Inner models from extended logics and the Delta-operation

This paper investigates the sensitivity of inner models $C(\mathcal{L})$, constructed by replacing first-order logic with an abstract logic $\mathcal{L}$ in Gödel's definition of $L$, to the $\Delta$-extension of $\mathcal{L}$, demonstrating that $C(\mathcal{L})$ can be consistently strictly smaller than $C(\Delta(\mathcal{L}))$ and that this strict inequality follows from the existence of $0^{\sharp}$ in specific cases.

The Gist

Imagine the universe of mathematics as a giant, infinite library called The Set Theory Library. Inside this library, there are books (sets) and rules for how those books can be written and organized.

For decades, mathematicians have tried to find the "perfect" way to build this library. The most famous blueprint was created by Kurt Gödel, a genius who designed a very strict, minimalist version of the library called $L$ (the Constructible Universe). In Gödel's library, you can only build a new book if you can describe it using the simplest, most basic language possible (First-Order Logic). It's a very orderly, safe place, but some mathematicians feel it's too small to contain all the interesting things that might exist in the real mathematical universe.

The New Idea: "Extended Logics"

In recent years, researchers (like the authors of this paper, Jouko Väänänen and Ur Ya'ar) started asking: What if we gave the librarians a bigger dictionary?

They proposed using Extended Logics. Imagine taking the basic language and adding special "super-words" or "super-quantifiers."

• Instead of just saying "There exists a number," you might say "There exist uncountably many numbers."
• Instead of just checking a list, you might have a tool that checks if a pattern holds for infinite groups at once.

When you use these super-words to build the library, you get a new, potentially larger version of the library, which they call $C(L)$.

The "Delta" Operation: The Safety Net

Now, here is the tricky part. When you add these super-words, you might accidentally create a situation where you can describe a group of books in two different ways that seem to contradict each other, or where the description is so complex it's hard to pin down.

To fix this, mathematicians invented the Delta ($\Delta$) Operation.
Think of the Delta operation as a Safety Net or a Quality Control Filter.

• If you can describe a group of books using the super-words in a way that is "positive" (it exists), and you can also describe the "not-them" group in a way that is "positive," then the Delta operation says, "Okay, this group is solid. Let's officially add it to the library."
• It's a way of saying: "If we can define something and its opposite clearly, then that thing is definitely real and belongs in our model."

Usually, mathematicians thought this Safety Net was a mild addition. They believed that if you built a library with the super-words ($L$) and then applied the Safety Net ($\Delta(L)$), you wouldn't get a significantly different library. It was thought to be like adding a few extra shelves to a room that was already full.

The Big Discovery: The Safety Net Changes Everything

This paper is a surprise party for mathematicians. The authors discovered that the Safety Net actually changes the library completely.

They found specific examples where:

1. Library A (built with the super-words) is small and simple.
2. Library B (built with the super-words plus the Safety Net) is huge and contains secrets that Library A doesn't even know exist.

The Analogy of the Detective:
Imagine a detective (the Logic) trying to solve a crime.

• Logic A can ask: "Was the butler in the room?"
• Logic B (with the Delta operation) can ask: "Can we prove the butler was in the room, AND can we prove the butler was not in the room? If we can't prove either, maybe the butler is a ghost?"

The authors show that sometimes, the "Ghost" (the Delta extension) reveals a whole new layer of reality. In one specific case, they showed that if a mysterious mathematical object called $0^\sharp$ (Zero Sharp) exists, then the Safety Net version of the library ($C(\Delta(L))$) contains this object, while the original version ($C(L)$) does not.

It's like discovering that your house has a hidden basement. You thought you had a two-story house (Logic A), but once you applied the "Delta" inspection, you realized there was a whole third floor (Logic B) that you couldn't see before.

Why Does This Matter?

This is a big deal because it shows that how we define our rules matters more than we thought.

• Two languages might look almost identical from the outside (like two dialects of English).
• But if you use them to build a mathematical universe, one might build a tiny village, and the other might build a sprawling metropolis.

