*Extractive statement; proof gist not independently reconstructed.*

Proof opening: We will consider the Leray spectral sequence of the inclusion of co-FI-spaces $\Conf(M)\into M^\bullet$. Recall that for any continuous map of topological spaces $f\colon X\to Y$, the Leray spectral sequence with $\Q$ coefficients is a cohomological spectral sequence converging to $H^*(X;\Q)$ with $E_2$ page $E_2^{p,q}\simeq H^p(Y;R^q f_*\underline{\Q})$. These entries are in general difficult to compute, with the exception of the edge terms $E_2^{p,0}\simeq H^p(Y;\Q)$. Recall further that the Leray spectral sequence is functorial: if $f'\colon X'\to Y'$ is another continuous map, and $g_X,g_Y$ are maps $g_X\colon X'\to X$ and $g_Y\colon Y'\to Y$ such that $f\circ g_X=g_Y\circ f'$, then $g$


Candidate references: `example:arnoldalgebra` (candidate_proof_dependency); `pr:cohoMbullet` (candidate_proof_dependency); `thm:fisubalgebra` (candidate_proof_dependency); `th:noetherian` (candidate_proof_dependency); `external:To` (external_reference).

### Remark 6.2.2 — `remark:line:1906:194152`

Source lines 1906–1908.

The key technical innovation in \cite{Ch} was the notion of monotonicity, which allows representation stability to be transmitted from the initial term of a spectral sequence to its $E_\infty$ term. We showed in the proof of Theorem~\ref{th:FIequiv} that, for FI-modules, finite generation implies monotonicity. Thus it is not surprising that we are able to imitate the argument of \cite{Ch} in the present paper.

*Extractive statement; proof gist not independently reconstructed.*


Candidate references: `th:FIequiv` (internal_reference); `external:Ch` (external_reference).

### Theorem 6.3.1 — `th:config`

Source lines 1918–1921.

Let $M$ be a connected, oriented manifold of dimension $d \geq 3$. For any $i\geq 0$, the FI-module $H^i(\Conf(M);\Q)$ has weight $\leq i$ and $H^i(\Conf(M);\Q)\stabT(i+2-d,i)$; in particular $\stabD(H^i(\Conf(M);\Q))\leq i$.

*Extractive statement; proof gist not independently reconstructed.*

Proof opening: We continue with the notation from the proof of Theorem~\ref{theorem:coho:config2}. Let $D=d-1$. We saw in that proof that $E_2^{p,\ast}=0$ unless $\ast=qD$. Moreover, Church~\cite[\S3.3]{Ch} gives an explicit description of $E_2^{p,qD}$ as a direct sum $E_2^{p,qD}\simeq \bigoplus_{k=0}^{p+2q}M(W_k)$, where $W_k$ is a certain representation of $S_k$. By Proposition~\ref{pr:stabdegprojective}, this implies that $E_2^{p,qD}\stabT(0,p+2q)$ and $\wt(E_2^{p,qD})\leq p+2q$. The argument proceeds by induction on the pages of the Leray spectral sequence. To be precise, the fact that $E_2^{p,\ast}=0$ unless $\ast=qD$ implies that the only nontrivial differentials occur on pages $E_{kD+1}^{p,qD}$, wit

- `lemma:homologyfacts` (proof_dependency; source_audit_annotation): We now deduce the theorem from the claims $(a_k)$ and $(b_k)$. The FI-module $H^i(\Conf(M);\Q)$ has a finite filtration with graded quotients those $E_\infty^{p,qD}$ with $p+qD=i$. Thus by Lemma~\ref{lemma:homologyfacts}(ii), it suffices to show that $E_\infty^{p,qD}\stabT(i+2-d,i)$ and $\wt(E_\infty^{p,qD})\leq i$ whenever $p+Dq=i$.

Candidate references: `theorem:coho:config2` (candidate_proof_dependency); `pr:stabdegprojective` (candidate_proof_dependency); `lemma:homologyfacts` (candidate_proof_dependency); `eq:kcomplex` (candidate_proof_dependency); `external:Ch` (external_reference).

### Lemma 6.3.2 — `lemma:homologyfacts`

Source lines 1923–1930.

Assume that $k$ contains $\Q$. [(i)] \item Consider a complex of FI-modules $X\xrightarrow{f} Y\xrightarrow{g} Z$ with $g\circ f=0$. If $X\stabT(A,B)$, $Y\stabT(C,D)$, and $Z\stabT(E,F)$, then the homology $\ker g/\im f$ satisfies $\ker g/\im f\stabT\big(\max(B,C),\ \max(D,E)\big)$. \item If $V$ is an FI-module with a finite filtration by sub-FI-modules $F^iV\subset V$, then $V\stabT(A,B)$ if and only if $F^iV/F^{i-1}V\stabT(A,B)$ for all $i$.

*Extractive statement; proof gist not independently reconstructed.*

### Remark 6.3.3 — `remark:line:1986:200582`

Source lines 1986–1989.

If $M$ is a connected oriented manifold of dimension 2 with $\dim_\Q(H^*(M;\Q))<\infty$, the weight of the FI-module $H^i(\Conf M;\Q)$ is bounded by that of $\bigoplus_i E^{p,i-p}_2$, just as in the proof of Theorem~\ref{th:config}. This weight is $p + 2(i-p)$, which is maximized when $p=0$. So in this case $H^i(\Conf M;\Q)$ is a finitely generated FI-module of weight at most $2i$, whence the characters and Betti numbers of $H^i(\Conf_n(M);\Q)$ are eventually polynomial of degree $\leq 2i$.

*Extractive statement; proof gist not independently reconstructed.*


Candidate references: `th:config` (internal_reference).

### Corollary 6.3.4 — `co:vakilwood`

Source lines 1996–2002.

