Q
stringlengths
4
3.96k
A
stringlengths
1
3k
Result
stringclasses
4 values
Theorem 4.8.2 (Miller[84]) Let \( G \) be a Polish group and \( H \) a Borel subgroup. Suppose the \( \sigma \) -algebra of invariant Borel sets is countably generated. Then \( H \) is closed.
Proof of 4.8.2. Let \( X = G/H \), the set of right cosets, and \( q : G \rightarrow \) \( G/H \) the quotient map. Equip \( G/H \) with the largest \( \sigma \) -algebra making \( q \) Borel measurable. By our hypothesis, \( X \) is a countably generated measurable space with singletons as atoms. Consider the action \( \left( {g,{g}^{\prime }H}\right) \rightarrow g \cdot {g}^{\prime }H \) of \( G \) on \( X \) . Let \( x = H \) . Then the stabilizer\n\n\[ \n{G}_{x} = \{ g \in G : g \cdot x = x\} = H.\n\]\n\nSince \( g \rightarrow g \cdot x \) is Borel, the result follows from 4.8.5.
Yes
Proposition 4.8.3 Let \( X \) be a Polish space and \( G \) a group of homeomorphisms of \( X \) such that for every pair \( U, V \) of nonempty open sets there is a \( g \in G \) with \( g\left( U\right) \cap V \neq \varnothing \) . Suppose \( A \) is a \( G \) -invariant Borel set; i.e., \( g\left( A\right) = A \) for all \( g \in G \) . Then either \( A \) or \( {A}^{c} \) is meager in \( X \) .
Proof. Suppose neither \( A \) nor \( {A}^{c} \) is meager in \( X \) . Then there exist nonempty open sets \( U, V \) such that \( A \) and \( {A}^{c} \) are comeager in \( U \) and \( V \) respectively. By our hypothesis, there is a \( g \in G \) such that \( g\left( U\right) \cap V \neq \varnothing \) . Let \( W = g\left( U\right) \bigcap V \) . It follows that \( W \) is meager. This contradicts the Baire category theorem.
Yes
Theorem 4.8.4 (Miller[84]) Let \( \\left( {G, \\cdot }\\right) \) be a Polish group, \( X \) a second countable \( {T}_{1} \) space, and \( \\left( {g, x}\\right) \\rightarrow g \\cdot x \) an action of \( G \) on \( X \) . Suppose that for a given \( x \), the map \( g \\rightarrow g \\cdot x \) is Borel. Then the stabilizer \( {G}_{x} \) is closed.
Proof. Let \( H = \\operatorname{cl}\\left( {G}_{x}\\right) \) . It is fairly easy to see that we can replace \( G \) by \( H \) . Hence, without loss of generality we assume that \( {G}_{x} \) is dense in \( G \) . \n\nSince \( X \) is second countable and \( {T}_{1},{G}_{x} \) is Borel. Therefore, by 3.5.13, we shall be done if we show that \( {G}_{x} \) is nonmeager. Suppose not. We shall get a contradiction. Take a countable base \( \\left( {U}_{n}\\right) \) for \( X \) . Let \( f\\left( g\\right) = g \\cdot x \) . As \( f \) is Borel, \( {f}^{-1}\\left( {U}_{n}\\right) = {A}_{n} \), say, is Borel. For every \( h \\in {G}_{x},{A}_{n} \\cdot h = {A}_{n} \) . Since \( X \) is \( {T}_{1} \), for any two \( g, h \) we have \n\n\[ \n g \\cdot x = h \\cdot x \\Leftrightarrow \\forall n\\left( {g \\in {A}_{n} \\Leftrightarrow h \\in {A}_{n}}\\right) . \n\] \n\nHence, for any \( g \\in G \)\n\n\[ \ng{G}_{x} = \\bigcap \\left\\{ {{A}_{n} : g \\in {A}_{n}}\\right\\} \n\] \n\nApplying 4.8.3 to the group of homeomorphisms of \( G \) induced by right multiplication by elements of \( {G}_{x} \), we see that \( {A}_{n} \) is either meager or comeager. Since \( {G}_{x} \) is meager, there exists \( n \) such that \( g \\in {A}_{n} \) and \( {A}_{n} \) is meager. Hence, \n\n\[ \nG = \\bigcup \\left\\{ {{A}_{n} : {A}_{n}\\text{ meager }}\\right\\} . \n\] \n\nThis contradicts the Baire category theorem, and our result is proved.
Yes
Theorem 4.8.6 Let \( G \) be a Polish group, \( X \) a Polish space, and \( a\left( {g, x}\right) = \) \( g \cdot x \) an action of \( G \) on \( X \) . Assume that \( g \cdot x \) is continuous in \( x \) for all \( g \) and Borel in \( g \) for all \( x \) . Then the action is continuous.
Proof. By 3.1.30, the action \( a : G \times X \rightarrow X \) is Borel. Let \( \left( {V}_{n}\right) \) be a countable base for \( X \) . Put \( {C}_{n} = {a}^{-1}\left( {V}_{n}\right) \) . Then \( {C}_{n} \) is Borel with open sections. By 4.7.2, write\n\n\[ \n{C}_{n} = \mathop{\bigcup }\limits_{m}\left( {{B}_{nm} \times {W}_{nm}}\right)\n\]\n\nthe \( {B}_{nm} \)’s Borel, the \( {W}_{nm} \)’s open. By 3.5.1, \( {B}_{nm} \) has the Baire property. Let \( {I}_{nm} \) be a meager set in \( G \) such that \( {B}_{nm}\Delta {I}_{nm} \) is open. Put \( I = \mathop{\bigcup }\limits_{{nm}}{I}_{nm} \) . Then \( I \) is meager in \( G \) and \( a \mid \left( {G \smallsetminus I}\right) \times X \) is continuous.\n\nNow take a sequence \( \left( {{g}_{k},{x}_{k}}\right) \) in \( G \times X \) converging to \( \left( {g, x}\right) \), say. We need to show that \( {g}_{k} \cdot {x}_{k} \rightarrow g \cdot x \) . Let\n\n\[ \nJ = \mathop{\bigcup }\limits_{k}I \cdot {g}_{k}^{-1}\bigcup I \cdot {g}^{-1}.\n\]\n\nSince \( G \) is a topological group, \( J \) is meager in \( G \) . By the Baire category theorem, \( G \neq J \) . Take any \( h \in G \smallsetminus J \) . Then \( h \cdot g, h \cdot {g}_{k} \in G \smallsetminus I \) . As \( {g}_{k} \rightarrow g \) , \( h \cdot {g}_{k} \rightarrow h \cdot g \) . Since \( a \mid \left( {G \smallsetminus I}\right) \times X \) is continuous, \( \left( {h \cdot {g}_{k}}\right) \cdot {x}_{k} \rightarrow \left( {h \cdot g}\right) \cdot x \) . Since the action is continuous in the second variable,\n\n\[ \n{g}_{k} \cdot {x}_{k} = {h}^{-1} \cdot \left( {\left( {h \cdot {g}_{k}}\right) \cdot {x}_{k}}\right) \rightarrow {h}^{-1} \cdot \left( {\left( {h \cdot g}\right) \cdot x}\right) = g \cdot x.\n\]
Yes
Lemma 4.8.8 If \( \left( {G, \cdot }\right) \) is a group with a Polish topology such that the group operation \( \left( {g, h}\right) \rightarrow g \cdot h \) is Borel, then \( g \rightarrow {g}^{-1} \) is continuous.
Proof. Since \( \left( {g, h}\right) \rightarrow g \cdot h \) is Borel, the graph\n\n\[ \n\{ \left( {g, h}\right) : g \cdot h = e\}\n\]\n\nof \( g \rightarrow {g}^{-1} \) is Borel. Hence, by 4.5.2, \( g \rightarrow {g}^{-1} \) is Borel measurable. An imitation of the proof of 3.5.9 shows that \( g \rightarrow {g}^{-1} \) is continuous.
No
Proposition 4.8.9 If \( \left( {G, \cdot }\right) \) is a group with a Polish topology such that the group operation is separately continuous in each variable, then \( G \) is a topological group.
Proof. In view of 4.8.8, we have only to show that the group operation is jointly continuous. This we get immediately by applying 4.8.6 to \( X = G \) and action \( g \cdot x \) the group operation.
