Euler-admissible pairs and the Ext-Euler characteristic #
Let C be a k-linear abelian category with Ext groups. The Ext-Euler characteristic of a
pair of objects is the alternating sum
χ(X, Y) = ∑ n, (-1)ⁿ dim_k Extⁿ(X, Y).
This is a number only under two genuinely separate finiteness hypotheses: every Extⁿ(X, Y) must
be a finite-dimensional k-vector space, and Extⁿ(X, Y) must vanish for all large n. Neither
implies the other — over the dual numbers k[ε]/(ε²) the simple module S has
Extⁿ(S, S) ≅ k for every n, so the first holds and the second fails, while an
infinite-dimensional vector space is Ext-bounded but not Ext-finite against itself — and
a definition that totalizes the sum would return a junk value in exactly the first
situation. This file therefore keeps
the two conditions apart, as IsExtFinite and IsExtBounded, packages them as
IsEulerAdmissible, and defines the Euler characteristic only from a witness of both.
The value is defined by truncating the sum to the degrees below an explicit bound
(truncatedExtEuler); the whole content of extEuler_eq is that every bound beyond which the
Ext groups vanish gives the same answer.
Main definitions #
TauCeti.IsExtFinite: everyExtⁿ(X, Y)is a finite-dimensionalk-vector space.TauCeti.IsExtBoundedByandTauCeti.IsExtBounded:Extⁿ(X, Y)vanishes fornat least a given bound, resp. for all largen.TauCeti.IsEulerAdmissible: the conjunction of the two, the hypothesis under which the Ext-Euler characteristic of the pair(X, Y)exists.TauCeti.IsExtBoundedOnandTauCeti.IsEulerAdmissibleOn: respectively, uniform boundedness and pointwise Euler-admissibility for a pair of object properties; the latter is the hypothesis Layer 5's descent to Grothendieck groups will consume.TauCeti.truncatedExtEulerandTauCeti.extEuler: the truncated alternating sum, and the Ext-Euler characteristic of an Euler-admissible pair.
Main results #
TauCeti.extEuler_eq: the Ext-Euler characteristic is the alternating sum truncated at any bound beyond which theExtgroups vanish.TauCeti.IsEulerAdmissible.congrandTauCeti.extEuler_congr: degreewise linear equivalences ofExtpreserve the hypothesis and the value.TauCeti.IsEulerAdmissible.of_isoandTauCeti.extEuler_of_iso: isomorphism invariance of the hypothesis and of the value.TauCeti.IsEulerAdmissible.of_shortExact₂andTauCeti.IsEulerAdmissible.of_shortExact₂': Euler-admissibility is closed under extensions in either variable, andTauCeti.IsEulerAdmissible.biprod,TauCeti.IsEulerAdmissible.biprod'under binary direct sums; the empty direct sum is covered byTauCeti.IsEulerAdmissible.of_isZero_leftandTauCeti.IsEulerAdmissible.of_isZero_right.TauCeti.IsExtFinite.of_shortExact₃':Ext-finiteness also passes from the subobject and the middle object of a short exact sequence to its quotient, in the first variable.TauCeti.IsExtFinite.finiteDimensional_hom: Hom-finiteness is the degree-zero consequence ofExt-finiteness, and is kept as a separate predicate.TauCeti.extEuler_projective: projective evaluation,χ(P, Y) = dim_k Hom(P, Y).
The remaining Layer 5 targets — the long exact sequences cut off by the vanishing bound, the
resulting additivity of χ in both variables, and the descent of χ to a biadditive pairing on
Grothendieck groups — are not proved here; this file supplies the finiteness interface they are
stated over.
References #
- Charles A. Weibel, An Introduction to Homological Algebra, Cambridge Studies in Advanced
Mathematics 38, Cambridge University Press (1994), Sections 2.4--2.7 and Chapter 4, for
Ext, its long exact sequences, and projective dimension. - Ibrahim Assem, Daniel Simson, and Andrzej Skowroński, Elements of the Representation Theory of Associative Algebras, Volume 1, LMS Student Texts 65, Cambridge University Press (2006), Chapter III, Section 3, Proposition 3.13, for the Euler form of an algebra of finite global dimension.