Let $M$ be a connected, oriented manifold of dimension $\geq 3$. Then for all $n \geq i + \abs{\mu}$, \[ H^i(B_{n,\mu}(M);\Q) \simeq H^i(B_{n+1,\mu}(M);\Q). \]

*Extractive statement; proof gist not independently reconstructed.*

Proof opening: The invariant subspace $H^i(\Conf_n(M);\Q)^{S_{n-\abs{\mu}}}$ carries an action of $S_{\abs{\mu}}$, and the dimension of $H^i(B_{n,\mu}(M);\Q)$ is the dimension of the space fixed by the action of \beq S_\mu \coloneq S_{\mu_1} \times \ldots \times S_{\mu_k} \subset S_{\abs{\mu}}. \eeq The statement that $H^i(\Conf_n(M);\Q)$ has stability degree $\leq i$ means by definition that the isomorphism class of $H^i(\Conf_n(M);\Q)^{S_{n-\abs{\mu}}}\simeq H^i(\Conf_n(M);\Q)_{S_{n-\abs{\mu}}}$ as an $S_{\abs{\mu}}$-representation is constant for $n\geq i+\abs{\mu}$. The dimension of the $S_{\abs{\mu}}$-invariant subspace is thus constant for $n\geq i+\abs{\mu}$ as well.

### Definition 6.4.1 — `definition:line:2019:203395`

Source lines 2019–2021.

An homotopy $\FIsharp$-space is a functor $X$ from $\FIsharp$ to $\hTop$, the category of topological spaces and homotopy classes of continuous maps. Concretely, this means that for each $n$ we have a space $X_n$, for each partial injection $f\in \Hom_{\FIsharp}(\m,\n)$ we have a map $f_*\colon X_m\to X_n$ (or rather a homotopy class of maps), and the corresponding diagrams all commute up to homotopy. Since a homotopy class of maps induces a well-defined map on homology and cohomology, both the homology groups $H_i(X;k)$ and cohomology groups $

*Extractive statement; proof gist not independently reconstructed.*

### Proposition 6.4.2 — `pr:CMhTop`

Source lines 2023–2026.

Let $M$ be the interior of a connected, oriented, compact manifold $\overline{M}$ of dimension $d\geq 2$ with nonempty boundary $\partial\overline{M}$. We can extend the FI-space $\Conf(M)\colon \coFI\to \Top$ to a homotopy $\FIsharp$-space $\Conf(M)\colon \FIsharp\to \hTop$.

*Extractive statement; proof gist not independently reconstructed.*

### Theorem 6.4.3 — `thm:ConfMFIsharpfg`

Source lines 2044–2048.

Let $M$ be a connected, oriented manifold of dimension $d\geq 2$ which is the interior of a compact manifold with nonempty boundary. Let $k$ be a Noetherian ring of finite Krull dimension. For each $i\geq 0$, the $\FIsharp$-module $H^i(\Conf(M);k)$ is finitely generated in dimension $\leq i$ if $d\geq 3$, and in dimension $\leq 2i$ if $d=2$.

*Extractive statement; proof gist not independently reconstructed.*

Proof opening: We have already proved the theorem in the case $k=\Q$; the content lies in the case when $k$ has positive characteristic, or is not a field. We recall from the proof of Theorem~\ref{th:config} that each entry $E_2^{pq}$ of the Leray spectral sequence $\Conf(M) \into M^\bullet$ is a finitely generated $\FIsharp$-module. However, we cannot immediately conclude that $E_\infty^{pq}$ or $H^i(\Conf(M);k)$ is finitely generated, even though this spectral sequence converges to $H^*(\Conf(M);k)$. The problem is that $\Conf(M)$ is only a homotopy $\FIsharp$-space, and the Leray spectral sequence is not homotopy invariant. As a result it is not a spectral sequence of $\FIsharp$-modules. Our hypotheses

- `th:config` (proof_dependency; source_audit_annotation): We recall from  the proof of Theorem~\ref{th:config} that each entry $E_2^{pq}$  of the Leray spectral sequence $\Conf(M) \into M^\bullet$ is a finitely generated $\FIsharp$-module. However, we cannot immediately conclude that $E_\infty^{pq}$ or $H^i(\Conf(M);k)$ is finitely generated, even though this spectral sequence converges to $H^*(\Conf(M);k)$. The problem is that $\Conf(M)$ is only a \emph{homotopy} $\FIsharp$-space, and the Leray spectral sequence is not homotopy invariant. As a result it is not a spectral sequence of $\FIsharp$-modules.
- `thm:polyFIsharp` (proof_dependency; source_audit_annotation): First assume that $k$ is a field. From the proof of Theorem~\ref{th:config}, each entry $E_2^{p,q(d-1)}$ of the Leray spectral sequence is a finitely generated $\FIsharp$-module generated in degree $\leq p+2q$. When $d\geq 3$, it follows that $\dim_k (E_2^{pq})_n=O(n^{p+q})$. As a vector space $(E_\infty^{p,q})_n$ is a subquotient of $(E_2^{pq})_n$, so $\dim_k (E_\infty^{pq})_n=O(n^{p+q})$ as well. Finally, $H^i(\Conf_n(M);k)$ has a finite filtration with quotients $(E_\infty^{p,i-p})_n$, so we conclude that $\dim_k H^i(\Conf_n(M);k) = O(n^i)$. By Theorem~\ref{thm:polyFIsharp}, this implies that the $\FIsharp$-module $H^i(\Conf(M);k)$ is finitely generated in degree $\leq i$, as the theorem claims. When $d=2$ we have $\dim_k(E_2^{pq})_n=O(n^{p+2q})$, so $\dim_k H^i(\Conf_n(M);k) = O(n^{2i})$.

Candidate references: `th:config` (candidate_proof_dependency); `thm:polyFIsharp` (candidate_proof_dependency); `external:Forster` (external_reference).

### Theorem 7.1.1 — `theorem:line:2099:213199`

Source lines 2099–2103.

