Wiki
Wiki

Research notes on every problem, and a library of the papers behind them. Built from the open erdos repository.

Updated

../


Source. E. Glazer, Erdős Problem 501 after adding ω2\omega_2 random reals, draft rev10, Theorem 5.1 (forcing interface), physical pp. 6--7, in the eight-page PDF held by its library source card, Glazer (2026). It uses Definition 3.1 from the Theorem 3.2 page, Proposition 4.4 and Lemma 4.5.

Standing. This is an author-recorded reconstruction. It is not an independent review and changes no status and assigns no tier. The conventions (R1)--(R5) of the Lemma 4.1 page are assumed; outer regularity of Lebesgue outer measure is used as a definition. The instance of the map ρ\rho under Definitions is a compilation fill.

Definitions

κ=ω2\kappa=\omega_2, Θ=κ×ω\Theta=\kappa\times\omega, Dα={α}×ωD_\alpha=\{\alpha\}\times\omega and B=B(Θ)\mathbb B=\mathbb B(\Theta), as on the Proposition 4.4 page; O\mathcal O, U(c)U(c), Im=[m,m+1)I_m=[m,m+1) and profile certificates as on the Theorem 3.2 page. Fix a Borel measure-preserving map ρ ⁣:2ω→[0,1)\rho\colon2^\omega\to[0,1) (fair-coin measure to Lebesgue measure) with null point fibers. One instance: ρ(r)=∑nr(n)2−n−1\rho(r)=\sum_nr(n)2^{-n-1}, redefined as 00 on the null set of sequences that are eventually 11; it pushes the fair-coin measure to Lebesgue measure on [0,1)[0,1), and each fiber is countable, hence null. Through the enumeration n↦(α,n)n\mapsto(\alpha,n) of DαD_\alpha, 2Dα2^{D_\alpha} is identified with 2ω2^\omega, and through the enumeration ⟨dn⟩\langle d_n\rangle of DD from Proposition 4.4, so is 2D2^D.

A name A˙=(A˙y)y∈R\dot{\mathcal A}=(\dot A_y)_{y\in\mathbb R} is a name for a function from the reals of the extension to subsets of those reals; for a name x˙\dot x for a real, A˙x˙\dot A_{\dot x} names its value at x˙\dot x.

Statement

ZFC+CH⊢ Bω2⊩∀A [(∀y∈R λ∗(Ay)<1)⟶Prof(A)]\mathrm{ZFC}+\mathrm{CH}\vdash\ \mathbb B_{\omega_2}\Vdash \forall\mathcal A\,\bigl[(\forall y\in\mathbb R\ \lambda^*(A_y)<1) \longrightarrow\mathrm{Prof}(\mathcal A)\bigr]

(the source's (5.1)).

Proof

Work in a ground model M⊨ZFC+CHM\models\mathrm{ZFC}+\mathrm{CH} and let p∈Bp\in\mathbb B force

A˙=(A˙y)y∈R  satisfies  λ∗(A˙y)<1  for every y∈R\dot{\mathcal A}=(\dot A_y)_{y\in\mathbb R}\ \text{ satisfies }\ \lambda^*(\dot A_y)<1\ \text{ for every }y\in\mathbb R

(the source's (5.2)). We show that pp forces Prof(A˙)\mathrm{Prof}(\dot{\mathcal A}).

Random points. For α<κ\alpha<\kappa let r˙α=G˙↾Dα∈2ω\dot r_\alpha=\dot G\restriction D_\alpha\in2^\omega, the random real read from the block DαD_\alpha, and for m∈Zm\in\mathbb Z put

x˙α,m=m+ρ(r˙α)∈Im\dot x_{\alpha,m}=m+\rho(\dot r_\alpha)\in I_m

(the source's (5.3)).