- The Tau Ceti Grothendieck groups, Cartan maps, and Euler forms roadmap, Layer 5, whose "Ext-finite pairs" bullet and the first half of its "Ext-Euler value" bullet are the targets proved here.
The two finiteness conditions #
Every Extⁿ(X, Y) is a finite-dimensional k-vector space. This is one of the two
independent halves of TauCeti.IsEulerAdmissible; on its own it does not make the alternating
sum of the dimensions a finite sum.
- finiteDimensional (n : ℕ) : FiniteDimensional k (CategoryTheory.Abelian.Ext X Y n)
Each
Extgroup of the pair is finite-dimensional.
Instances For
Extⁿ(X, Y) vanishes in every degree n ≥ N.
The
Extgroups of the pair vanish from degreeNon.
Instances For
Extⁿ(X, Y) vanishes for all large n. This is the other half of
TauCeti.IsEulerAdmissible, and the one that makes the Ext-Euler characteristic a finite sum.
- exists_bound : ∃ (N : ℕ), IsExtBoundedBy X Y N
Some degree bounds the
Ext-support of the pair.
Instances For
A pair (X, Y) is Euler-admissible when all of its Ext groups are finite-dimensional
and all but finitely many of them vanish. This is exactly the hypothesis under which the
alternating sum ∑ n, (-1)ⁿ dim_k Extⁿ(X, Y) is a well-defined integer.
- isExtFinite : IsExtFinite k X Y
All
Extgroups of the pair are finite-dimensional. - isExtBounded : IsExtBounded X Y
All but finitely many
Extgroups of the pair vanish.
Instances For
Elementary consequences and monotonicity #
A vanishing bound may always be raised.
An explicit vanishing bound witnesses eventual Ext-vanishing.
Hom-finiteness is the degree-zero part of Ext-finiteness, and is kept as a separate, weaker
predicate: it says nothing about the higher Ext groups.
Versions for a pair of object properties #
A uniform Ext-vanishing bound for a pair of object properties: Extⁿ(X, Y) = 0 in every
degree n ≥ N, for all X satisfying P and all Y satisfying Q. For a category of modules
over a k-algebra of finite global dimension this is the bound supplied by that dimension, and it
is a strictly stronger hypothesis than pointwise Ext-boundedness.
The bound
Nworks for every pair of objects drawn fromPandQ.
Instances For
Every pair of objects drawn from P and Q is Euler-admissible, with a bound that may depend
on the pair. This is the hypothesis under which the Ext-Euler characteristic descends to a pairing
between the Grothendieck groups of the two subcategories. A shared bound is the separate, stronger
predicate TauCeti.IsExtBoundedOn.
Each pair of objects drawn from
PandQis Euler-admissible.
Instances For
A uniform bound is in particular a pointwise one.
A uniform vanishing bound passes to smaller object properties and may always be raised, in particular it restricts to full subcategories of the ones considered.
Euler-admissibility on a pair of object properties passes to smaller properties, in particular to full subcategories of the ones considered.
A uniform bound and pointwise Ext-finiteness give Euler-admissibility on the pair.
The Ext-Euler characteristic #
The raw alternating sum ∑_{n < N} (-1)ⁿ * Module.finrank k Extⁿ(X, Y) over the cohomological
degrees below N. This definition takes no admissibility witness: a summand is the dimension of
the corresponding Ext group only where that group is finite-dimensional, Module.finrank
returning its 0 fallback elsewhere. Truncating at an explicit bound is what keeps the sum finite;
TauCeti.extEuler is the version that supplies both a TauCeti.IsExtFinite witness, making every
summand a genuine dimension, and a vanishing bound for N.
Equations
- TauCeti.truncatedExtEuler k X Y N = ∑ n ∈ Finset.range N, (-1) ^ n * ↑(Module.finrank k (CategoryTheory.Abelian.Ext X Y n))
Instances For
The empty truncation of the alternating sum is zero.