For each $g\geq 2$ the tautological ring $\cR(\cM_{g,\bullet})$ is a graded FI-algebra of finite type. Thus for each $j\geq 1$ the characters $\chi_{\cR^j(\cM_{g,n})}$ are eventually polynomial of degree $\leq j$. In particular, $\dim\cR^j(\cM_{g,n})$ is eventually polynomial in $n$ of degree $\leq j$.

*Extractive statement; proof gist not independently reconstructed.*

Proof opening: Let $\cM_{g,S}$ be the moduli space of genus $g$ Riemann surfaces $X$ endowed with an injection $S\into X$. These spaces form a co-FI-space $\cM_{g,\bullet}$ just as in \S\ref{section:fgconfig}, so the cohomology $H^*(\cM_{g,\bullet};\Q)$ is a graded FI-algebra. For each $s\in S$ we have a line bundle $L_s\to \cM_{g,S}$ with first Chern class $\psi_s\coloneq c_1(L_i)\in H^2(\cM_{g,S};\Q)$. Similarly, we consider the map $p_\star\colon \cM_{g,S\disjoint\{\star\}}\to \cM_{g,S}$, and for each $j\geq 0$ we define $\kappa_j\coloneq (p_\star)_!(\psi_\star^{j+1})\in H^{2j}(\cM_{g,S};\Q)$. These classes define a map of graded FI-modules $t\colon V\to H^*(\cM_{g,\bullet};\Q)$, where $V$ is the graded


Candidate references: `section:fgconfig` (candidate_proof_dependency); `thm:fisubalgebra` (candidate_proof_dependency); `thm:persinomial` (candidate_proof_dependency).

### Remark 7.1.2 — `remark:line:2112:215077`

Source lines 2112–2115.

Jimenez Rolland~\cite{J1} proved the related theorem that for each fixed $g\geq 2$ and $i\geq 0$, the sequence $\{H^i(\cM_{g,n};\Q)\}$ is a uniformly representation stable sequence of $S_n$-representations. Since $H^i(\cM_{g,\bullet};\Q)$ is an FI-module, and since $H^\ast(\cM_{g,n};\Q)$ is finite dimensional, Theorem~\ref{th:FIequiv} together with Jimenez Rolland's theorem shows that $\{H^\ast(\cM_{g,\bullet};\Q)\}$ is a graded FI-module of finite type. This result, with bounds on the stability degree, etc., has been worked out by Jimenez Roll

*Extractive statement; proof gist not independently reconstructed.*


Candidate references: `th:FIequiv` (internal_reference); `external:J1` (external_reference); `external:J2` (external_reference).

### Definition 7.2.1 — `definition:line:2124:216199`

Source lines 2124–2127.

{\bf Albanese cohomology}

Let $\Gamma$ be a finitely generated group, let $\Gamma^{\rm ab}$ be its abelianization, and let $\psi\colon\Gamma\to \Gamma^{\rm ab}$ be the natural quotient map. Since $H^1(\Gamma^{\ab};\Q)\simeq H^1(\Gamma;\Q)$ and $H^*(\Gamma^{\rm ab};\Q)\simeq \bwedge^*H^1(\Gamma^{\rm ab};\Q)$, the map $\psi$ induces a homomorphism \[\psi^*\colon H^*(\Gamma^{\rm ab};\Q)\simeq \bwedge^*H^1(\Gamma;\Q)\to H^*(\Gamma;\Q).\] We define the Albanese cohomology $H_{\rm Alb}^*(\Gamma;\Q)$ of $\Gamma$ to be the image of this map: \[H_{\rm Alb}^*(\Gamma;\Q)\coloneq \

*Extractive statement; proof gist not independently reconstructed.*

### Theorem 7.2.2 — `theorem:albanese1`

Source lines 2147–2150.

For each $i\geq 0$ there exists a polynomial $P_i(T)$ of degree at most $3i$ such that $\dim H^i_{\rm Alb}(\I^1_g;\Q)=P_i(g)$ for $g\gg i$.

*Extractive statement; proof gist not independently reconstructed.*

Proof opening: We define the FI-group up to conjugacy $\I_\bullet^1$ as follows. Fix for each $g$ an ordered symplectic basis $\{a_1,b_1,\ldots,a_g,b_g\}$ for $H_1(S_g^1;\Z)$. Any injection $f\colon \m\into \n$ determines an injection $f_*\colon H_1(S_m^1;\Z)\into H_1(S_n^1;\Z)$ by $f_*(a_i)=a_{f(i)}$ and $f_*(b_i)=b_{f(i)}$. Any embedding $S_m^1\into S_n^1$ determines an inclusion $H_1(S_m^1;\Z)\into H_1(S_n^1;\Z)$ as a symplectic summand. Such an embedding also determines an injection $\I_m^1\into \I_n^1$, given by extending each map to be the identity in the complement. Moreover, although there are many embeddings $S_m^1\into S_n^1$ inducing a given map on homology, Johnson \cite[Theorem 1A]{JohnsonCR}


Candidate references: `eq:Johnsonhom` (candidate_proof_dependency); `pr:schurfunctor` (candidate_proof_dependency); `pr:cofisubalgebra` (candidate_proof_dependency); `thm:persinomial` (candidate_proof_dependency); `external:JohnsonCR` (external_reference).

### Theorem 7.2.3 — `theorem:albanese2`

Source lines 2181–2184.

For each $i\geq 0$ there exists a polynomial $P_i(T)$ of degree at most $3i$ such that $\dim H^i_{\rm Alb}(\IAn;\Q)=P_i(n)$ for $n\gg i$.