The paper proves that the "Delta" extension, which was thought to be a minor tweak, is actually a powerful engine that can generate entirely new mathematical worlds. It forces us to be very careful about exactly how we define our logical rules, because a tiny change in the definition can lead to a massive change in the reality we construct.

Summary in a Nutshell

• The Goal: Build a perfect mathematical universe.
• The Tool: Use "super-languages" (Extended Logics) to define what exists.
• The Twist: Adding a "Safety Net" (the Delta operation) to these languages doesn't just tidy things up; it fundamentally changes the size and content of the universe you build.
• The Result: Sometimes, the "Safe" version of the universe is much bigger and more complex than the "Raw" version, containing deep secrets (like $0^\sharp$) that the raw version misses.

This paper tells us that in the world of math, precision is everything. A slight shift in how we define "truth" can open the door to entirely new dimensions of reality.

Technical Summary

1. Problem Statement

The paper investigates the sensitivity of inner models of set theory constructed from abstract logics to the specific definition of the logic used. Specifically, it examines the relationship between an abstract logic $L$ and its $\Delta$-extension, denoted $\Delta(L)$.

• Context: In the theory of abstract logics, $\Delta(L)$ is the minimal extension of $L$ such that every model class definable by both a $\Sigma(L)$ and a $\Pi(L)$ formula (i.e., a $\Delta(L)$-definable class) becomes explicitly definable in $\Delta(L)$. From a model-theoretic perspective, $\Delta(L)$ is often considered a "mild" extension that preserves key properties like compactness and Löwenheim numbers.
• The Core Question: Does the "mildness" of the $\Delta$-operation extend to the set-theoretic strength of the resulting inner models? Specifically, is the inner model $C(L)$ (the class of sets constructible using logic $L$) always equal to $C(\Delta(L))$?
• Hypothesis: While $L$ and $\Delta(L)$ may be model-theoretically similar, the authors hypothesize that their set-theoretic power (the size and complexity of $C(L)$ vs. $C(\Delta(L))$) can differ significantly, particularly under large cardinal assumptions or specific forcing conditions.

2. Methodology

The authors employ a combination of abstract model theory, inner model theory, and forcing techniques to construct counterexamples.

• Definitions:
  • $C(L)$: Defined analogously to Gödel's $L$, but replacing the first-order definability operator with the $L$-definability operator ($Def_L$).
  • $\Delta(L)$: Defined via the $\Delta$-operation, which captures implicit definability. The syntax of $\Delta(L)$ can be non-absolute (dependent on the set-theoretic universe), which is a crucial feature exploited in the proofs.
• Comparative Analysis: The paper compares $C(L)$ and $C(\Delta(L))$ across various logics:
  • Standard Logics: First-order logic with the quantifier $Q_0$ ("there are $\ge \aleph_0$"), Second-order logic ($L_2$), and Henkin quantifiers ($H$).
  • Extended Logics: Logics involving the quantifier $Q_1$ ("there are $\ge \aleph_1$") and the well-ordering quantifier $W$.
• Forcing Constructions:
  • The authors use finite support products of forcing notions (specifically involving Suslin trees and specializing functions) to manipulate the properties of trees in the universe.
  • They utilize Martin's Axiom (MA) to ensure that all Aronszajn trees in the generic extension are special.
  • They construct models where specific sets (like a Cohen real $a$) are definable in $C(\Delta(L))$ but not in $C(L)$, thereby proving the models are distinct.
• Large Cardinal Assumptions: The paper utilizes the existence of $0^\sharp$ (zero sharp) to demonstrate that $0^\sharp$ itself can be captured in $C(\Delta(L))$ for specific logics, a feat impossible in $C(L)$.

3. Key Contributions and Results

The paper establishes that $C(L)$ and $C(\Delta(L))$ can be distinct, even when $L$ and $\Delta(L)$ are model-theoretically close. The results are categorized by the logic $L$:

A. Cases Where $C(L) = C(\Delta(L))$

• $L(Q_0)$: For the logic with the quantifier "there are $\ge \aleph_0$," it is provable in ZF that $C(L(Q_0)) = C(\Delta(L(Q_0))) = L$ (Gödel's constructible universe), despite $L(Q_0) \neq \Delta(L(Q_0))$.
• $L_2$ (Second-order logic): Assuming AC, $C(L_2) = C(\Delta(L_2)) = \text{HOD}$ (Hereditarily Ordinal Definable sets).