Yes
Theorem 4.8.10 (S. Solecki and S. M. Srivastava[109]) Let \( \left( {G, \cdot }\right) \) be a group with a Polish topology such that \( h \rightarrow g \cdot h \) is continuous for every \( g \in G \), and \( g \rightarrow g \cdot h \) Borel for all \( h \) . Then \( G \) is a topological group.
Proof. By 4.8.9, we only have to show that the group operation \( g \cdot h \) is jointly continuous. A close examination of the proof of 4.8.6 shows that this follows from the following result.\n\nLemma 4.8.11 Let \( G \) satisfy the hypothesis of our theorem. Then for every meager set \( I \) and every \( g \) ,\n\n\[ \n{Ig} = \{ h \cdot g : h \in I\} \n\]\n\n is meager.\n\nProof.\n\nClaim. If \( I \) is meager in \( G \), so is \( {I}^{-1} = \left\{ {h \in G : {h}^{-1} \in I}\right\} \) .\n\nAssuming the claim, we prove the lemma as follows. Let \( I \) be meager in \( G \) and \( g \in G \) . By the claim, \( {I}^{-1} \) is meager. Since the group operation is continuous in the second varible, \( J = {g}^{-1} \cdot {I}^{-1} \) is meager. As \( I \cdot g = {J}^{-1} \) , it is meager by our claim.\n\nProof of the claim. Let \( I \) be meager. Since every meager set is contained in a meager \( {F}_{\sigma } \), without any loss of generality we assume that \( I \) is Borel. By 3.1.30, the group operation \( \left( {g, h}\right) \rightarrow g \cdot h \) is a Borel map. Since the graph of \( g \rightarrow {g}^{-1} \) is Borel, \( g \rightarrow {g}^{-1} \) is Borel measurable (4.5.2). Hence, \( \left( {g, h}\right) \rightarrow {g}^{-1} \cdot h \) is Borel measurable. Let\n\n\[ \n\widehat{I} = \left\{ {\left( {h, g}\right) : {g}^{-1} \cdot h \in I}\right\} .\n\]\n\nSince \( \widehat{I} \) is a Borel set, it has the Baire property. Now, for every \( g \in G \) ,\n\n\[ \n{\widehat{I}}^{g} = \left\{ {h \in G : {g}^{-1} \cdot h \in I}\right\} = g \cdot I.\n\]\n\nHence, by our hypothesis, \( {\widehat{I}}^{g} \) is meager for every \( g \) . Therefore, by the Kura-towski - Ulam theorem (3.5.16), the set \( \left\{ {h : {\widehat{I}}_{h}}\right. \) is meager \( \} \) is comeager and hence nonempty by the Baire category theorem. In particular, there exists \( h \in G \) such that \( {\widehat{I}}_{h} = h \cdot {I}^{-1} \) is meager. It follows that \( {I}^{-1} = {h}^{-1}\left( {h{I}^{-1}}\right) \) is meager.
Yes
Lemma 4.8.11 Let \( G \) satisfy the hypothesis of our theorem. Then for every meager set \( I \) and every \( g \) , \[ {Ig} = \{ h \cdot g : h \in I\} \] is meager.
Proof. Claim. If \( I \) is meager in \( G \), so is \( {I}^{-1} = \left\{ {h \in G : {h}^{-1} \in I}\right\} \) . Assuming the claim, we prove the lemma as follows. Let \( I \) be meager in \( G \) and \( g \in G \) . By the claim, \( {I}^{-1} \) is meager. Since the group operation is continuous in the second varible, \( J = {g}^{-1} \cdot {I}^{-1} \) is meager. As \( I \cdot g = {J}^{-1} \) , it is meager by our claim. Proof of the claim. Let \( I \) be meager. Since every meager set is contained in a meager \( {F}_{\sigma } \), without any loss of generality we assume that \( I \) is Borel. By 3.1.30, the group operation \( \left( {g, h}\right) \rightarrow g \cdot h \) is a Borel map. Since the graph of \( g \rightarrow {g}^{-1} \) is Borel, \( g \rightarrow {g}^{-1} \) is Borel measurable (4.5.2). Hence, \( \left( {g, h}\right) \rightarrow {g}^{-1} \cdot h \) is Borel measurable. Let \[ \widehat{I} = \left\{ {\left( {h, g}\right) : {g}^{-1} \cdot h \in I}\right\} . \] Since \( \widehat{I} \) is a Borel set, it has the Baire property. Now, for every \( g \in G \) , \[ {\widehat{I}}^{g} = \left\{ {h \in G : {g}^{-1} \cdot h \in I}\right\} = g \cdot I. \] Hence, by our hypothesis, \( {\widehat{I}}^{g} \) is meager for every \( g \) . Therefore, by the Kura-towski - Ulam theorem (3.5.16), the set \( \left\{ {h : {\widehat{I}}_{h}}\right. \) is meager \( \} \) is comeager and hence nonempty by the Baire category theorem. In particular, there exists \( h \in G \) such that \( {\widehat{I}}_{h} = h \cdot {I}^{-1} \) is meager. It follows that \( {I}^{-1} = {h}^{-1}\left( {h{I}^{-1}}\right) \) is meager.
No
Lemma 4.9.3 Let \( X \) be a Polish space, \( A \subseteq X \) coanalytic, and \( \varphi \) a norm on \( A \) . Then \( \varphi \) is a \( {\mathbf{\Pi }}_{1}^{1} \) -norm if and only if both \( { \leq }_{\varphi }^{ * },{ < }_{\varphi }^{ * } \) are coanalytic.
Proof. We first prove the \
No
Example 4.9.4 Let \( X = {2}^{\mathbb{N} \times \mathbb{N}} \) and \( A = {WO} \). For \( x \in {WO} \), Let \( \left| x\right| < {\omega }_{1} \) be the order type of \( x \).
For \( x \in {2}^{\mathbb{N} \times \mathbb{N}} \), define\n\n\[ m{ < }_{x}n \Leftrightarrow x\left( {m, n}\right) = 1\& x\left( {n, m}\right) = 0. \]\n\nFor \( x, y \) in \( {2}^{\mathbb{N} \times \mathbb{N}} \), set\n\n\[ x{ \leq }_{\left| \cdot \right| }^{{\sum }_{1}^{1}}y \Leftrightarrow \exists z \in {\mathbb{N}}^{\mathbb{N}}\forall m\forall n\left\lbrack {m{ < }_{x}n \Leftrightarrow z\left( m\right) { < }_{y}z\left( n\right) }\right\rbrack \]\n\nand\n\n\[ x{ < }_{\left| \cdot \right| }^{{\sum }_{1}^{1}}y\; \Leftrightarrow \;\exists k\exists z \in {\mathbb{N}}^{\mathbb{N}}\forall m\forall n\lbrack z\left( m\right) { < }_{y}k \]\n\n\[ \left. {\& \left( {m{ < }_{x}n \Leftrightarrow z\left( m\right) { < }_{y}z\left( n\right) }\right) }\right\rbrack . \]\n\nThus, \( x{ \leq }_{\left| \cdot \right| }^{{\sum }_{1}^{1}}y \) if and only if there is an order-preserving map from \( x \) to \( y \), and \( x{ < }_{\left| \cdot \right| }^{{\sum }_{1}^{1}}y \) if and only if there is an order-preserving map from \( x \) into an initial segment of \( y \). The sets \( { \leq }_{\left| \cdot \right| }^{{\sum }_{1}^{1}} \) and \( { < }_{\left| \cdot \right| }^{{\sum }_{1}^{1}} \) are clearly \( {\mathbf{\sum }}_{1}^{1} \). Further, for \( y \in {WO} \), \n\n\[ x{ \leq }_{\left| \cdot \right| }^{{\sum }_{1}^{1}}y \Leftrightarrow x \in {WO}\& \left| x\right| \leq \left| y\right| \]\n\nand\n\n\[ x{ < }_{\left| \cdot \right| }^{{\sum }_{1}^{1}}y \Leftrightarrow x \in {WO}\& \left| x\right| < \left| y\right| . \]\n\nTherefore, by 4.9.2, \( \left| \cdot \right| \) is a \( {\mathbf{\Pi }}_{1}^{1} \) -norm on \( {WO} \), which we shall call the canonical norm on \( {WO} \).
Yes
Theorem 4.9.8 (Boundedness theorem for \( {\mathbf{\Pi }}_{1}^{1} \) -norms) Suppose \( A \) is a \( {\mathbf{\Pi }}_{1}^{1} \) set in a Polish space \( X \) and \( \varphi \) a norm on \( A \) as defined in 4.9.1. Then for every \( {\mathbf{\sum }}_{1}^{1} \) set \( B \subseteq A,\sup \{ \varphi \left( x\right) : x \in B\} < {\omega }_{1} \) .