*Extractive statement; proof gist not independently reconstructed.*

Proof opening: The proof proceeds along much the same lines as that of Theorem~\ref{theorem:albanese1}. One simplifying factor is that in this case we can actually define an FI-group $\IA_\bullet$ as follows. To $\n$ we associate the group $\IA_n$ defined above, and to an injection $f\colon \m\to \n$ we associate the following homomorphism $f_*\colon \IA_m\into\IAn$. Let $f_*\colon F_m\into F_n$ denote the inclusion induced by $x_i\mapsto x_{f(i)}$. Given $\varphi\in \IA_m$, we define $f_*\varphi\in \IAn$ to be the automorphism sending $x_{f(i)}\mapsto f_*(\varphi(x_i))$ and $x_j\mapsto x_j$ for those $j\not\in f(\m)$. It is easy to check that $(f\circ g)_*=f_*\circ g_*$, so this defines an FI-group $\IA_\


Candidate references: `theorem:albanese1` (candidate_proof_dependency); `eq:H1IAn` (candidate_proof_dependency); `pr:tensor` (candidate_proof_dependency); `pr:schurfunctor` (candidate_proof_dependency); `pr:cofisubalgebra` (candidate_proof_dependency); `thm:persinomial` (candidate_proof_dependency).

### Theorem 7.3.1 — `theorem:assoc:graded`

Source lines 2223–2226.

{\bf Finite generation of $\gr(\Gamma)$}

If $\Gamma$ is an FI-group up to conjugacy (e.g.\ the fundamental group of a homotopy FI-space) and the FI-module $H_1(\Gamma;k)$ is finitely generated, the graded FI-module $\gr(\Gamma)$ is of finite type.

*Extractive statement; proof gist not independently reconstructed.*

Proof opening: For any group $\Gamma$, the terms $\Gamma_i$ of the lower central series are by definition generated by iterated commutators of elements of $\Gamma$. This shows that $\gr(\Gamma)$ is always generated as a Lie algebra by $\gr(\Gamma)_1\simeq H_1(\Gamma;k)$. Thus if $H_1(\Gamma;k)$ is finitely generated, Theorem~\ref{thm:fisubalgebra} implies the graded FI-algebra $\gr(\Gamma)$ is of finite type.


Candidate references: `thm:fisubalgebra` (candidate_proof_dependency).

### Example 7.3.2 — `xample:line:2233:228503`

Source lines 2233–2235.

{\bf Free groups}

Let $\Gamma$ be the $\FIsharp$-group which assigns to $S$ the free group $F_S=\langle x_s\,|\, s\in S\rangle$, and to a morphism $f\colon S\supset A\xrightarrow{\phi}B\subset T$ assigns the map $F_S\to F_T$ given by $\phi(x_s)=x_{\phi(s)}$ if $i\in A$, and $\phi(x_s)=1$ if $i\in S-A$. It is easy to compute that $\gr(\Gamma)_1\simeq M(1)$. Since $\gr(\Gamma)_1$ is finitely generated, Theorem~\ref{theorem:assoc:graded} implies that $\gr(\Gamma)$ is of finite type. It is known that $\gr(F_n)$ is isomorphic to the free Lie algebra on $n$ variables.

*Extractive statement; proof gist not independently reconstructed.*


Candidate references: `theorem:assoc:graded` (internal_reference); `th:FIequiv` (internal_reference); `external:CF` (external_reference).

### Theorem 7.3.4 — `theorem:malcevpn`

Source lines 2248–2252.

{\bf Representation stability for $\fp_n$}

For each fixed $i\geq 1$, the sequence $\{\fp_n^i\}$ of grading-$i$ pieces of $\fp_n$ is a uniformly representation stable sequence of $S_n$--representations.

*Extractive statement; proof gist not independently reconstructed.*

### Remark 7.3.5 — `remark:line:2256:230890`

Source lines 2256–2259.

The pure braid groups $P_n$ are examples of pseudo-nilpotent groups, so that $H^*(\fp_n;\Q)\simeq H^*(P_n;\Q)$ for all $n$. It was already proved in \cite{CF} that the sequence $\{H^i(P_n;\Q)\}$ is uniformly representation stable for each $i \geq 0$, so one is tempted to derive Theorem~\ref{theorem:malcevpn} directly from \cite[Theorem 5.3]{CF}, which states the equivalence of uniform representation stability for a Lie algebra and for its (co)homology. However, that theorem was only proved in \cite{CF} in the context of stability for $\SL_n\C$-

*Extractive statement; proof gist not independently reconstructed.*


Candidate references: `theorem:malcevpn` (internal_reference); `external:CF` (external_reference).

### Example 7.3.3 — `example:pbgraded`

Source lines 2239–2260.

{\bf Pure braid groups}

We proved in Proposition~\ref{pr:CMhTop} that the configuration space $\Conf_n(\R^2)$ of ordered $n$-tuples of distinct points in the plane is a homotopy $\FIsharp$-space. The pure braid group $P_n$ on $n$ strands is the fundamental group $P_n=\pi_1(\Conf_n(\R^2))$, so this shows that $P_\bullet$ is an $\FIsharp$-group up to conjugacy. \remove{We proved in Example~\ref{example:arnoldalgebra} that $H_1(P_\bullet;\Q)\simeq M(\y{2})$, } \antiremove{One can compute directly that $H_1(P_\bullet;\Q)\simeq M(\y{2})$, } so Theorem~\ref{theorem:assoc:gr

*Extractive statement; proof gist not independently reconstructed.*


Candidate references: `pr:CMhTop` (internal_reference); `example:arnoldalgebra` (internal_reference); `theorem:assoc:graded` (internal_reference); `th:FIequiv` (internal_reference); `theorem:malcevpn` (internal_reference); `thm:fisubalgebra` (internal_reference); `external:Ha` (external_reference); `external:CF` (external_reference); `external:Ko` (external_reference).

### Theorem 7.3.7 — `theorem:line:2268:232565`

Source lines 2268–2269.

For each $i\geq 0$ the dimension $\dim(\gr(\I_g^1)^i)$ is polynomial in $g$ for $g\gg i$.

