An imaginary mathematical history.🧵
I should like to sketch an imaginary mathematical history, an alternative history by which the continuum hypothesis (CH) might have come naturally to be seen as a core axiom of set theory and one furthermore necessary for ordinary mathematics.
Let us imagine that in the early days of calculus, the theory was founded upon infinitesimals, but in a clear manner that identified the fundamental principles relating the real field ℝ to the hyperreals ℝ*, identifying the transfer principle and saturation.
And let us further imagine that in the heady developments at the end of the 19th and into the early 20th centuries providing the categorical characterizations of ℕ and ℝ, that a Zermelo-like figure provided also a categorical characterization of the hyperreal field ℝ*.
To be sure, this is not possible in ZFC, since there are different nonisomorphic ultrapowers ℝ^ℕ/U. The basic mathematical fact is that "the hyperreals" is not meaningful in plain ZFC—it does not determine a unique structure up to isomorphism.
But if we assume CH, however, then all such models are isomorphic, and we are able to provide a categorical characterization of the hyperreals in ZFC+CH. Namely, there is a unique saturated model of size continuum satisfying the transfer principle with the real field.
What I am asking you to imagine is that this key theorem, providing a fundamental characterization of the hyperreals, and thus of all the core concepts of calculus, was presented by some Zermelo-like figure at the dawn of the 20th century, as a theorem of set theory.
Of course, the figure would have to have provided the axioms of set theory supporting the proof, and these axioms would have included CH, much like the actual Zermelo had included the axiom of choice in his formalization in set theory of his proof of the well-order theorem.
Thus, CH would have been on the official axiom list, seen as necessary for proving the very welcome fact that the hyperreals are a definite mathematical structure. In this way, CH could have come to be seen as a core principle, necessary for the success of ordinary mathematics.
Later discoveries via forcing that without CH there can be multiple non-isomorphic hyperreal fields would have been seen as prima facie evidence for CH, since dropping it gives rise to the chaos of bizarre multiple interpretations for a fundamental mathematical idea.
@UnrollHelper unroll please
• • •
Missing some Tweet in this thread? You can try to
force a refresh
Three variations on Tarski.
Let me present three variations of Tarski's theorem on the non-definablity of truth. The first is proved in Gödelian style using Gödelian methods; the second is Russellian; and the last is purely Cantorian. But all express the nondefinability of truth.
First, the sentential variation, in Gödelian style. Truth is not definable, in the sense that there is no arithmetically expressible formula T(x) such that T('σ')⟷σ for every arithmetic sentence σ.
The proof amounts to the reasoning of the Liar paradox, since the fixed-point σ would assert its own falsity, if T did indeed have the stated truth-predicate property. Because it appeals to the fixed-point lemma and the coding of syntax, this proof is Gödelian.
The countable random graph.
I realized something today about the countable random graph. Namely, although this graph is highly homogeneous and has numerous automorphisms, it is nevertheless definably rigid, meaning that it has no definable automorphisms. Let me explain.
The countable random graph arises from a countable set of vertices by flipping independent fair coins to determine whether there is an edge between any two nodes.
Almost surely such a process fulfills what is called the finite pattern property: for any two finite disjoint sets of vertices A,B, there is a node x adjacent to every node of A and to no node of B.
Consider several structures on the natural numbers:
• Successor ⟨ℕ,0,S⟩
• Order ⟨ℕ,0,<⟩
• Addition ⟨ℕ,0,+⟩
What I claim is that each structure is definable in the next, but not conversely.
The positive claim is easy, for we can easily define the successor function from the order—the successor Sx is simply the least element that is larger than x. And similarly, we can define the order from +, since
x < y if and only if ∃z (z≠0 and x+z = y).
The converse nondefinability claims are more subtle.
To see that we cannot define < from successor, consider a nonstandard elementary extension
⟨ℕ,0,S⟩ ≺ ⟨ℕ*,0,S*⟩, which must consist of an ℕ-chain growing out of 0 together with a number of dissconnected ℤ-chains.
Truth and provability, a thread. Let us compare Tarski's theorem on the nondefinability of truth with Gödel's incompleteness theorem. Smullyan advanced the view that much of the fascination with Gödel's theorem should be better directed toward Tarski's theorem.
Both theorems begin with the profound arithmetization idea, Gödel's realization that arithmetic interprets essentially any finite combinatorial process, including arithmetic syntax itself. Every arithmetic assertion φ is thus represented in arithmetic by Gödel code 'φ', a number.
Gödel's theorem (the first incompleteness theorem) is the claim that no computability axiomatizable true arithmetic theory can prove all arithmetic truths. Every such theory will admit true but unprovable assertions.
I once saw an incredible lecture in Berkeley by a historian of science, one of the best talks I have ever seen. He started his talk sitting on the desk amongst a huge pile of physics textbooks from history, each in its day "the best textbook of its time."
He began by showing us the text written by a student of Isaac Newton, which explained Newtonian forces and the motion of planets. Nevertheless, it had various mistakes and misunderstandings. He set it aside and moved to the next book, also excellent, but also flawed in some way.
And then the next. The speaker proceeded systematically through the historical procession of physics texts, highlighting how each had improved upon its predecessors, but also explaining in specific detail ways in which each of them was flawed.
Foreshadowing — the algebra of orders. Here is an addition table for some simple finite orders. Did you know that you can add and multiply any two orders? A+B means a copy of A with a copy of B above. Stay tuned for multiplication...