Proof. Suppose \( \sup \{ \varphi \left( y\right) : y \in B\} = {\omega }_{1} \) . Take any \( {\mathbf{\Pi }}_{1}^{1} \) set \( C \) that is not \( {\mathbf{\sum }}_{1}^{1} \) . Fix a Borel function \( g \) such that\n\n\[ x \in C \Leftrightarrow g\left( x\right) \in {WO}. \]\n\nThen,\n\n\[ x \in C\; \Leftrightarrow \;\exists y\left( {y \in B\& \left| {g\left( x\right) }\right| \leq \varphi \left( y\right) }\right) \]\n\n\[ \Leftrightarrow \exists y\left( {y \in B\& g\left( x\right) { \leq }_{\left| .\right| }^{{\sum }_{1}^{1}}f\left( y\right) }\right) , \]\n\nwhere \( f \) is as in 4.9.1. This contradicts the fact that \( C \) is not \( {\mathbf{\sum }}_{1}^{1} \) . Hence,\n\n\[ \sup \{ \varphi \left( x\right) : x \in B\} < {\omega }_{1}. \]\n\nIf \( A \) is Borel, then taking \( B \) to be \( A \), we see that \( \sup \{ \varphi \left( x\right) : x \in A\} < {\omega }_{1} \) . On the other hand, if \( \sup \{ \varphi \left( x\right) : x \in A\} < {\omega }_{1} \), then \( A \) is a union of countably many Borel sets of the form \( \{ x \in A : \varphi \left( x\right) = \xi \} ,\xi < {\omega }_{1} \) . So \( A \) is Borel.
Yes
Example 4.9.11 (A. Maitra and C. Ryll-Nardzewski[76]) Let \( X, Y \) be uncountable Polish spaces. Let \( U \subseteq X \times X \) be universal analytic and \( C \subseteq Y \) an uncountable coanalytic set not containing a perfect set. We mentioned earlier that Gödel's axiom of constructibility implies the existence of such a set. The set \( C \) does not contain any uncountable Borel set. Take \( A = Y \smallsetminus C \) . Then \( U \) and \( A \) are not Borel isomorphic.
Here is a proof. Suppose they are Borel isomorphic. Take a Borel isomorphism \( f : U \rightarrow A \) . By 3.3.5, there exist Borel sets \( {B}_{1} \supseteq U,{B}_{2} \supseteq A \) and a Borel isomorphism \( g : {B}_{1} \rightarrow {B}_{2} \) extending \( f \) . Let \( \varphi \) be a \( {\mathbf{\Pi }}_{1}^{1} \) norm on \( {U}^{c} \) as defined in 4.9.8. It is easy to verify that for uncountably many \( \xi < {\omega }_{1},\left\{ {\left( {x, y}\right) \in {U}^{c} : \varphi \left( {x, y}\right) = }\right. \) \( \xi \} \) is uncountable. By 4.9.8, \( \sup \left\{ {\varphi \left( {x, y}\right) : \left( {x, y}\right) \in {B}_{1}^{c}}\right\} < {\omega }_{1} \) . Therefore, \( {B}_{1} \smallsetminus U \) contains an uncountable Borel set. It follows that \( C \) contains an uncountable Borel set, which is not the case. Hence, \( U \) and \( A \) are not Borel isomorphic.
Yes
Theorem 4.9.14 (The reduction principle for coanalytic sets) (Kuratowski) Let \( \\left( {A}_{n}\\right) \) be sequence of \( {\\mathbf{\\Pi }}_{1}^{1} \) sets in a Polish space \( X \) . Then there is a sequence \( \\left( {A}_{n}^{ * }\\right) \) of \( {\\mathbf{\\Pi }}_{1}^{1} \) sets such that they are pairwise disjoint, \( {A}_{n}^{ * } \\subseteq {A}_{n} \) , and \( \\mathop{\\bigcup }\\limits_{n}{A}_{n}^{ * } = \\mathop{\\bigcup }\\limits_{n}{A}_{n} \) .
Proof. Consider \( A \\subseteq X \\times \\mathbb{N} \) given by\n\n\[ \n\\left( {x, n}\\right) \\in A \\Leftrightarrow x \\in {A}_{n} \n\] \n\nClearly, \( A \) is \( {\\mathbf{\\Pi }}_{1}^{1} \) with projection \( \\mathop{\\bigcup }\\limits_{n}{A}_{n} \) . Let \( \\varphi \) be a \( {\\mathbf{\\Pi }}_{1}^{1} \)-norm on \( A \) . Define \( {A}^{ * } \\subseteq X \\times \\mathbb{N} \) by\n\n\[ \n\\left( {x, n}\\right) \\in {A}^{ * }\\; \\Leftrightarrow \\;\\left( {x, n}\\right) \\in A\\& \\forall m\\left\\lbrack {\\left( {x, n}\\right) { \\leq }_{\\varphi }^{ * }\\left( {x, m}\\right) }\\right\\rbrack \n\] \n\n\[ \n\\text{&}\\forall m\\left\\lbrack {\\left( {x, n}\\right) { < }_{\\varphi }^{ * }\\left( {x, m}\\right) \\text{or}n \\leq m}\\right\\rbrack \\text{.} \n\] \n\nThus, for each \( x \) in the projecton of \( A \) we first look at the set of integers \( n \) with \( \\left( {x, n}\\right) \\in A \) such that \( \\varphi \\left( {x, n}\\right) \) is the minimum. Then we choose the least among these integers. Note that \( {A}^{ * } \) is \( {\\mathbf{\\Pi }}_{1}^{1},{A}^{ * } \\subseteq A \), and for every \( x \\in \\mathop{\\bigcup }\\limits_{n}{A}_{n} \) there is exactly one \( n \) such that \( \\left( {x, n}\\right) \\in {A}^{ * } \) . Let\n\n\[ \n{A}_{n}^{ * } = \\{ x : \\left( {x, n}\\right) \\in A * \\} . \n\] \n\nClearly, \( {A}_{n}^{ * } \) is \( {\\mathbf{\\Pi }}_{1}^{1} \) . It is easy to check that the \( {A}_{n}^{ * } \\)’s are pairwise disjoint and \( \\mathop{\\bigcup }\\limits_{n}{A}_{n}^{ * } = \\mathop{\\bigcup }\\limits_{n}{A}_{n} \) .
Yes
Corollary 4.9.15 Let \( X \) be Polish and \( {A}_{0},{A}_{1} \) coanalytic subsets of \( X \) . Then there exist pairwise disjoint coanalytic sets \( {A}_{0}^{ * },{A}_{1}^{ * } \) contained in \( {A}_{0} \) , \( {A}_{1} \) respectively such that \( {A}_{0}^{ * }\bigcup {A}_{1}^{ * } = {A}_{0}\bigcup {A}_{1} \) .
Proof. In the above theorem, take \( {A}_{n} = \varnothing \) for \( n > 1 \) .
Yes
Theorem 4.9.19 Let \( X \) be a Polish space. Then there exist sets \( C \in \) \( {\mathbf{\Pi }}_{1}^{1}\left( {\mathbb{N}}^{\mathbb{N}}\right) \) and \( V \in {\mathbf{\Pi }}_{1}^{1}\left( {{\mathbb{N}}^{\mathbb{N}} \times X}\right), U \in {\mathbf{\sum }}_{1}^{1}\left( {{\mathbb{N}}^{\mathbb{N}} \times X}\right) \) such that for every \( \alpha \in C,{U}_{\alpha } = {V}_{\alpha } \) and\n\n\[ \n{\mathbf{\Delta }}_{1}^{1}\left( X\right) = \left\{ {{U}_{\alpha } : \alpha \in C}\right\} \n\] \n\nIn particular, there are a coanalytic set and an analytic set contained in \( {\mathbb{N}}^{\mathbb{N}} \times X \) that are universal for \( {\mathbf{\Delta }}_{1}^{1}\left( X\right) \) .
