Harvard Logic Seminar
The Harvard Logic Seminar meets Tuesdays from 5:15pm to 6:15pm in Science Center 507. Email Will Boney (wboney@math.harvard.edu) if you are interested in speaking or being added to the mailing list.
The schedule for the Logic Colloquium can be found here.
Click the date to expand abstracts and names to go to the speaker's website (when available).
Next talk:

 Abstract: CW complexes are used extensively in algebraic topology as a suitable class of spaces to work with, but the product of two CW complexes need not be a CW complex, as shown by Dowker.
Whitehead and Milnor gave sufficient conditions on the two spaces for the product to be a CW complex, but until now the known characterisations of those pairs of CW complexes with product a CW complex relied on settheoretic assumptions about the whole universe, such as the Continuum Hypothesis. In this talk I will present a complete characterisation, valid without assuming any extra settheoretic axioms, of those pairs of CW complexes whose product is a CW complex.
Future talks:

 Abstract: The FefermanVaught Theorem in model theory gives a sort of upper bound on the complexity of definable subsets in a product structure. We show how this theorem implies that subsets whose boundary is dense (in the product topology where every factor structure has the discrete topology) are undefinable. We also show how the upper bound in the F.V. Theorem is the best possible, in the case of the language of rings where each factor structure is an integral domain. Finally, we apply these results to the ring $\prod_{p prime} F_p$, the product of all finite prime fields, and obtain a quantifier elimination result for the structure.

 Abstract: Spring Break is a common week (or more) long break in American universities, often observed in March or April. This talk details new insight into the ways that such breaks disrupt weekly scheduled events.


 Abstract: Two results are proved:
 Presburger Arithmetic is not finitely axiomatizable, and
 There is a consistent recursive extension of Robinson Arithmetic that decides every diophantine sentence.

Upcoming Logic Colloquia:
 January 25: Aki Kanamori
 March 8: Dima Sinapova
 March 22: Tony Martin
 April 12: Boris Zilber
 April 19: Alekos Kechris
Past talks:

 Abstract: We meditate on a particularly naive notion of a limit of a sequence of
theories: a union of conservative expansions. That is, we consider a
sequence of nested signatures $L_1 \subset L_2 \subset \ldots$, each
one a subsignature of the next, and a sequence of $L_i$theories
$T_i$, where each $T_i$ is precisely the set of $L_i$consequences of
$T_{i+1}$ (and hence is a subset of $T_{i+1}$).
It turns out that many modeltheoretic properties then pass from all
$T_i$ to their union $T$; these include consistency, completeness,
quantifier elimination, partial quantifier elimination such a
modelcompleteness, elimination of imaginaries, stable embeddedness of
some definable set, characterization of algebraic closure; stability,
simplicity, rosiness, dependence.
Our motivating example is the theory $T$ of fields with an action by
$(Q; +)$, seen as a limit of (theories of) fields with $(Z;
+)$actions.

February 6: Jesse Han (McMaster University), "Strong conceptual completeness for \omegacategorical theories"
 Abstract: Suppose we have some process to attach to every model of a firstorder theory some (permutation) representation of its automorphism group, compatible with elementary embeddings. How can we tell if this is really an imaginary sort of our theory?
In the '80s, Michael Makkai proved that the answer to our question is yes if and only if our given process is compatible with all ultraproducts and all "formal comparison maps" between them (generalizing e.g. the diagonal embedding into an ultrapower). This is known as /strong conceptual completeness/; formally, the statement is that the category Def(T) of definable sets can be reconstructed up to biinterpretability as the category of "ultrafunctors" Mod(T) \to Set.
\omegacategorical structures, having few definable sets, are exceptionally simple to understand, and in fact are determined up to biinterpretability by the action of their automorphism groups. Any general framework which reconstructs theories from their categories of models should therefore be considerably simplified for \omegacategorical theories.
Indeed, we show:
1. If T is ωcategorical, then X : Mod(T) → Set is definable, i.e. isomorphic to (M \mapsto ψ(M)) for some formula ψ ∈ T, if and only if X preserves ultraproducts and diagonal embeddings into ultrapowers. This means that all the preservation requirements for ultramorphisms, which a priori get unboundedly complicated, collapse to just diagonal embeddings when T is ωcategorical.
2. This definability criterion fails if we remove the ωcategoricity assumption. We construct examples of theories and nondefinable functors Mod(T) \to Set exhibiting this.