*Extractive statement; proof gist not independently reconstructed.*

### Theorem 7.3.8 — `theorem:line:2270:232684`

Source lines 2270–2272.

For each $i\geq 0$ the dimension $\dim(\gr(\IAn)^i)$ is polynomial in $n$ for $n\gg i$.

*Extractive statement; proof gist not independently reconstructed.*

Proof opening: We showed in the proof of Theorems~\ref{theorem:albanese1} and \ref{theorem:albanese2} that $\I_\bullet^1$ is an FI-group up to conjugacy and $\IA_\bullet$ is an FI-group. Moreover, we saw that the FI-modules $H_1(\I_\bullet^1;\Q)\simeq \bwedge^3(M(1)\oplus M(1))$ and $H_1(\IA_\bullet;\Q)\simeq \bwedge^2 M(1) \otimes M(1)$ are finitely generated. Theorem~\ref{theorem:assoc:graded} now implies that the graded FI-algebras $\gr(\I_\bullet^1)$ and $\gr(\IA_\bullet)$ are of finite type, and the claim follows from Theorem~\ref{thm:persinomial}.


Candidate references: `theorem:albanese1` (candidate_proof_dependency); `theorem:albanese2` (candidate_proof_dependency); `theorem:assoc:graded` (candidate_proof_dependency); `thm:persinomial` (candidate_proof_dependency).

### Example 7.3.6 — `xample:line:2263:231683`

Source lines 2263–2278.

{\bf The Torelli group}

In \S\ref{subsection:albanese} we encountered the Torelli group $\I_g^1$. The action of the mapping class group $\Mod(S_g^1)$ on $\I_g^1$ by conjugation induces a well-defined action of $\Sp_{2g}\Z$ on $\gr(\I_g^1)$, taking $k=\Q$. A finite presentation for $\gr(\I_g^1)$ as a Lie algebra has been given by Habegger--Sorger~\cite{HS}, extending the fundamental computation of Hain~\cite{Ha} in the case of closed surfaces. Hain also worked out the first few graded terms of this Lie algebra explicitly as $\Sp_{2g}\Z$-representations. Getzler--Garouf

*Extractive statement; proof gist not independently reconstructed.*


Candidate references: `subsection:albanese` (internal_reference); `external:HS` (external_reference); `external:Ha` (external_reference).

## External inputs and references

### [AM] Aguiar — `external:Aguiar`

Source lines 2286–2288.

M. Aguiar and S. Mahajan, Monoidal Functors, Species, and Hopf Algebras, CRM Monograph Series, Vol 29 (2010).


Candidate references: `section:1` (external_reference); `section:setup` (external_reference); `section:3` (external_reference).

### [AAB] AAB — `external:AAB`

Source lines 2288–2292.

S. Ashraf, H. Azam and B. Berceanu, Representation stability of power sets and square free polynomials, \arXiv{1106.4926v1}, preprint 2011. \remove{


Candidate references: `xample:line:1663:164760` (external_reference).

### [Ar] Ar — `external:Ar`

Source lines 2292–2300.

V.I. Arnol'd, The cohomology ring of the colored braid group, Mathematical Notes 5, no. 2 (1969), 138--140. }


Candidate references: `example:arnoldalgebra` (external_reference).

### [Be] Be — `external:Be`

Source lines 2300–2305.

F. Bergeron, Multivariate diagonal coinvariant spaces for complex reflection groups, Adv. Math. 239 (2013), 97--108. \arXiv{1105.4358}.


Candidate references: `section:1` (external_reference).

### [BS] BS — `external:BS`

Source lines 2305–2308.

M. Bhargava and M. Satriano, On a notion of ``Galois closure'' for extensions of rings, to appear in J. Eur. Math. Soc. \arXiv{1006.2562}.


Candidate references: `section:FIalgebrasandcoinv` (external_reference); `proposition:line:1760:176167` (external_reference).

### [Che] Chevalley — `external:Chevalley`

Source lines 2308–2310.

C. Chevalley, Invariants of finite groups generated by reflections, Amer. J. Math. 77 (1955), 778--782.


Candidate references: `section:1` (external_reference); `section:FIalgebrasandcoinv` (external_reference).

### [CE] CE — `external:CE`

Source lines 2310–2312.

T. Church and J. S. Ellenberg, Homology of FI-modules and stability, in preparation.


Candidate references: `section:1` (external_reference); `remark:line:800:69509` (external_reference).

### [CEF] CEF2 — `external:CEF2`

Source lines 2312–2314.

T. Church, J. S. Ellenberg, and B. Farb, Representation stability in cohomology and asymptotics for families of varieties over finite fields, Contemp. Math. 620 (2014), 1--54. \arXiv{1309.6038}.


Candidate references: `section:1` (external_reference).

### [CEFN] CEFN — `external:CEFN`

Source lines 2314–2316.

T. Church, J. S. Ellenberg, B. Farb, and R. Nagpal, FI-modules over Noetherian rings, to appear in Geometry \& Topology. \arXiv{1210.1854}.


Candidate references: `section:1` (external_reference).

### [CF] CF — `external:CF`

Source lines 2316–2318.

T. Church and B. Farb, Representation theory and homological stability, Adv. Math. (2013), 250--314. \arXiv{1008.1368}.


Candidate references: `section:1` (external_reference); `theorem:intro:FIequiv` (external_reference); `section:3` (external_reference); `definition:repstab` (external_reference); `example:arnoldalgebra` (external_reference); `xample:line:1736:172087` (external_reference); `xample:line:2233:228503` (external_reference); `example:pbgraded` (external_reference); `remark:line:2256:230890` (external_reference).

### [CF2] CF2 — `external:CF2`

Source lines 2318–2321.