Proof. Let \( {W}_{0},{W}_{1} \) be coanalytic subsets of \( {\mathbb{N}}^{\mathbb{N}} \times X \) such that for every pair \( \left( {{C}_{0},{C}_{1}}\right) \) of sets in \( {\mathbf{\Pi }}_{1}^{1}\left( X\right) \) there is an \( \alpha \) with \( {C}_{i} = {\left( {W}_{i}\right) }_{\alpha }, i = 0 \) or 1 . By the reduction principle for coanalytic sets, (4.9.15), there are pairwise disjoint coanalytic sets \( {V}_{i} \subseteq {W}_{i}, i = 0 \) or 1, such that \( {V}_{0}\bigcup {V}_{1} = {W}_{0}\bigcup {W}_{1} \) . Define\n\n\[ \nC = \left\{ {\alpha : \forall x\left( {\left( {\alpha, x}\right) \in {V}_{0}\bigcup {V}_{1}}\right) }\right\} \n\] \n\nSo, \( C \) is coanalytic. Take \( V = {V}_{0} \) and \( U = {V}_{1}^{c} \) . A routine argument shows that \( C, U \), and \( V \) have the desired properties. So, we have proved the first part of the result.\n\nTo see the second part, note that \( V\bigcap \left( {C \times X}\right) \) is a coanalytic set universal for \( {\mathbf{\Delta }}_{1}^{1}\left( X\right) \), and its complement is an analytic set universal for \( {\mathbf{\Delta }}_{1}^{1}\left( X\right) \) .
Yes
Example 4.10.1 Let \( \mu \) be a finite Borel measure on a Polish space \( X \) and \( {\mu }^{ * } \) the associated outer measure. Thus, for any \( A \subseteq X \) , \[ {\mu }^{ * }\left( A\right) = \inf \{ \mu \left( B\right) : B \supseteq A, B\text{ Borel }\} . \]
It is easy to check that \( {\mu }^{ * } \) is a capacity on \( X \) .
No
Proposition 4.10.4 Let \( I \) be a capacity on a Polish space \( X \) and that \( {I}^{ * } : \mathcal{P}\left( X\right) \rightarrow \left\lbrack {0,\infty }\right\rbrack \) be defined by\n\n\[ \n{I}^{ * }\left( A\right) = \inf \{ I\left( B\right) : B \supseteq A, B\text{ Borel }\} .\n\]\n\nThen \( {I}^{ * } \) is a capacity on \( X \) .
Proof. Clearly, \( {I}^{ * } \) is monotone. Further, \( {I}^{ * } \) and \( I \) coincide on Borel sets. As \( I \) is a capacity, it follows that \( {I}^{ * }\left( K\right) < \infty \) for every compact \( K \) and that \( {I}^{ * } \) is right-continuous over compacta.\n\nTo show that \( {I}^{ * } \) is going up, take any nondecreasing sequence \( \left( {A}_{n}\right) \) of subsets of \( X \) . Set \( A = \mathop{\bigcup }\limits_{n}{A}_{n} \) . Note that for every \( C \subseteq X \), there is a Borel \( D \supseteq C \) such that \( {I}^{ * }\left( C\right) = I\left( D\right) \) . Hence, for every \( n \) there is a Borel \( {B}_{n} \supseteq {A}_{n} \) such that \( I\left( {B}_{n}\right) = {I}^{ * }\left( {A}_{n}\right) \) . Replacing \( {B}_{n} \) by \( \mathop{\bigcap }\limits_{{m > n}}{B}_{m} \), we may assume that \( \left( {B}_{n}\right) \) is nondecreasing. Set \( B = \mathop{\bigcup }\limits_{n}{B}_{n} \) . Clearly,\n\n\[ \nI\left( B\right) \geq {I}^{ * }\left( A\right) \geq {I}^{ * }\left( {A}_{n}\right) = I\left( {B}_{n}\right)\n\]\n\nfor every \( n \) . Since \( I \) is going up, \( \lim I\left( {B}_{n}\right) = I\left( B\right) \) . It follows that \( \left. {\lim {I}^{ * }\left( {A}_{n}\right) = {I}^{ * }A}\right) \) .
Yes
Proposition 4.10.4 Let \( I \) be a capacity on a Polish space \( X \) and that \( {I}^{ * } : \mathcal{P}\left( X\right) \rightarrow \left\lbrack {0,\infty }\right\rbrack \) be defined by\n\n\[ \n{I}^{ * }\left( A\right) = \inf \{ I\left( B\right) : B \supseteq A, B\text{ Borel }\} .\n\]\n\nThen \( {I}^{ * } \) is a capacity on \( X \) .
Proof. Clearly, \( {I}^{ * } \) is monotone. Further, \( {I}^{ * } \) and \( I \) coincide on Borel sets. As \( I \) is a capacity, it follows that \( {I}^{ * }\left( K\right) < \infty \) for every compact \( K \) and that \( {I}^{ * } \) is right-continuous over compacta.\n\nTo show that \( {I}^{ * } \) is going up, take any nondecreasing sequence \( \left( {A}_{n}\right) \) of subsets of \( X \) . Set \( A = \mathop{\bigcup }\limits_{n}{A}_{n} \) . Note that for every \( C \subseteq X \), there is a Borel \( D \supseteq C \) such that \( {I}^{ * }\left( C\right) = I\left( D\right) \) . Hence, for every \( n \) there is a Borel \( {B}_{n} \supseteq {A}_{n} \) such that \( I\left( {B}_{n}\right) = {I}^{ * }\left( {A}_{n}\right) \) . Replacing \( {B}_{n} \) by \( \mathop{\bigcap }\limits_{{m > n}}{B}_{m} \), we may assume that \( \left( {B}_{n}\right) \) is nondecreasing. Set \( B = \mathop{\bigcup }\limits_{n}{B}_{n} \) . Clearly,\n\n\[ \nI\left( B\right) \geq {I}^{ * }\left( A\right) \geq {I}^{ * }\left( {A}_{n}\right) = I\left( {B}_{n}\right)\n\]\n\nfor every \( n \) . Since \( I \) is going up, \( \lim I\left( {B}_{n}\right) = I\left( B\right) \) . It follows that \( \left. {\lim {I}^{ * }\left( {A}_{n}\right) = {I}^{ * }A}\right) \) .
Yes
Proposition 4.10.10 Let \( I \) be a capacity on a Polish space \( X \) and \( A \subseteq X \) universally capacitable. Then\n\n\[ I\left( A\right) = {I}^{ * }\left( A\right) \]\n\nwhere \( {I}^{ * } \) is as defined in 4.10.4.
Proof. By 4.10.4, \( {I}^{ * } \) is a capacity. Now note the following.\n\n\[ {I}^{ * }\left( A\right) = \sup \left\{ {{I}^{ * }\left( K\right) : K \subseteq A\text{ compact }}\right\} \;\text{ (as }A\text{ is }{I}^{ * } - \text{ capacitable) }\n\n= \;\sup \{ I\left( K\right) : K \subseteq A\text{ compact}\} \n\n= I\left( A\right) \]
Yes
Proposition 4.10.11 \( {\mathbb{N}}^{\mathbb{N}} \) is universally capacitable.
Proof. For any \( s = \left( {{n}_{0},{n}_{1},\ldots ,{n}_{k - 1}}\right) \in {\mathbb{N}}^{ < \mathbb{N}} \), set\n\n\[ \n{\sum }^{ * }\left( s\right) = \left\{ {\alpha \in {\mathbb{N}}^{\mathbb{N}} : \left( {\forall i < k}\right) \left( {\alpha \left( i\right) \leq {n}_{i}}\right) }\right\} \n\]\n\nTake any capacity \( I \) on \( {\mathbb{N}}^{\mathbb{N}} \) and a real number \( t \) such that \( I\left( {\mathbb{N}}^{\mathbb{N}}\right) > t \) . To prove our result, we shall show that there is a compact set \( K \) such that \( I\left( K\right) \geq t \) .\n\nSince the sequence \( \left( {{\sum }^{ * }\left( n\right) }\right) \) increases to \( {\mathbb{N}}^{\mathbb{N}} \), there is a natural number \( {n}_{0} \) such that \( I\left( {{\sum }^{ * }\left( {n}_{0}\right) }\right) > t \) . Again, since \( \left( {{\sum }^{ * }\left( {{n}_{0}n}\right) }\right) \) increases to \( {\sum }^{ * }\left( {n}_{0}\right) \), there is a natural number \( {n}_{1} \) such that \( I\left( {{\sum }^{ * }\left( {{n}_{0}{n}_{1}}\right) }\right) > t \) . Proceeding similarly, we get a sequence \( {n}_{0},{n}_{1},{n}_{2},\ldots \) of natural numbers such that\n\n\[ \nI\left( {{\sum }^{ * }\left( {{n}_{0}{n}_{1}\ldots {n}_{k - 1}}\right) }\right) > t \n\]\n\nfor every \( k \) . Now consider\n\n\[ \nK = \left\{ {\alpha \in {\mathbb{N}}^{\mathbb{N}} : \alpha \left( i\right) \leq {n}_{i}\text{ for every }i}\right\} .\n\]\n\nClearly, \( K \) is compact. We claim that \( I\left( K\right) \geq t \) . Suppose not. Since \( I \) is right-continuous over compacta, there is an open set \( U \supseteq K \) such that \( I\left( U\right) < t \) . It is not very hard to show that \( U \supseteq {\sum }^{ * }\left( {{n}_{0}{n}_{1}\ldots {n}_{k - 1}}\right) \) ) for some \( k \) . Since \( I \) is monotone, we have arrived at a contradiction.