 Abstract: Szemer\‘{e}di’s regularity lemma for graphs says, roughly speaking, that any finite graph can be approximated by one that has a small structural “skeleton” overlaid with randomness. Malliaris and Shelah have shown that under the modeltheoretic tameness condition of stability, this approximation is especially wellbehaved: bounds on the size of the socalled “regularity partition” are significantly improved; there are no “irregular pairs” in the partition; and randomness is essentially eliminated. In this talk I will describe a generalisation of the MalliarisShelah result to stable finite structures in a finite relational language. This is joint work with Nate Ackerman and Cameron Freer.

 Abstract: Given an abelian group G and a subset A of G, one can construct a graph on G in which distinct elements x,y in G are connected if x+y is in A. If this graph is stable, then work of Malliaris and Shelah implies that it satisfies a strong form of Szemeredi's regularity lemma (and this has nothing to do with groups). A corollary of recent work of Terry and Wolf is that if G is a finite dimensional vector space over a prime field, then the regular partition of such a stable graph can be obtained using cosets of a subgroup. This motivates a statement of ``coset regularity" for subsets A of arbitrary finite groups G, such that "xy in A" is a stable binary relation. We prove this statement using local stable group theory and an ultraproduct construction. Joint with A. Pillay and C. Terry.

 Abstract: Thanksgiving break is a common tradition in American universities that cancels class for two or three days the week of Thanksgiving. In this cancelled talk, we explore how this tradition creeps earlier into the week and causes earlier meetings to be cancelled as well.

November 14: Linda Westrick (University of Connecticut), "Towards a notion of computable reducibility for discontinuous functions"
 Abstract: If X and Y are computably presented uncountable metric spaces, the collection of all functions from X to Y has cardinality too large to allow such functions to be represented as elements of Baire space. Nevertheless, we have some intuitive idea of what it should mean for one discontinuous function to compute another. I will discuss the problem of defining an appropriate notion of computable reducibility on this space. Joint work with Adam Day, Rod Downey and Takayuki Kihara.


 Abstract: The classification theory of elementary classes was started by
Michael Morley in the early sixties, when he proved that a countable
firstorder theory with a single model in some uncountable cardinal has
a single model in all uncountable cardinals. The proof of this result,
now called Morley's categoricity theorem, led to the development of
forking, a joint generalization of linear independence in vector spaces
and algebraic independence of fields, which is now a central pillar of
modern model theory.
In recent years, it has become apparent that the theory of forking can
also be developed in several nonelementary contexts. Prime among those
is the axiomatic framework of abstract elementary classes (AECs),
encompassing the class of models of any L_{infinity, omega}theory and
closely connected to the more general framework accessible categories. A
test question to judge progress in this direction is the forty year old
eventual categoricity conjecture of Shelah, which says that a version of
Morley's categoricity theorem should hold of any AEC. I will survey
recent developments, including the connections with category theory and
large cardinals as well as my resolution of the eventual categoricity
conjecture for classes of models axiomatized by a universal L_{infty,
omega}theory.

 Abstract: For a logic $\mathcal{L}$, interpolation holds in $\mathcal{L}$ iff for every implication $\phi \to \psi$, there is a sentence $\chi$ in their common language (in $\mathcal{L}$) such that $\phi \to \chi$ and $\chi \to \psi$. It is wellknown that $\mathbb{L}_{\omega, \omega}$ and $\mathbb{L}_{\omega_1, \omega}$ satisfy interpolation, but no other $\mathbb{L}_{\kappa, \lambda}$ does. We discuss a logic $\mathbb{L}^1_\kappa$ developed by Shelah that (for $\kappa = \beth_\kappa$) is intermediate between $\mathbb{L}_{\kappa, \omega}$ and $\mathbb{L}_{\kappa, \kappa}$ and satisfies interpolation.

 Abstract: The notion of a feedback query is a natural generalization of choosing for an oracle the set of indices of halting computations. Notice that, in that setting, the computations being run are different from the computations in the oracle: the former can query an oracle, whereas the latter cannot. A feedback computation is one that can query an oracle, which itself contains the halting information about all feedback computations. Although this is selfreferential, sense can be made of at least some such computations.
We'll discuss feedback around Turing computability. In one direction, we examine feedback Turing machines, and show that they provide exactly hyperarithmetic computability. In the other direction, Turing computability is itself feedback primitive recursion (at least, one version thereof). We'll also briefly consider notions for parallel computation, and for Borel maps on Cantor space.
Joint work with Nate Ackerman and Bob Lubarsky.

