Paper XIIReading edition

Finite Hecke packets and infinite occupancy

Reading tools
Reading libraryFind statementsInspect Lean
Switch work

Saved reading

Read side by side

Start reading

Resources & source

Original TeX with line links · Download TeX

The wording and mathematics follow the retained manuscript. This reading format does not add a proof review or extend Lean coverage.

Skip contents and start reading

8 Finite Hecke packets and infinite occupancy

Finite Hecke carriers admit canonical packets even when the Hecke action is not semisimple. The packets come from the Artin local factors of the finite commutative Hecke algebra.

Standard fact 8.1: Artin decomposition

Let \(E'\) be a field and let \({\mathbb T}\) be a finite dimensional commutative \(E'\) algebra. Then

\[ \boxed{ {\mathbb T} \cong \prod_{{\mathfrak m}\in{\operatorname{Max}}{\mathbb T}}{\mathbb T}_{{\mathfrak m}}. } \]

Each factor is an Artin local algebra. Equivalently, there is a unique complete family of central idempotents \((e_{{\mathfrak m}})\) such that

\[ e_{{\mathfrak m}}e_{\mathfrak n}=0 \quad({\mathfrak m}\neq\mathfrak n), \qquad \sum_{{\mathfrak m}}e_{{\mathfrak m}}=1, \]

and \(e_{{\mathfrak m}}{\mathbb T}\cong{\mathbb T}_{{\mathfrak m}}\) [2], [15].

Theorem 8.2: Canonical Hecke packet decomposition

Let \({\mathbb T}\) be a finite dimensional commutative algebra over a field \(E'\), and let \(V\) be a finite \({\mathbb T}\) module. Then

\[ \boxed{ V = \bigoplus_{{\mathfrak m}\in{\operatorname{Max}}{\mathbb T}}e_{{\mathfrak m}}V. } \]

For every maximal ideal \({\mathfrak m}\), the canonical map gives

\[ \boxed{ e_{{\mathfrak m}}V\cong V_{{\mathfrak m}}.} \]

The packet \(e_{{\mathfrak m}}V\) is nonzero exactly when \({\mathfrak m}\in{\operatorname{Supp}}_{{\mathbb T}}(V)\).

Proof of Theorem 8.2

The idempotent relations in Standard fact 8.1 give the direct sum decomposition. Localization at \({\mathfrak m}\) kills every factor except the \({\mathfrak m}\) factor, where \(e_{{\mathfrak m}}\) becomes one. Thus the localization map identifies \(e_{{\mathfrak m}}V\) with \(V_{{\mathfrak m}}\). The final statement is the definition of support.

Corollary 8.3: Simultaneous generalized eigenspaces

Assume that \(E'\) is algebraically closed and that \({\mathbb T}\) is generated by commuting operators \(T_1,\ldots,T_d\) on \(V\). For a maximal ideal

\[ {\mathfrak m} = (T_1-\lambda_1,\ldots,T_d-\lambda_d), \]

there is an integer \(N\geq1\) such that

\[ \boxed{ e_{{\mathfrak m}}V = \left\{ v\in V: (T_i-\lambda_i)^Nv=0 \text{ for every }i \right\}. } \]

Thus the canonical Artin packet is the simultaneous generalized eigenspace.

Proof of Corollary 8.3

The maximal ideal of the Artin local algebra \({\mathbb T}_{{\mathfrak m}}\) is nilpotent. Hence a common power of every \(T_i-\lambda_i\) kills the local packet. On every other local factor, at least one \(T_i-\lambda_i\) is a unit, so no nonzero vector can satisfy all the displayed nilpotence conditions.

Conversely, suppose every \((T_i-\lambda_i)^N\) kills \(v\). Every monomial of total degree greater than \(d(N-1)\) contains one factor to exponent at least \(N\). Therefore a power of \({\mathfrak m}\) kills \(v\), which places \(v\) in the \({\mathfrak m}\) primary packet.

Corollary 8.4: Rank and trace detection in characteristic zero

Assume that \(E'\) has characteristic zero. For every maximal Hecke packet,

\[ \boxed{ {\operatorname{Tr}}_{E'}(e_{{\mathfrak m}}|V) = {\operatorname{rank}}_{E'}(e_{{\mathfrak m}}|V) = \dim_{E'}e_{{\mathfrak m}}V. } \]

In particular, the packet is occupied exactly when its projector has positive rank, equivalently nonzero trace.

Proof of Corollary 8.4

An idempotent linear map has minimal polynomial dividing \(X(X-1)\), which has distinct roots. It is diagonalizable with eigenvalues zero and one. Its trace and rank both equal the multiplicity of the eigenvalue one. A positive integer is nonzero in characteristic zero.

Remark 8.5

The trace criterion can fail in positive characteristic when the packet dimension is divisible by the characteristic. The rank and localization criteria remain valid.

A finite commutative Hecke algebra splits into canonical Artin local packets.
A finite commutative Hecke algebra splits into canonical Artin local packets.

View full-size figure (new tab)

An Artin packet is a simultaneous generalized eigenspace, not necessarily an ordinary eigenspace.
An Artin packet is a simultaneous generalized eigenspace, not necessarily an ordinary eigenspace.

View full-size figure (new tab)

Example 8.6: A nonsemisimple Hecke packet

Let

\[ {\mathbb T} = E'[u]/(u^2) \times E'[v]/(v^3) \]

and let \(V={\mathbb T}\) be the regular module. The two canonical central idempotents split

\[ V=V_u\oplus V_v \]

with dimensions two and three. Multiplication by \(u\) is nilpotent on \(V_u\), and multiplication by \(v\) is nilpotent on \(V_v\). Neither local factor is semisimple, but the two maximal Hecke packets are isolated exactly by the Artin idempotents.