Yes
Theorem 4.10.12 (Choquet capacitability theorem [30], [107]) Every analytic subset of a Polish space is universally capacitable.
Proof. Let \( X \) be a Polish space and \( A \subseteq X \) analytic. Let \( I \) be any capacity on \( X \) . Suppose \( I\left( A\right) > t \) . Let \( f : {\mathbb{N}}^{\mathbb{N}} \rightarrow X \) be a continuous map with range \( A \) . By 4.10.11, there is a compact \( K \subseteq {\mathbb{N}}^{\mathbb{N}} \) such that \( {I}_{f}\left( K\right) > t \) . Plainly \( I\left( {f\left( K\right) }\right) > t \), and our result is proved.
No
Proposition 4.10.13 Let \( X \) be a Polish space and \( I \) the separation capacity on \( X \times X \) as defined in 4.10.2. Assume that a rectangle \( {A}_{1} \times {A}_{2} \) be universally capacitable. If \( I\left( {{A}_{1} \times {A}_{2}}\right) = 0 \), then there is a Borel rectangle \( B = {B}_{1} \times {B}_{2} \) containing \( {A}_{1} \times {A}_{2} \) of \( I \) -capacity 0 .
Proof of 4.10.13. Set \( {C}_{0} = {A}_{1} \times {A}_{2} \) . By 4.10.10, there is a Borel \( {C}_{1} \supseteq {C}_{0} \) such that \( I\left( {C}_{1}\right) = 0 \) . Set \( {C}_{2} = R\left\lbrack {C}_{1}\right\rbrack \) . (Recall that \( \mathrm{R}\left\lbrack \mathrm{A}\right\rbrack \) denotes the smallest rectangle containing \( A \) .) Clearly \( I\left( {C}_{2}\right) = 0 \) . Since \( {C}_{2} \) is analytic, by 4.10.12, it is universally capacitable. By 4.10.10, there is a Borel \( {C}_{3} \supseteq {C}_{2} \) such that \( I\left( {C}_{3}\right) = 0 \) . Set \( {C}_{4} = R\left\lbrack {C}_{3}\right\rbrack \) . Proceeding similarly, we get a nondecreasing sequence \( \left( {C}_{n}\right) \) of subsets of \( X \times X \) such that \( {C}_{n} \) is a rectangle for even \( n \) and \( {C}_{n} \) ’s are Borel for odd \( n \) . Further \( I\left( {C}_{n}\right) = 0 \) for all \( n \) . Take \( B = \mathop{\bigcup }\limits_{n}{C}_{n} \) .
Yes
Theorem 4.11.1 (Second separation theorem for analytic sets) (Kuratowski) Let \( X \) be a Polish space and \( A, B \) two analytic subsets. There exist disjoint coanalytic sets \( C \) and \( D \) such that\n\n\[ A \smallsetminus B \subseteq C \\text{and} B \smallsetminus A \subseteq D. \]
Proof. By 4.1.20, there exist Borel maps \( f : X \rightarrow {LO}, g : X \rightarrow {LO} \) such that \( {f}^{-1}\left( {WO}\right) = {A}^{c} \) and \( {g}^{-1}\left( {WO}\right) = {B}^{c} \). \n\nFor \( \alpha ,\beta \) in \( {LO} \), define\n\n\[ \alpha \preccurlyeq \beta \; \Leftrightarrow \;\exists f \in {\\mathbb{N}}^{\\mathbb{N}}(f \\mid D\\left( \\alpha \\right) \\text{ is one-to-one }\n\n\\left. {\\& \\forall m\\forall n\\left( {\\alpha \\left( {m, n}\\right) = 1 \\Leftrightarrow \\beta \\left( {f\\left( m\\right), f\\left( n\\right) }\\right) = 1}\\right) }\\right) .\n\n(Recall that for any \( \\alpha \\in {\\mathbb{N}}^{\\mathbb{N}}, n \\in D\\left( \\alpha \\right) \\Leftrightarrow \\alpha \\left( {n, n}\\right) = 1 \).) So \( \\preccurlyeq \) is an analytic subset of \( {\\mathbb{N}}^{\\mathbb{N}} \\times {\\mathbb{N}}^{\\mathbb{N}} \). Let\n\n\[ C = {B}^{c}\\bigcap \\{ x \\in X : f\\left( x\\right) \\preccurlyeq g\\left( x\\right) {\\} }^{c}\n\nand\n\n\[ D = {A}^{c}\\bigcap \\{ x \\in X : g\\left( x\\right) \\preccurlyeq f\\left( x\\right) {\\} }^{c}.\n\nClearly, \( C \) and \( D \) are coanalytic. We claim that \( C \) and \( D \) are disjoint. Suppose not. Take any \( x \\in C \\cap D \). Then both \( f\\left( x\\right) \) and \( g\\left( x\\right) \) are in \( {WO} \). Therefore, either \( \\left| {f\\left( x\\right) }\\right| \\leq \\left| {g\\left( x\\right) }\\right| \) or \( \\left| {g\\left( x\\right) }\\right| \\leq \\left| {f\\left( x\\right) }\\right| \). Since \( x \\in C \\cap D \), this is impossible.\n\nFinally, we show that \( A \\smallsetminus B \\subseteq C \). Let \( x \\in A \\smallsetminus B \). Then, of course, \( x \\in {B}^{c} \). As \( f\\left( x\\right) \\notin {WO} \) and \( g\\left( x\\right) \\in {WO} \), there is no order-preserving one-to-one map from \( D\\left( {f\\left( x\\right) }\\right) \) into \( D\\left( {g\\left( x\\right) }\\right) \). So, \( x \\in C \). Similarly it follows that \( B \\smallsetminus A \\subseteq D \).