T. Church and B. Farb, Parameterized Abel--Jacobi maps and abelian cycles in the Torelli group, J. Topol. 5 (2012), no. 1, 15--38. \arXiv{1001.1114}.


Candidate references: `section:moduli` (external_reference).

### [Ch] Ch — `external:Ch`

Source lines 2321–2324.

T. Church, Homological stability for configuration spaces of manifolds, Invent. Math. 188 (2012), no. 2, 465--504. \arXiv{1103.2441}.


Candidate references: `section:1` (external_reference); `section:3` (external_reference); `th:FIequiv` (external_reference); `pr:Vlambda` (external_reference); `corollary:line:1223:116053` (external_reference); `section:configspaces` (external_reference); `remark:line:1906:194152` (external_reference); `th:config` (external_reference).

### [CP] CP — `external:CP`

Source lines 2324–2328.

T. Church and A. Putman, Generating the Johnson filtration, to appear in Geometry \& Topology. \arXiv{1311.7150}.


Candidate references: `section:1` (external_reference).

### [De] De — `external:De`

Source lines 2328–2331.

P. Deligne, La Cat\'{e}gorie des Repr\'{e}sentations du Groupe Sym\'{e}trique $S_t$, lorsque $t$ n'est pas un Entier Naturel, in Algebraic groups and homogeneous spaces, 209--273, Tata Inst. Fund. Res. Stud. Math., Mumbai, 2007.


Candidate references: `section:1` (external_reference).

### [DK] DK — `external:DK`

Source lines 2331–2334.

J. Draisma and J. Kuttler, Bounded-rank tensors are defined in bounded degree, Duke Math. J. 163 (2014), no. 1, 35--63, 2014. \arXiv{1103.5336}.


Candidate references: `section:FIalgebrasandcoinv` (external_reference).

### [DV] DV — `external:DV`

Source lines 2334–2339.

A. Djament and C. Vespa, Sur l'homologie des groupes orthogonaux et symplectiques \`{a} coefficients tordus, \emph{Ann. Sci. \'{E}cole Norm. Sup} 43 (2010) 4, 395--459. \arXiv{0808.4035}.


Candidate references: `section:1` (external_reference).

### [ES] ES — `external:ES`

Source lines 2339–2342.

D. Eisenbud and D. Saltman, Rank varieties of matrices, in Commutative Algebra, ed. by M. Hochster, C. Huneke and J.D. Sally, Springer-Verlag, 1989.


Candidate references: `section:FIalgebrasandcoinv` (external_reference).

### [Et] Et — `external:Et`

Source lines 2342–2344.

P. Etingof, Representation theory in complex rank, Conference talk at the Isaac Newton Institute for Mathematical Sciences, 2009. Available at:\newline \url{http://www.newton.ac.uk/programmes/ALT/seminars/032716301.html}.


Candidate references: `section:1` (external_reference).

### [FM] FM — `external:FM`

Source lines 2344–2347.

B. Farb and D. Margalit, A Primer on Mapping Class Groups, Princeton Mathematical Series 49, Princeton University Press, 2012.


Candidate references: `section:moduli` (external_reference).

### [Fo] Forster — `external:Forster`

Source lines 2347–2350.

O. Forster, \"{U}ber die Anzahl der Erzeugenden eines Ideals in einem Noetherschen Ring, Math. Z. 84 (1964), 80--87.


Candidate references: `thm:ConfMFIsharpfg` (external_reference).

### [FH] FH — `external:FH`

Source lines 2350–2354.

W. Fulton and J. Harris, Representation theory. A first course, Graduate Texts in Mathematics 129, Readings in Mathematics, Springer-Verlag, New York, 1991.


Candidate references: `section:3` (external_reference); `pr:Vlambda` (external_reference).

### [FMac] FuM — `external:FuM`

Source lines 2354–2358.

W. Fulton and R. MacPherson, A compactification of configuration spaces, Ann. Math 139 (1994) 1, 183--225.


Candidate references: `section:1` (external_reference).

### [GG] GG — `external:GG`

Source lines 2358–2360.

A. Garsia and A. Goupil, Character polynomials, their $q$-analogs and the Kronecker product, Elect. Jour. Combin. 16(2) 2009, \#R19.


Candidate references: `thm:persinomial` (external_reference).

### [GL] GanLi — `external:GanLi`

Source lines 2360–2362.

W. L. Gan and L. Li, Noetherian property of infinite EI categories, \arXiv{1407.8235v1}, preprint 2014.


Candidate references: `section:1` (external_reference).

### [Ha] Ha — `external:Ha`

Source lines 2362–2366.

R. Hain, Infinitesimal presentations of the {T}orelli groups, J. Amer. Math. Soc. 10 (1997), no. 3, 597--651.


Candidate references: `section:moduli` (external_reference); `example:pbgraded` (external_reference); `xample:line:2263:231683` (external_reference).

### [Hai] Hai — `external:Hai`

Source lines 2366–2370.

M. Haiman, Combinatorics, symmetric functions, and Hilbert schemes, in Current developments in mathematics, 2002, 39--111, Int. Press, Somerville, MA, 2003.


Candidate references: `section:1` (external_reference).

### [Hai2] HaimanInventiones — `external:HaimanInventiones`

Source lines 2370–2373.

M. Haiman, Vanishing theorems and character formulas for the Hilbert scheme of points in the plane, Invent. Math. 149 (2002), 371--407. \arXiv{math/0201148}.


Candidate references: `section:1` (external_reference).

### [HHLRU] HHLRU — `external:HHLRU`

Source lines 2373–2376.

J. Haglund, M. Haiman, N. Loehr, J. Remmel and A. Ulyanov, A combinatorial formula for the character of the diagonal coinvariants, Duke Math. J. 126 (2005), pp. 195-232. \arXiv{math/0310424}.


Candidate references: `section:1` (external_reference).

### [HS] HS — `external:HS`

Source lines 2376–2380.