B. Cases Where $C(L) \neq C(\Delta(L))$ (Consistency Results)

• Henkin Quantifier ($H$): The authors prove that it is consistent with ZF that $C(L(H)) = \text{HOD}_1$ (a model slightly larger than $L$ but smaller than HOD), while $C(\Delta(L(H))) = \text{HOD}$. Thus, $C(L(H)) \neq C(\Delta(L(H)))$.
• Quantifier $Q_1$ ("$\ge \aleph_1$"):
  • Theorem 16: It is consistent with ZF that $C(L(Q_1)) = L$ while $C(\Delta(L(Q_1))) \neq L$.
  • Mechanism: The proof relies on the fact that the property of a tree being "non-Aronszajn" (having an uncountable branch) is $\Delta(L(Q_1))$-definable under Martin's Axiom, allowing the construction of a Cohen real $a$ inside $C(\Delta(L(Q_1)))$ that is not in $L$.
• Symbiotic Logics ($L(Q_1, W)$):
  • The paper addresses the concept of symbiosis (a tight connection between a logic and set-theoretic predicates). Even for $L(Q_1, W)$, which is symbiotic with the predicate $R_{\aleph_1}$, it is consistent that $C(L) = L \neq C(\Delta(L))$.

C. The Role of Large Cardinals ($0^\sharp$)

• Theorem 24: If $0^\sharp$ exists, then $0^\sharp \in C(\Delta(Q_1, W))$.
• Significance: Since $0^\sharp \notin L$ (and thus $0^\sharp \notin C(L(Q_1, W))$ if that model is $L$), this provides a strong separation. The $\Delta$-operation allows the logic to "see" the existence of indiscernibles for $L$, effectively capturing $0^\sharp$ within the inner model, whereas the base logic $L$ cannot.

D. The $\Delta^1_1$ Operation

• The authors distinguish between the full $\Delta$-operation (allowing new sorts) and the $\Delta^1_1$-operation (allowing only new predicates).
• They show that $C(\Delta^1_1(L(H))) = C(L(H))$, but $C(\Delta(L(H))) \neq C(L(H))$. This highlights that the ability to add new sorts in the $\Delta$-extension is crucial for increasing the set-theoretic power of the inner model in certain cases.

4. Significance

1. Decoupling Model Theory and Set Theory: The paper demonstrates that two logics can be indistinguishable in terms of classical model-theoretic properties (compactness, Löwenheim numbers, interpolation) yet generate vastly different inner models of set theory. This challenges the intuition that "mild" logical extensions yield "mild" set-theoretic consequences.
2. Sensitivity of Inner Models: It reveals that the construction of inner models $C(L)$ is highly sensitive to the exact syntactic definition of the logic, particularly regarding implicit definability ($\Delta$-definability).
3. New Inner Models: The results contribute to the emerging spectrum of inner models between $L$ and HOD, showing that $\Delta$-extensions can generate models with large cardinal properties (like containing $0^\sharp$) that are inaccessible via the base logic.
4. Open Questions on AC: The paper raises a critical open question regarding the Axiom of Choice (AC). Standard proofs that AC holds in $C(L)$ rely on the logic having a recursive syntax. Since $\Delta(L)$ syntax is often non-absolute and non-recursive, it is currently unknown whether AC holds in models like $C(\Delta(Q_1))$.

5. Conclusion

Väänänen and Ya'ar successfully demonstrate that the $\Delta$-operation, while preserving many model-theoretic features of a logic, can drastically alter the set-theoretic landscape of the associated inner model. Through the use of forcing, tree properties, and large cardinal assumptions, they provide concrete examples where $C(L) \subsetneq C(\Delta(L))$, proving that the "mildness" of the $\Delta$-extension does not extend to the constructible universe it generates.