Yes
Corollary 4.11.3 Suppose \( X \) is a Polish space and \( \left( {A}_{n}\right) \) a sequence of analytic subsets of \( X \) . Then there exists a sequence \( \left( {C}_{n}\right) \) of pairwise disjoint coanalytic sets such that
Proof. By the second separation theorem, for each \( n \) there exist pairwise disjoint coanalytic sets \( {C}_{n}^{\prime } \) and \( {D}_{n}^{\prime } \) such that\n\n\[ {A}_{n} \smallsetminus \mathop{\bigcup }\limits_{{m \neq n}}{A}_{m} \subseteq {C}_{n}^{\prime }\text{ and }\mathop{\bigcup }\limits_{{m \neq n}}{A}_{m} \smallsetminus {A}_{n} \subseteq {D}_{n}^{\prime } \]\n\nTake\n\n\[ {C}_{n} = {C}_{n}^{\prime } \cap \mathop{\bigcap }\limits_{{m \neq n}}{D}_{m}^{\prime } \]
Yes
Proposition 4.11.4 Suppose \( X \) is a Polish space and \( \left( {A}_{n}\right) \) a sequence of analytic subsets of \( X \) . Then there exists a sequence \( \left( {C}_{n}\right) \) of coanalytic subsets of \( X \) such that\n\n\[ \n{A}_{n} \smallsetminus \lim \sup {A}_{m} \subseteq {C}_{n} \n\]\n\n(1)\n\nand\n\n\[ \n\lim \sup {C}_{n} = \varnothing \text{.} \n\]\n\n(2)
Proof. For each \( n \), set \( {\beta }_{n} = {\beta }_{{A}_{n}} \), where \( {\beta }_{{A}_{n}} \) is as defined in 4.11.2. Let\n\n\[ \n{Q}_{nm} = \left\{ {x \in X : {\beta }_{n}\left( x\right) \leq {\beta }_{m}\left( x\right) }\right\} \n\]\n\n\( {Q}_{nm} \) is analytic by 4.11.2. Take\n\n\[ \n{C}_{n} = {\left\lbrack \mathop{\limsup }\limits_{m}\left\{ {Q}_{nm}\right\} \right\rbrack }^{c}. \n\]\n\nThen \( {C}_{n} \) is coanalytic and\n\n\[ \nx \notin {C}_{n} \Leftrightarrow \exists \eta \subseteq \mathbb{N}\left( {\eta \text{ infinite,}\& x \in \mathop{\bigcap }\limits_{{m \in \eta }}{Q}_{nm}}\right) . \n\]\n\n\( \left( *\right) \)\n\nProof of (1): Let \( x \in {A}_{n} \smallsetminus \lim \sup {A}_{m} \) . Then \( {\beta }_{n}\left( x\right) = {\omega }_{1} \) . Let \( \eta \) be any infinite subset of \( \mathbb{N} \) . Find \( m \in \eta \) such that \( x \notin {A}_{m} \) . Then \( {\beta }_{m}\left( x\right) < {\omega }_{1} = \) \( {\beta }_{n}\left( x\right) \) . So, \( x \notin {Q}_{nm} \) . By \( \left( \star \right) x \in {C}_{n} \) .\n\nProof of (2): Suppose \( \lim \sup {C}_{n} \neq \varnothing \) . Take any \( x \in \lim \sup {C}_{n} \) . Choose an infinite subset \( \eta \) of \( \mathbb{N} \) such that \( x \in {C}_{n} \) for all \( n \in \eta \) . Choose \( {n}_{0} \in \eta \) such that \( {\beta }_{{n}_{0}}\left( x\right) = \min \left\{ {{\beta }_{n}\left( x\right) : n \in \eta }\right\} \) . So, \( x \notin {C}_{{n}_{0}} \) by \( \left( \star \right) \) . This is a contradiction. Hence, \( \lim \sup {C}_{n} = \varnothing \) .
Yes
Theorem 4.12.3 (Lusin[71]) If \( X, Y \) are Polish and \( B \) a Borel subset of \( X \times Y \) such that for every \( x \in X \) the section \( {B}_{x} \) is countable, then \( {\pi }_{X}\left( B\right) \) is Borel.
Proof. Let \( E \subseteq {\mathbb{N}}^{\mathbb{N}} \) be a closed set and \( f : E \rightarrow X \times Y \) a one-to-one continuous map from \( E \) onto \( B \) . Consider \( g = {\pi }_{X} \circ f \) . For every \( x \in {\pi }_{X}\left( B\right) \) , \( {g}^{-1}\left( x\right) \) is a countable closed subset of \( E \) . Hence, by the Baire category theorem, \( {g}^{-1}\left( x\right) \) has an isolated point. Let \( {g}_{s} = g \mid \sum \left( s\right), s \in {\mathbb{N}}^{ < \mathbb{N}} \) . As\n\n\[ \n{\pi }_{X}\left( B\right) = \mathop{\bigcup }\limits_{s}{Z}_{{g}_{s}} \n\]\n\nit is coanalytic by 4.12.1. The result follows from Souslin's theorem.
Yes
Theorem 4.12.4 Suppose \( X, Y \) are Polish spaces and \( f : X \rightarrow Y \) is a countable-to-one Borel map. Then \( f\left( B\right) \) is Borel for every Borel set \( B \) in \( X \) .
Proof. The result follows from 4.12.3 and the identity\n\n\[ f\left( B\right) = {\pi }_{Y}\left( {\operatorname{graph}\left( f\right) \bigcap \left( {B \times Y}\right) }\right) .\n\]
Yes
Theorem 4.12.5 (Purves [93]) Let \( X \) be a standard Borel space, \( Y \) Polish, and \( f : X \rightarrow Y \) a bimeasurable map. Then\n\n\[ \left\{ {y \in Y : {f}^{-1}\left( y\right) \text{ is uncountable }}\right\} \] \n\nis countable.
Proof of 4.12.5. Assume that \( {f}^{-1}\left( y\right) \) is uncountable for uncountably many \( y \) . We shall show that there is a Borel \( B \subseteq X \) such that \( f\left( B\right) \) is not Borel.\n\nCase 1: \( f \) is continuous.\n\nFix a countable base \( \left( {U}_{n}\right) \) for the topology of \( X \) . Let \( G = \operatorname{graph}\left( f\right) \) . For each \( n \), let\n\n\[ {E}_{n} = \left\{ {y \in Y : {U}_{n}\bigcap {G}^{y}\text{ is countable }}\right\} \]\n\nand\n\n\[ A = G \smallsetminus \mathop{\bigcup }\limits_{n}\left( {{U}_{n} \times {E}_{n}}\right) \]\n\nBy 4.3.7, \( {E}_{n} \) is coanalytic. Hence, \( A \) is analytic. Further, \( {\pi }_{Y}\left( A\right) \) is uncountable and \( {A}^{y} \) is perfect for every \( y \in {\pi }_{Y}\left( A\right) \) . By 4.12.6, there is a homeomorph of the Cantor set \( C \) contained in \( {\pi }_{Y}\left( A\right) \) and a one-to-one Borel map \( g : {2}^{\mathbb{N}} \times C \rightarrow A \) such that \( {\pi }_{Y}\left( {g\left( {\alpha, y}\right) }\right) = y \) . Let \( D \) be a Borel subset of \( {2}^{\mathbb{N}} \times C \) such that \( {\pi }_{C}\left( D\right) \) is not Borel and let \( B = {\pi }_{X}\left( {g\left( D\right) }\right) \) . Since \( {\pi }_{X} \circ g \) is one-to-one, \( B \) is Borel by 4.5.4. Since \( f\left( B\right) = {\pi }_{C}\left( D\right) \), the result follows in this case.\n\nThe general case follows from case 1 by replacing \( X \) by \( \operatorname{graph}\left( f\right) \) and \( f \) by \( {\pi }_{Y} \mid \operatorname{graph}\left( f\right) \) .
Yes
Lemma 4.12.6 Let \( X \) be a standard Borel space, \( Y \) Polish, and \( A \subseteq X \times \) \( Y \) analytic with \( {\pi }_{X}\left( A\right) \) uncountable. Suppose that for every \( x \in {\pi }_{X}\left( A\right) \) , the section \( {A}_{x} \) is perfect. Then there is a \( C \subseteq {\pi }_{X}\left( A\right) \) homeomorphic to the Cantor set and a one-to-one Borel map \( f : C \times {2}^{\mathbb{N}} \rightarrow A \) such that \( {\pi }_{X}\left( {f\left( {x,\alpha }\right) }\right) = x \) for every \( x \) and every \( \alpha \) .
Proof of 4.12.6.\n\nFix a compatible complete metric on \( Y \) and a countable base \( \left( {U}_{n}\right) \) for the topology of \( Y \) . For each \( s \in {2}^{ < \mathbb{N}} \), we define a map \( {n}_{s}\left( x\right) : {\pi }_{X}\left( A\right) \rightarrow \mathbb{N} \) satifying the following conditions.\n\n(i) \( x \rightarrow {n}_{s}\left( x\right) \) is \( \sigma \left( {\mathbf{\sum }}_{1}^{1}\right) \) -measurable,\n\n(ii) diameter \( \left( {U}_{{n}_{s}\left( x\right) }\right) < \frac{1}{{2}^{\left| s\right| }} \) ,\n\n(iii) \( {U}_{{n}_{s}\left( x\right) } \cap {A}_{x} \neq \varnothing \) for all \( x \in {\pi }_{X}\left( A\right) \) ,\n\n(iv) \( \operatorname{cl}\left( {U}_{{n}_{s{}^{ \frown }\epsilon }\left( x\right) }\right) \subseteq {U}_{{n}_{s}\left( x\right) },\epsilon = 0 \) or \( 1 \), and\n\n(v)\n\n\[ \operatorname{cl}\left( {U}_{{n}_{{s}^{\prime }0}\left( x\right) }\right) \cap \operatorname{cl}\left( {U}_{{n}_{{s}^{\prime }1}\left( x\right) }\right) = \varnothing . \]\n\nSuch a system of functions is defined by induction on \( \left| s\right| \) . This is a fairly routine exercise, which we leave for the reader. Now fix a continuous probability measure \( P \) on \( X \) such that \( P\left( {{\pi }_{X}\left( A\right) }\right) = 1 \) . Since every set in \( \sigma \left( {\mathbf{\sum }}_{1}^{1}\right) \) is \( P \) -measurable and since \( {\pi }_{X}\left( A\right) \) is uncountable, there is a homeomorph \( C \) of the Cantor set contained in \( {\pi }_{X}\left( A\right) \) such that \( {n}_{s} \mid C \) is Borel measurable for all \( s \in {2}^{ < \mathbb{N}} \) . Take \( x \in C \) and \( \alpha \in {2}^{\mathbb{N}} \) . Note that \( \mathop{\bigcap }\limits_{k}{U}_{{n}_{\alpha \mid k\left( x\right) }} \) is a singleton, say \( \{ y\} \) . Put \( f\left( {x,\alpha }\right) = \left( {x,y}\right) \).