N. Habegger and C. Sorger, An infinitesimal presentation of the Torelli group of a surface with boundary, preprint 2000.


Candidate references: `xample:line:2263:231683` (external_reference).

### [He] Helmstutler — `external:Helmstutler`

Source lines 2380–2382.

R. Helmstutler, Model category extensions of the Pirashvili-S{\l}omi\'{n}ska theorems, \arXiv{0806.1540v1}, preprint 2008.


Candidate references: `section:1` (external_reference).

### [JR1] J1 — `external:J1`

Source lines 2382–2385.

R. Jimenez Rolland, Representation stability for the cohomology of moduli space ${\cal M}_{g}^{n}$, Algebr. Geom. Topol. 11 (2011), 3011--3041. \arXiv{1106.0947}.


Candidate references: `section:1` (external_reference); `remark:line:2112:215077` (external_reference).

### [JR2] J2 — `external:J2`

Source lines 2385–2391.

R. Jimenez Rolland, On the cohomology of pure mapping class groups as FI-modules, to appear in J. Homotopy Relat. Struct. \arXiv{1207.6828}.


Candidate references: `remark:line:2112:215077` (external_reference).

### [Jo2] JohnsonCR — `external:JohnsonCR`

Source lines 2391–2395.

D. Johnson, Conjugacy relations in subgroups of the mapping class group and a group-theoretic description of the Rochlin invariant, Math. Ann. 249 (1980), no. 3, 243--263.


Candidate references: `theorem:albanese1` (external_reference).

### [Ka] Ka — `external:Ka`

Source lines 2395–2398.

N. Kawazumi, Cohomological aspects of Magnus expansions, \arXiv{math/0505497v3}, preprint 2006.


Candidate references: `section:moduli` (external_reference).

### [Kn] Kn — `external:Kn`

Source lines 2398–2401.

F. Knop, Tensor envelopes of regular categories, Adv. Math. 214 (2007) 2, 571--617. \arXiv{math/0610552}. \remove{


Candidate references: `section:1` (external_reference).

### [KT] KT — `external:KT`

Source lines 2401–2404.

A. Knutson and T. Tao, The honeycomb model of $\GL_n(\C)$ tensor products I: Proof of the saturation conjecture, J. Amer. Math. Soc. 12 (1999) 4, 1055--1090. \arXiv{math/9807160}. }


Candidate references: `section:4` (external_reference).

### [Ko] Ko — `external:Ko`

Source lines 2404–2408.

T. Kohno, S\'{e}rie de Poincar\'{e}-Koszul associ\'{e}e aux groupes de tresses pures, Invent. Math. 82 (1985), 57--75.


Candidate references: `example:pbgraded` (external_reference).

### [Li] Li — `external:Li`

Source lines 2408–2411.

D.E. Littlewood, Products and plethysms of characters with orthogonal, symplectic and symmetric groups, Canad. J. Math. 10 (1958), 17--32.


Candidate references: `section:1` (external_reference); `section:3` (external_reference).

### [L] Lue — `external:Lue`

Source lines 2411–2414.

W. L\"{u}ck, Transformation Groups and Algebraic $K$-Theory, Lecture Notes in Mathematics 1408, Mathematica Gottingensis, Springer-Verlag, Berlin, 1989.


Candidate references: `section:1` (external_reference); `rem:projective` (external_reference).

### [Mac] Mac — `external:Mac`

Source lines 2414–2417.

I.G. Macdonald, {\it Symmetric functions and Hall polynomials}, second ed., Oxford Math. Mon., Clarendon Press, Oxford, 1995.


Candidate references: `section:1` (external_reference); `thm:persinomial` (external_reference).

### [McD] McD — `external:McD`

Source lines 2417–2419.

D. McDuff, Configuration spaces of positive and negative particles, Topology 14 (1975), 91--107.


Candidate references: `section:1` (external_reference).

### [MacL] MacL — `external:MacL`

Source lines 2419–2421.

S. Mac Lane, Categories for the working mathematician, Springer GTM Series Vol.\ 5, 2nd ed., 1998.


Candidate references: `def:MW` (external_reference).

### [Me] Mess — `external:Mess`

Source lines 2421–2423.

G. Mess, The Torelli groups for genus 2 and 3 surfaces, Topology 31 (1992), 775--790.


Candidate references: `section:moduli` (external_reference).

### [Pe] Pe — `external:Pe`

Source lines 2423–2426.

A. Pettet, The Johnson homomorphism and the second cohomology of $\IA_n$, Algebr. Geom. Topol. 5 (2005), 725--740. \arXiv{math/0501053}.


Candidate references: `section:moduli` (external_reference).

### [Pi] Pi — `external:Pi`

Source lines 2426–2429.

T. Pirashvili, Dold-Kan type theorem for $\Gamma$-groups, Math. Ann. 318 (2000) 2, 277--298.


Candidate references: `section:1` (external_reference); `remark:line:1296:122151` (external_reference).

### [Pu] Pu — `external:Pu`

Source lines 2429–2432.

A. Putman, Stability in the homology of congruence subgroups, \arXiv{1201.4876v4}, preprint 2012.


Candidate references: `section:1` (external_reference).

### [PuS] PutmanSam — `external:PutmanSam`

Source lines 2432–2437.

A. Putman and S. Sam, Representation stability and finite linear groups, \arXiv{1408.3694v1}, preprint 2014.


Candidate references: `section:1` (external_reference).

### [RS] RS — `external:RS`

Source lines 2437–2440.

Q. Ren and T. Schedler, On the asymptotic $S_n$-structure of invariant differential operators on symplectic manifolds, J. Algebra 356 (1), 39--89. \arXiv{1006.0268}.


Candidate references: `section:1` (external_reference).

### [Sa] Sa — `external:Sa`

Source lines 2440–2444.