Names for envelopes. Fix α<κ\alpha<\kappa and m∈Zm\in\mathbb Z. In any extension by a generic containing pp, λ∗(Axα,m)<1\lambda^*(A_{x_{\alpha,m}})<1, so by the definition of outer measure there is an open set of measure below one containing Axα,mA_{x_{\alpha,m}}, and every open set has a code. By the maximum principle (R3) there is a name c˙α,m\dot c_{\alpha,m} with

p⊩A˙x˙α,m⊆U(c˙α,m)  and  λ(U(c˙α,m))<1p\Vdash\dot A_{\dot x_{\alpha,m}}\subseteq U(\dot c_{\alpha,m}) \ \text{ and }\ \lambda(U(\dot c_{\alpha,m}))<1

(the source's (5.4)). Mix c˙α,m\dot c_{\alpha,m} with a fixed default code below the complement of pp, so that the top condition forces c˙α,m∈O\dot c_{\alpha,m}\in\mathcal O; this makes it a name for an element of the standard Borel space O\mathcal O in the sense of (R4). Bundle the countable sequence into

w˙α=⟨c˙α,m:m∈Z⟩∈OZ,\dot w_\alpha=\langle\dot c_{\alpha,m}:m\in\mathbb Z\rangle \in\mathcal O^{\mathbb Z},

a name for an element of the standard Borel space OZ\mathcal O^{\mathbb Z}.

Homogeneous reading. Apply Proposition 4.4 (in MM, which satisfies CH) to pp, X=OZX=\mathcal O^{\mathbb Z} and (w˙α)α<κ(\dot w_\alpha)_{\alpha<\kappa}. Obtain J⊆κJ\subseteq\kappa of size κ\kappa, the root RR, the petals (Pα)α∈J(P_\alpha)_{\alpha\in J} with Dα⊆PαD_\alpha\subseteq P_\alpha, the pair D⊆PD\subseteq P with bijections πα ⁣:P→Pα\pi_\alpha\colon P\to P_\alpha sending dnd_n to (α,n)(\alpha,n), and a Borel map

F=⟨Fm:m∈Z⟩ ⁣:2R×2P→OZF=\langle F_m:m\in\mathbb Z\rangle\colon2^R\times2^P\to\mathcal O^{\mathbb Z}

(the source's (5.5)) with ⊩w˙α=F(G˙↾R,(G˙↾Pα)∘πα)\Vdash\dot w_\alpha=F(\dot G\restriction R,(\dot G\restriction P_\alpha)\circ\pi_\alpha) for α∈J\alpha\in J. The set PP is countably infinite, since D⊆PD\subseteq P is in bijection with DαD_\alpha.

The certificate. Let G∋pG\ni p be B\mathbb B-generic over MM and work in M[G]M[G]. Set g=G↾R∈2Rg=G\restriction R\in2^R and

zα=(G↾Pα)∘πα∈Ω:=2P(α∈J)z_\alpha=(G\restriction P_\alpha)\circ\pi_\alpha\in\Omega:=2^P \qquad(\alpha\in J)

(the source's (5.6)). Let ν\nu be the fair-coin product measure on Ω\Omega, a standard Borel probability space, and put Z={zα:α∈J}Z=\{z_\alpha:\alpha\in J\}. Lemma 4.5, applied in MM to the petals (Pα)α∈J(P_\alpha)_{\alpha\in J} (uncountable in MM) and Γ=Θ∖⋃α∈JPα\Gamma=\Theta\setminus\bigcup_{\alpha\in J}P_\alpha, gives ν∗(Z)=1\nu^*(Z)=1 in M[G]M[G] (the source's (5.7)). This is (P1).

For z∈Ωz\in\Omega and m∈Zm\in\mathbb Z define the raw code cm0(z)=Fm(g,z)c^0_m(z)=F_m(g,z), a Borel function of zz (a section of the Borel map FmF_m at the fixed gg), and the truncated code

cm(z)={cm0(z),λ(U(cm0(z)))<1,c∅,λ(U(cm0(z)))≥1c_m(z)=\begin{cases} c^0_m(z),&\lambda(U(c^0_m(z)))<1,\\ c_\varnothing,&\lambda(U(c^0_m(z)))\ge1 \end{cases}

(the source's (5.8)), where U(c∅)=∅U(c_\varnothing)=\varnothing. Since c↦λ(U(c))c\mapsto\lambda(U(c)) is Borel, the case split is over a Borel set and cm ⁣:Ω→Oc_m\colon\Omega\to\mathcal O is Borel; and λ(U(cm(z)))<1\lambda(U(c_m(z)))<1 for every z∈Ωz\in\Omega (the source's (5.9)). This is (P3). The truncation is what makes (P3) hold on all of Ω\Omega rather than only on ZZ.