No
Lemma 5.1.2 Suppose \( Y \) is metrizable, \( G : X \rightarrow Y \) strongly \( \mathcal{A} \) - measurable, and \( \mathcal{A} \) closed under countable unions. Then \( G \) is \( \mathcal{A} \) -measurable.
Proof. Let \( U \) be open in \( Y \) . Since \( Y \) is metrizable, \( U \) is an \( {F}_{\sigma } \) set in \( Y \) . Let \( U = \mathop{\bigcup }\limits_{n}{C}_{n},{C}_{n} \) closed. Then\n\n\[ \n{G}^{-1}\left( U\right) = \mathop{\bigcup }\limits_{n}{G}^{-1}\left( {C}_{n}\right) \n\]\n\nSince \( G \) is strongly \( \mathcal{A} \) -measurable and \( \mathcal{A} \) closed under countable unions, \( {G}^{-1}\left( U\right) \in \mathcal{A} \) .
Yes
Lemma 5.1.4 Suppose \( \left( {X,\mathcal{A}}\right) \) is a measurable space, \( Y \) a Polish space, and \( G : X \rightarrow Y \) a closed-valued measurable multifunction. Then \( \operatorname{gr}\left( G\right) \in \) \( \mathcal{A} \otimes {\mathcal{B}}_{Y} \)
Proof. Let \( \left( {U}_{n}\right) \) be a countable base for \( Y \) . Note that\n\n\[ y \notin G\left( x\right) \Leftrightarrow \exists n\left\lbrack {G\left( x\right) \bigcap {U}_{n} = \varnothing \& y \in {U}_{n}}\right\rbrack . \]\n\nTherefore,\n\n\[ \left( {X \times Y}\right) \smallsetminus \operatorname{gr}\left( G\right) = \mathop{\bigcup }\limits_{n}\left\lbrack {{\left( {G}^{-1}\left( {U}_{n}\right) \right) }^{c} \times {U}_{n}}\right\rbrack \]\n\nand the result follows.
Yes
Proposition 5.1.9 Suppose \( X \) is a Polish space and \( \mathbf{\Pi } \) a Borel equivalence relation on \( X \) . Then the following statements are equivalent.\n\n(i) \( \Pi \) has a Borel section.\n\n(ii) II admits a Borel cross section.
Proof. If \( f \) is a Borel section of \( \mathbf{\Pi } \), then the corresponding cross section is clearly Borel. On the other hand, let \( S \) be a Borel cross section of \( \mathbf{\Pi } \) . Let \( f\left( x\right) \) be the unique point of \( S \) equivalent to \( x \) . It is clearly a section of \( \mathbf{\Pi } \) . Note that\n\n\[ y = f\left( x\right) \Leftrightarrow x \coprod y\& y \in S. \]\n\nTherefore, as \( \mathbf{\Pi } \) and \( S \) are Borel, the graph of \( f \) is Borel. Hence, \( f \) is Borel measurable by 4.5.2.
Yes
Proposition 5.1.11 Every closed equivalence relation \( \mathbf{\Pi } \) on a Polish space \( X \) is countably separated.
Proof. Take any countable base \( \left( {U}_{n}\right) \) for the topology of \( X \) . For every \( x, y \) in \( X \) such that \( \left( {x, y}\right) \notin \mathbf{\Pi } \), there exist basic open sets \( {U}_{n} \) and \( {U}_{m} \) containing \( x \) and \( y \) respectively with \( {U}_{n} \times {U}_{m} \subseteq \left( {X \times Y}\right) \smallsetminus \Pi \) . In particular, \( {U}_{n}^{ * } \) is disjoint from \( {U}_{m} \) . Since \( {U}_{n}^{ * } \) is the projection onto the first coordinate axis of \( {\pi }_{X}\left( {\mathbf{\Pi } \cap \left( {X \times {U}_{n}}\right) }\right) \), which is Borel, \( {U}_{n}^{ * } \) is analytic. Thus \( {U}_{n}^{ * } \) is an invariant analytic set disjoint from \( {U}_{m} \) . Hence, by 4.4.5, there exists an invariant Borel set \( {B}_{n} \) containing \( {U}_{n}^{ * } \) and disjoint from \( {U}_{m} \) . It is now fairly easy to see that\n\n\[ \nX \times Y \smallsetminus \mathbf{\Pi } = \mathop{\bigcup }\limits_{n}\left( {{B}_{n} \times {B}_{n}^{c}}\right) \n\]\n\nThe result follows from 5.1.10.
Yes
Proposition 5.1.12 Every Borel measurable partition of a Polish space into \( {G}_{\delta } \) sets is countably separated.
Proof. Let \( X \) be a Polish space and \( \mathbf{\Pi } \) a Borel measurable partition of \( X \) into \( {G}_{\delta } \) sets. Take \( Y = F\left( X\right) \), the Effros Borel space of \( X \) . Then \( Y \) is standard Borel (3.3.10). For \( x \in X \), let \( \left\lbrack x\right\rbrack \) be the equivalence class containing \( x \) and \( p\left( x\right) = \operatorname{cl}\left( \left\lbrack x\right\rbrack \right) \) . For any open \( U \subseteq X \) ,\n\n\[ \n\{ x \in X : p\left( x\right) \bigcap U \neq \varnothing \} = {U}^{ * },\n\]\n\nwhich is Borel, since \( \mathbf{\Pi } \) is measurable. Therefore, \( p : X \rightarrow Y \) is Borel measurable (5.1.1). We now show that:\n\n\[ \nx \equiv y \Leftrightarrow p\left( x\right) = p\left( y\right) .\n\]\n\n\( \left( *\right) \)\n\nClearly, \( x \equiv y \Rightarrow p\left( x\right) = p\left( y\right) \) . Suppose \( x ≢ y \) but \( p\left( x\right) = p\left( y\right) \) . Then \( \left\lbrack x\right\rbrack \) and \( \left\lbrack y\right\rbrack \) are two disjoint dense \( {G}_{\delta } \) sets in \( p\left( x\right) \) . This contradicts the Baire category theorem, and we have proved \( \left( \star \right) \) . Thus, \( p : X \rightarrow Y \) witnesses the fact that \( \mathbf{\Pi } \) is countably separated.
Yes
Lemma 5.1.16 Let \( \Pi \) be a Borel partition of a Polish space \( X \) . The following statements are equivalent.\n\n(i) \( \Pi \) is countably separated.\n\n(ii) The \( \sigma \) -algebra \( {\mathcal{B}}^{ * } \) of \( \mathbf{\Pi } \) -invariant Borel sets is countably generated.
Proof. (i) implies (ii): Let \( \mathbf{\Pi } \) be countably separated. Take a Polish space \( Y \) and \( f : X \rightarrow Y \) a Borel map such that\n\n\[ x \coprod {x}^{\prime } \Leftrightarrow f\left( x\right) = f\left( {x}^{\prime }\right) .\n\]\n\nWe show that \( {\mathcal{B}}^{ * } = {f}^{-1}\left( {\mathcal{B}}_{Y}\right) \), which will then show that \( \mathbf{\Pi } \) satisfies (ii). Clearly, \( {\mathcal{B}}^{ * } \supseteq {f}^{-1}\left( {\mathcal{B}}_{Y}\right) \) . To prove the reverse inclusion, let \( B \subseteq X \) be an invariant Borel set. Then \( f\left( B\right) \) and \( f\left( {B}^{c}\right) \) are disjoint analytic subsets of \( Y \) . By the first separation principle for analytic sets (4.4.1), there is a Borel set \( C \) such that\n\n\[ f\left( B\right) \subseteq C\text{ and }C\bigcap f\left( {B}^{c}\right) = \varnothing .\n\]\n\nTherefore, \( B = {f}^{-1}\left( C\right) \in {f}^{-1}\left( {\mathcal{B}}_{Y}\right) \) . Hence,(i) implies (ii).\n\n(ii) implies (i): Let \( {\mathcal{B}}^{ * } \) be countably generated. Take any countable generator \( \left( {A}_{n}\right) \) of \( {\mathcal{B}}^{ * } \) . Note that the atoms of \( {\mathcal{B}}^{ * } \) are precisely the \( \mathbf{\Pi } \) - equivalence classes. Therefore, for any \( x,{x}^{\prime } \) in \( X \) ,\n\n\[ x \coprod {x}^{\prime } \Leftrightarrow \forall n\left( {x \in {A}_{n} \Leftrightarrow {x}^{\prime } \in {A}_{n}}\right) .\n\]\n\nFrom this and 5.1.10, it follows that (ii) implies (i).