 Abstract: We will show there is a topological space for which presheaves are the same thing as trees. We will further show that there is a sheaf on this topological space which has an important relationship with Baire space. We will then use these connections to show how a definition by transfinite recursion can be thought of as an operation on sheaves, and how the welldefinedness of such a definition can be thought of as a property of the sheaf we are working on. This will then allow us to define a second order tree as a sheaf on a tree and to expand our notion of definition by transfinite recursion to all wellfounded second order trees (even those which are illfounded as normal trees). We will then mention how these techniques can be used to prove a variant of the SuslinKleene Separation theorem.

 Abstract: (Joint work with Tim McNicholl.) In this talk we will define a new uniform computable reducibility between computable Polish spaces. No specialized knowledge of computability theory is required.
Given computably presented Polish spaces X and Y, we say x in X is reducible to y in Y if there is a Pi^0_1 subset P of Y and a computable map f : P > X such that f(y)=x. For each space X one may consider the corresponding degree structure deg(X). For example, deg(2^omega) is (isomorphic to) the truthtable degrees, whereas both deg(omega^omega) and deg(reals) are proper extensions of deg(2^omega).
This new reducibility has many motivations. First, it is based on the notion of truthtable reducibility (which we will define). Truthtable reducibility on 2^omega is too restrictive of a setting for working within Baire space or the real numbers. For example, there are functions f in omega^omega not truthtable reducible to any X in 2^omega and sequences X in 2^omega such that X/3 is not truthtable reducible to X. Our reducibility gives the correct generalization of truthtable reducibility to these spaces. Second, this project mirrors Miller's nontrivial work extending Turing reducibility to computably presented Polish spaces. Last, our reducibility grew naturally out of work of the first author on computable arcs and the second author on Schnorr randomness. For example, we show that, for the vector space R^d, every Schnorr random is found in some computable arc.

 Abstract: It is consistent that the least strongly compact cardinal is the least supercompact cardinal, but it is also consistent that the least strongly compact cardinal is the least measurable cardinal. Which is it? The Ultrapower Axiom is an abstract comparison principle motivated by inner model theory that roughly states that any pair of ultrapowers can be ultrapowered to a common ultrapower. We give a characterization of supercompact cardinals in terms of the Mitchell order and use this to prove that the least strongly compact cardinal is supercompact assuming the Ultrapower Axiom and the GCH.

 Abstract: Compact cardinals get their names from a characterization in terms of the compactness of L_{\kappa, \kappa}. Measurable and supercompact cardinals also have characterizations in these terms, and Magidor has used secondorder logic to characterize supercompacts and extendible cardinals in this way. We will continue this line of modeltheoretic characterizations and discuss the characterizations of large cardinals in terms of compactness for omitting types focusing on three logics: L_{\kappa, \kappa}, secondorder, and sort logic.

 Abstract: The internal size of an object $M$ inside a given category
is, roughly, the least infinite cardinal $\lambda$ such that any
morphism from M into the colimit of a $\lambda^+$directed system
factors through one of the components of the system. In the category of
set, the internal size of an object is its cardinality. In the category
of vector spaces, the internal size is the dimension, and in the
category of metric spaces, the internal size is the least cardinality of
a dense subset.
We will discuss questions around internal sizes in the framework of
$\mu$abstract elementary classes ($\μ$AECs), which are, up to
equivalence of categories, the same as accessible categories with all
morphisms monomorphisms. We will in particular examine an example of
Shelaha certain class of sufficientlyclosed constructible models of
set theorywhich shows that the categoricity spectrum can behave very
differently depending on whether we look at categoricity in
cardinalities or in internal sizes. This is joint work with Michael
Lieberman and Jiří Rosický.

 Abstract: In this talk we will give background on $\mathcal{L}_{\infty, \omega}(L)$, categorical logic as well as Grothendieck toposes. We will then show how to make precise a version of Vaught's conjecture for a Grothendieck topos as well as discuss various analogs of Morley's theorem which hold in all Grothendeick toposes (under mild set theoretic assumptions). If we have time we will also discuss analogs of other theorems of $\mathcal{L}_{\infty, \omega}$ for Grothendieck toposes.
Back to Harvard Logic.
Back to my homepage.