T. Sakasai, The Johnson homomorphism and the third rational cohomology group of the Torelli group, {\em Topology and its Applications} 148 (2005) 83--111.


Candidate references: `section:moduli` (external_reference).

### [SSc] SagaveSch — `external:SagaveSch`

Source lines 2444–2446.

S. Sagave and C. Schlichtkrull, Diagram spaces and symmetric spectra, Adv. Math. 231 (2012), no. 3-4, 2116--2193. \arXiv{1103.2764}.


Candidate references: `remark:FIspaces` (external_reference).

### [SS] SS — `external:SS`

Source lines 2446–2449.

S. Sam and A. Snowden, $\GL$-equivariant modules over polynomial rings in infinitely many variables, \arXiv{1206.2233v2}, preprint 2013.


Candidate references: `section:1` (external_reference).

### [SS2] SSsurvey — `external:SSsurvey`

Source lines 2449–2452.

S. Sam and A. Snowden, Introduction to twisted commutative algebras, \arXiv{1209.5122v1}, preprint 2012.


Candidate references: `section:1` (external_reference).

### [SS3] SnowdenSamGrobner — `external:SnowdenSamGrobner`

Source lines 2452–2457.

S. Sam and A. Snowden, {G}r\"{o}bner methods for representations of combinatorial categories, preprint 2014.


Candidate references: `section:1` (external_reference).

### [Se] Se — `external:Se`

Source lines 2457–2459.

G. Segal, The topology of spaces of rational functions, Acta Mathematica 143 (1979), no. 1, 39--72.


Candidate references: `section:1` (external_reference).

### [S\l] Slominska — `external:Slominska`

Source lines 2459–2461.

J. S{\l}omi\'{n}ska, Dold--Kan type theorems and Morita equivalences of functor categories, J. Algebra 274 (2004), no. 1, 118--137.


Candidate references: `section:1` (external_reference).

### [Sn] Snowden — `external:Snowden`

Source lines 2461–2464.

A. Snowden, Syzygies of Segre embeddings and $\Delta$-modules, Duke Math J. 162 (2013), no. 2, 225--277. \arXiv{1006.5248}.


Candidate references: `section:1` (external_reference).

### [So] So — `external:So`

Source lines 2464–2469.

L. Solomon, Representations of the rook monoid, J. Algebra 256 (2002), no. 2, 309--342.


Candidate references: `remark:FIsharpremarks` (external_reference).

### [To] To — `external:To`

Source lines 2469–2473.

B. Totaro, Configuration spaces of algebraic varieties, Topology 35 (1996), no. 4, 1057--1067.


Candidate references: `section:1` (external_reference); `theorem:coho:config2` (external_reference).

### [Va] Va — `external:Va`

Source lines 2473–2476.

R. Vakil, The moduli space of curves and its tautological ring, {\it Notices Amer. Math. Soc.} 50 (2003), no. 6, 647-658. Available at: \url{http://www.ams.org/notices/200306/fea-vakil.pdf}


Candidate references: `section:moduli` (external_reference).

### [VW] VakilWood — `external:VakilWood`

Source lines 2476–2478.

R. Vakil and M.M. Wood, Discriminants in the Grothendieck ring, \arXiv{1208.3166v1}, preprint 2012.


Candidate references: `section:configspaces` (external_reference).

### [Wei] We — `external:We`

Source lines 2478–2480.

C. Weibel, An introduction to homological algebra, Cambridge Studies in Advanced Mathematics 38, Cambridge University Press, Cambridge, 1994.


Candidate references: `remark:line:622:50374` (external_reference); `rem:projective` (external_reference).

### [Wey] Wey — `external:Wey`

Source lines 2480–2483.

J. Weyman, Cohomology of Vector Bundles and Syzygies, Cambridge Tracts in Math., Vol.\ 149, Cambridge Univ. Press, 2003.


Candidate references: `section:FIalgebrasandcoinv` (external_reference).

### [Weyl] Weyl — `external:Weyl`

Source lines 2483–2486.

H. Weyl, The Classical Groups, their Invariants and Representations, Princeton Math. Series, Vol.~1, Princeton, 1946.


Candidate references: `section:1` (external_reference).

### [Wi1] WiRepstab — `external:WiRepstab`

Source lines 2486–2488.

J. Wilson, Representation stability for the cohomology of the pure string motion groups, {\em Alg. Geom. Top.} 12 (2012) 909--931. \arXiv{1108.1255}.


Candidate references: `section:1` (external_reference).

### [Wi2] JennyFIW1 — `external:JennyFIW1`

Source lines 2488–2490.

J. Wilson, $\FI_W$-modules and stability criteria for representations of classical Weyl groups. Available at: \url{http://math.uchicago.edu/~wilsonj/research.html}


Candidate references: `section:1` (external_reference).

### [Wi3] JennyFIW2 — `external:JennyFIW2`

Source lines 2490–2494.

J. Wilson, $\FI_W$-modules and constraints on classical Weyl group characters. Available at:\linebreak \url{http://math.uchicago.edu/~wilsonj/research.html}


Candidate references: `section:1` (external_reference).

## Coverage and limits

{"mode": "native_latex_index", "claims": 114, "sections": 7, "external_references": 66, "source_bibliography_entries": 66, "unresolved_bibliography": 0, "unresolved": 0}

- Reference edges are not certified proof dependencies. Edge direction is prerequisite → consumer.
- Numbers reflect simple source counters; labels and line anchors are authoritative.
- Extractive statements and proof openings are marked; they are not reconstructed proof gists.
- Macros, included files and custom counters require an expanded source or annotated fallback.
- Printed labels replay the frozen TeX declarations: subsection theorem counter, separate alphabetical counter, and separate introductory counters. This does not claim a full journal collation.
- Only the twelve explicitly located proof dependencies are role-classified. An index of 114 claim environments is not 114 completed audits; unclassified cross-references are not certified dependencies. Review statuses are excluded.