Proof

The product decomposition gives the two central idempotents \((1,0)\) and \((0,1)\). Their ranges are the two local factors. The classes of \(u\) and \(v\) are nilpotent of orders two and three, respectively.

The next two theorems distinguish a fixed finite carrier from a sequence of changing carriers.

Theorem 8.7: Nested packets in one finite carrier

Let \(V\) be a finite dimensional vector space. Let \(e_1,e_2,\ldots\) be idempotents such that

\[ e_{m+1}e_m=e_me_{m+1}=e_{m+1} \]

for every \(m\). If \(e_mV\neq0\) for every \(m\), then the descending chain stabilizes and

\[ \boxed{ \bigcap_{m\geq1}e_mV \neq0. } \]

Proof of Theorem 8.7

The displayed relations give

\[ e_{m+1}V\subseteq e_mV. \]

The nonnegative dimensions form a nonincreasing sequence of positive integers, so they eventually become constant. Once two consecutive dimensions agree, the corresponding inclusion is equality. Thus the ranges stabilize at a nonzero subspace, which is their intersection.

Corollary 8.8: Finite witness of incompatibility

Under the hypotheses of Theorem 8.7, if

\[ \bigcap_{m\geq1}e_mV=0, \]

then \(e_mV=0\) for some finite \(m\).

Proof of Corollary 8.8

This is the contrapositive of Theorem 8.7.

Nested nonzero packets in one finite carrier stabilize after finitely many steps.
Nested nonzero packets in one finite carrier stabilize after finitely many steps.

View full-size figure (new tab)

Definition 8.9: Stable core

Let

\[ V_1\longleftarrow V_2\longleftarrow V_3\longleftarrow\cdots \]

be an inverse system of finite dimensional vector spaces. For \(n\geq m\), write \(f_{n,m}:V_n\to V_m\). The stable core at level \(m\) is

\[ Y_m = \bigcap_{n\geq m}{\operatorname{im}}(f_{n,m}). \]

Lemma 8.10: Surjectivity on stable cores

For every \(m\), the transition map \(V_{m+1}\to V_m\) induces a surjection

\[ \boxed{ Y_{m+1}\twoheadrightarrow Y_m. } \]

Proof of Lemma 8.10

Fix \(y\in Y_m\). For \(n\geq m+1\), let

\[ A_n = \left\{ z\in{\operatorname{im}}(f_{n,m+1}): f_{m+1,m}(z)=y \right\}. \]

Each \(A_n\) is a nonempty affine subspace of \(V_{m+1}\), and the sequence \((A_n)\) is descending. A descending sequence of affine subspaces in a finite dimensional vector space stabilizes whenever every term is nonempty. Choose an element in its stable value. It lies in \(Y_{m+1}\) and maps to \(y\).

Theorem 8.11: Nonzero stable core criterion

Let \((V_m)\) be an inverse system of finite dimensional vector spaces. If

\[ Y_m\neq0 \]

for every \(m\), then

\[ \boxed{ \varprojlim_mV_m \neq0. } \]

More precisely, the inverse limit contains a compatible sequence whose first coordinate is any prescribed nonzero element of \(Y_1\).

Proof of Theorem 8.11

Choose \(0\neq y_1\in Y_1\). By Lemma 8.10, choose recursively \(y_{m+1}\in Y_{m+1}\) mapping to \(y_m\). Every \(y_m\) is nonzero because it maps to the nonzero vector \(y_1\). The sequence \((y_m)\) is a nonzero element of the inverse limit.

Corollary 8.12: Surjective systems

If every transition map \(V_{m+1}\to V_m\) is surjective and every \(V_m\) is nonzero, then the inverse limit is nonzero.

Proof of Corollary 8.12

Surjectivity gives \(Y_m=V_m\). Apply Theorem 8.11.

Example 8.13: Tail escape in an infinite carrier

Let \(V=\ell^2({\mathbb N})\), and let \(P_m\) be the orthogonal projection onto the closed span of

\[ e_m,e_{m+1},e_{m+2},\ldots. \]

Every range \(P_mV\) is nonzero and the ranges are nested, but

\[ \boxed{ \bigcap_{m\geq1}P_mV=0. } \]

Thus the finite carrier hypothesis in Theorem 8.7 cannot be removed.

Proof

A vector in every tail range has its first \(m-1\) coordinates equal to zero for every \(m\). All of its coordinates vanish.

Changing finite carriers require a nonzero stable core, not only nonzero individual levels.
Changing finite carriers require a nonzero stable core, not only nonzero individual levels.

View full-size figure (new tab)

Nested nonzero ranges can escape to infinity in an infinite dimensional carrier.
Nested nonzero ranges can escape to infinity in an infinite dimensional carrier.

View full-size figure (new tab)

Source and conversion record

Source SHA-256: f5137c0fffed58acc1590165c77a75848e4d97671dca9caafb7c147e6aa1112a

Unsupported prose commands stop the build. The mathematical expression sequence is checked against the source, with manuscript macros expanded for display.

Conversion records

Notation in Paper XII

Look up a symbol without leaving this chapter. Select text before opening help to search for it.

Search this work’s notation.

    Symbols can have different meanings in different passages. Read each result in its own context.

    Open the full notation index