Define

xm(z)=m+ρ(z↾D)x_m(z)=m+\rho(z\restriction D)

(the source's (5.10)). It is Borel with values in ImI_m. The restriction z↦z↾Dz\mapsto z\restriction D pushes ν\nu to the fair-coin measure on 2D2^D, and ρ\rho pushes that to Lebesgue measure on [0,1)[0,1), so for Borel B⊆RB\subseteq\mathbb R, ν(xm−1(B))=λ((B−m)∩[0,1))=λ(B∩Im)\nu(x_m^{-1}(B))=\lambda((B-m)\cap[0,1))=\lambda(B\cap I_m). This is (P2).

Finally let z=zα∈Zz=z_\alpha\in Z with α∈J\alpha\in J. In M[G]M[G], since p∈Gp\in G and w˙α\dot w_\alpha is read by FF,

c˙α,mG=Fm(g,zα)=cm0(zα)(m∈Z).\dot c_{\alpha,m}^G=F_m(g,z_\alpha)=c^0_m(z_\alpha)\qquad(m\in\mathbb Z).

Moreover xm(zα)=x˙α,mGx_m(z_\alpha)=\dot x_{\alpha,m}^G: by the choice of πα\pi_\alpha, (zα↾D)(dn)=G(α,n)=r˙αG(n)(z_\alpha\restriction D)(d_n)=G(\alpha,n)=\dot r_\alpha^G(n) for every nn, so zα↾Dz_\alpha\restriction D is r˙αG\dot r_\alpha^G under the fixed identifications, and xm(zα)=m+ρ(r˙αG)x_m(z_\alpha)=m+\rho(\dot r_\alpha^G). Hence, by the forced statement (5.4) with p∈Gp\in G,

λ(U(cm0(zα)))<1andAxm(zα)⊆U(cm0(zα)).\lambda(U(c^0_m(z_\alpha)))<1\quad\text{and}\quad A_{x_m(z_\alpha)}\subseteq U(c^0_m(z_\alpha)).

The first inequality says that the truncation leaves the code at zαz_\alpha unchanged, cm(zα)=cm0(zα)c_m(z_\alpha)=c^0_m(z_\alpha), so

Axm(zα)⊆U(cm(zα))A_{x_m(z_\alpha)}\subseteq U(c_m(z_\alpha))

(the source's (5.11)). This is (P4).

Thus (Ω,ν)(\Omega,\nu), ZZ, ⟨xm,cm:m∈Z⟩\langle x_m,c_m:m\in\mathbb Z\rangle is a profile certificate for A=A˙G\mathcal A=\dot{\mathcal A}^G in M[G]M[G].

Closing the quantifiers. Every generic G∋pG\ni p satisfies Prof(A˙G)\mathrm{Prof}(\dot{\mathcal A}^G), so p⊩Prof(A˙)p\Vdash\mathrm{Prof}(\dot{\mathcal A}) by the forcing theorem. If some condition forced (∀y λ∗(A˙y)<1)∧¬Prof(A˙)(\forall y\ \lambda^*(\dot A_y)<1)\wedge\neg\mathrm{Prof}(\dot{\mathcal A}), the argument applied to that condition would contradict it; hence the top condition forces the implication for A˙\dot{\mathcal A}. Every family in M[G]M[G] has a name, so the universal statement (5.1) follows.

Boundary. CH is used only through Proposition 4.4. Lemma 4.5 and Theorem 3.2 are ZFC theorems. The assembly into Theorem 1.1 is on the Theorem 1.1 page.