Raising the truncation bound by one adds the signed dimension of the next Ext group.
Raising the truncation bound past a degree from which the Ext groups vanish does not change
the alternating sum.
The Ext-Euler characteristic χ(X, Y) = ∑ n, (-1)ⁿ dim_k Extⁿ(X, Y) of an
Euler-admissible pair. The choice of vanishing bound made here is removed at once by
TauCeti.extEuler_eq, so no result depends on it.
Equations
- TauCeti.extEuler k h = TauCeti.truncatedExtEuler k X Y ⋯.choose
Instances For
Every degree from which the Ext groups of the pair vanish computes its Ext-Euler
characteristic.
The Ext-Euler characteristic of a pair with no Ext at all is zero.
Invariance under linear equivalences and isomorphisms #
Ext-finiteness transports along degreewise linear equivalences of Ext groups, including
between pairs in different categories.
A vanishing bound transports along bijections of the Ext groups from that degree on,
including between pairs in different categories.
Eventual Ext-vanishing transports along bijections of Ext groups in all large degrees,
including between pairs in different categories.
Euler-admissibility transports along degreewise linear equivalences of Ext groups,
including between pairs in different categories.
Degreewise linear equivalences of Ext groups preserve the Ext-Euler characteristic,
including when the pairs lie in different categories.
Ext-finiteness only depends on the isomorphism classes of the two objects.
A vanishing bound transports along isomorphisms of the two objects.
Eventual Ext-vanishing only depends on the isomorphism classes of the two objects.
Euler-admissibility only depends on the isomorphism classes of the two objects.
The Ext-Euler characteristic only depends on the isomorphism classes of the two objects.
Closure under extensions and finite direct sums #
Extension closure in the second variable for Ext-finiteness.
Extension closure in the first variable for Ext-finiteness.
Ext-finiteness passes to the quotient term of a short exact sequence, along the first
variable: the long exact sequence exhibits Extⁿ⁺¹(S.X₃, Y) between Extⁿ(S.X₁, Y) and
Extⁿ⁺¹(S.X₂, Y), and its degree-zero group is a subspace of Hom(S.X₂, Y).
Extension closure in the second variable for eventual Ext-vanishing: the middle term of
a short exact sequence inherits the larger of the two outer bounds.
Extension closure in the first variable for eventual Ext-vanishing.
Euler-admissibility is closed under extensions in the second variable.
Euler-admissibility is closed under extensions in the first variable.
Euler-admissibility is closed under binary direct sums in the second variable, because
Y₁ ⟶ Y₁ ⊞ Y₂ ⟶ Y₂ is short exact.
Euler-admissibility is closed under binary direct sums in the first variable.
A zero object is Euler-admissible against every object: all of its Ext groups vanish. This
is the empty case of TauCeti.IsEulerAdmissible.biprod', the direct sum of no objects being a
zero object.
Every object is Euler-admissible against a zero object. This is the empty case of
TauCeti.IsEulerAdmissible.biprod.
Projective evaluation #
All Ext groups of positive degree out of a projective object vanish, so a pair with
projective first entry is Ext-bounded by 1.
If every object satisfying P is projective, then 1 is a uniform Ext-vanishing bound for
P against an arbitrary Q. This is the degenerate case of a uniform global-dimension bound, and
the one Layer 4's Cartan comparison uses on the subcategory of projectives.
A Hom-finite pair with projective first entry is Ext-finite: apart from the degree-zero
group, which is the Hom space, all of its Ext groups vanish.
A Hom-finite pair with projective first entry is Euler-admissible.
Projective evaluation: the Ext-Euler characteristic of a pair with projective first entry
is the dimension of its Hom space, χ(P, Y) = dim_k Hom(P, Y).
Hom-finite projectives are Euler-admissible against every object satisfying Q: this is the
concrete source of Euler-admissibility on the projective side, and it is what makes
TauCeti.IsEulerAdmissibleOn a nonempty hypothesis.