Yes
Theorem 5.2.1 (Kuratowski and Ryll-Nardzewski [63]) Every \( {\mathcal{L}}_{\sigma } \) - measurable, closed-valued multifunction \( F : X \rightarrow Y \) admits an \( {\mathcal{L}}_{\sigma } \) - measurable selection.
Proof of 5.2.1. Inductively we define a sequence \( \left( {s}_{n}\right) \) of \( {\mathcal{L}}_{\sigma } \) -measurable maps from \( X \) to \( Y \) such that for every \( x \in X \) and every \( n \in \mathbb{N} \) ,\n\n(i) \( d\left( {{s}_{n}\left( x\right), F\left( x\right) }\right) < {2}^{-n} \), and\n\n(ii) \( d\left( {{s}_{n}\left( x\right) ,{s}_{n + 1}\left( x\right) }\right) < {2}^{-n} \) .\n\nTo define \( \left( {s}_{n}\right) \) we take a countable dense set \( \left( {r}_{n}\right) \) in \( Y \) . Define \( {s}_{0} \equiv {r}_{0} \) . Let \( n > 0 \) . Suppose that for every \( m < n,{s}_{m} \) satisfying conditions (i) and (ii) have been defined. Let\n\n\[ {E}_{k} = \left\{ {x \in X : d\left( {{s}_{n - 1}\left( x\right) ,{r}_{k}}\right) < {2}^{-n + 1}\& d\left( {{r}_{k}, F\left( x\right) }\right) < {2}^{-n}}\right\} .\n\]\n\nSo,\n\n\[ {E}_{k} = {s}_{n - 1}^{-1}\left( {B\left( {{r}_{k},{2}^{-n + 1}}\right) }\right) \bigcap {F}^{-1}\left( {B\left( {{r}_{k},{2}^{-n}}\right) }\right) ,\n\]\n\nwhere \( B\left( {y, r}\right) \) denotes the open ball in \( Y \) with center \( y \) and radius \( r \) . It follows that \( {E}_{k} \in {\mathcal{L}}_{\sigma } \) .\n\nFurther, \( \mathop{\bigcup }\limits_{k}{E}_{k} = X \) . To see this, take any \( x \in X \) . As \( d\left( {{s}_{n - 1}\left( x\right), F\left( x\right) }\right) < \) \( {2}^{-n + 1} \), there is a \( y \) in \( F\left( X\right) \) such that \( d\left( {y,{s}_{n - 1}\left( x\right) }\right) < {2}^{-n + 1} \) . Since \( \left( {r}_{k}\right) \) is dense, there exists an \( l \) such that \( d\left( {{r}_{l}, y}\right) < {2}^{-n} \) and \( d\left( {{r}_{l},{s}_{n - 1}\left( x\right) }\right) < {2}^{-n + 1} \) . Then \( x \in {E}_{l} \) .\n\nBy 5.2.2, there exist pairwise disjoint sets \( {D}_{k} \subseteq {E}_{k} \) in \( {\mathcal{L}}_{\sigma } \).
Yes
Lemma 5.2.2 Suppose \( {A}_{n} \in {\mathcal{L}}_{\sigma } \). Then there exist \( {B}_{n} \subseteq {A}_{n} \) such that the \( {B}_{n} \)’s are pairwise disjoint elements of \( {\mathcal{L}}_{\sigma } \) and \( \mathop{\bigcup }\limits_{n}{A}_{n} = \mathop{\bigcup }\limits_{n}{B}_{n} \).
Proof. Write\n\n\[ \n{A}_{n} = \mathop{\bigcup }\limits_{m}{C}_{nm} \n\] \n\n\( {C}_{nm} \in \mathcal{L} \). Enumerate \( \left\{ {{C}_{nm} : n, m \in \mathbb{N}}\right\} \) in a single sequence, say \( \left( {D}_{i}\right) \). Set\n\n\[ \n{E}_{i} = {D}_{i} \smallsetminus \mathop{\bigcup }\limits_{{j < i}}{D}_{j} \n\] \n\n\nClearly, \( {E}_{i} \in \mathcal{L} \). Take\n\n\[ \n{B}_{n} = \bigcup \left\{ {{E}_{i} : {E}_{i} \subseteq {A}_{n}\& \left( {\forall m < n}\right) \left( {{E}_{i} \nsubseteq {A}_{m}}\right) }\right\} .\n\]
Yes
Lemma 5.2.3 Suppose \( {f}_{n} : X \rightarrow Y \) is a sequence of \( {\mathcal{L}}_{\sigma } \) -measurable functions converging uniformly to \( f : X \rightarrow Y \) . Then \( f \) is \( {\mathcal{L}}_{\sigma } \) -measurable.
Proof. Replacing \( \left( {f}_{n}\right) \) by a subsequence if necessary, we assume that\n\n\[ \forall x\forall n\left( {d\left( {f\left( x\right) ,{f}_{n}\left( x\right) }\right) < 1/\left( {n + 1}\right) }\right) .\n\]\n\nLet \( F \) be a closed set in \( Y \) and\n\n\[ {F}_{n} = \operatorname{cl}\left( {\{ y \in Y : d\left( {y, F}\right) < 1/\left( {n + 1}\right) \} }\right) .\n\]\n\nThen\n\n\[ f\left( x\right) \in F \Leftrightarrow \forall n{f}_{n}\left( x\right) \in {F}_{n} \]\n\ni.e., \( {f}^{-1}\left( F\right) = \mathop{\bigcap }\limits_{n}{f}_{n}^{-1}\left( {F}_{n}\right) \in {\mathcal{L}}_{\delta } \), and our result is proved.
Yes
Corollary 5.2.4 Let \( X \) be a Polish space and \( F\left( X\right) \) the space of nonempty closed subsets of \( X \) with Effros Borel structure. Then there is a measurable \( s : F\left( X\right) \rightarrow X \) such that \( s\left( F\right) \in F \) for all \( F \in F\left( X\right) \) .
Proof. Apply 5.2.1 to the multifunction \( G : F\left( X\right) \rightarrow X \), where \( G\left( F\right) = \) \( F \), with \( \mathcal{L} \) the Effros Borel \( \sigma \) -algebra on \( F\left( X\right) \) .
Yes
Corollary 5.2.5 Let \( \\left( {T,\\mathcal{T}}\\right) \) be a measurable space and \( Y \) a separable metric space. Then every \( \\mathcal{T} \) -measurable, compact-valued multifunction \( F \) : \( T \\rightarrow Y \) admits a \( \\mathcal{T} \) -measurable selection.
Proof. Let \( X \) be the completion of \( Y \) . Then \( F \) as a multifunction from \( T \) to \( X \) is closed-valued and \( \\mathcal{T} \) -measurable. Apply 5.2.1 now.
No
Corollary 5.2.6 Suppose \( Y \) is a compact metric space, \( X \) a metric space, and \( f : Y \rightarrow X \) a continuous onto map. Then there is a Borel map \( s : X \rightarrow Y \) of class 2 such that \( f \circ s \) is the identity map on \( X \) .
Proof. Let \( F\left( x\right) = {f}^{-1}\left( x\right), x \in X \), and \( \mathcal{L} = {\mathbf{\Delta }}_{2}^{0} \) . For any closed set \( C \) in \( Y \) ,\n\n\[ \n{F}^{-1}\left( C\right) = {\pi }_{X}\left( {\operatorname{graph}\left( f\right) \bigcap \left( {X \times C}\right) }\right) .\n\]\n\nTherefore, by \( {2.3.24},{F}^{-1}\left( C\right) \) is closed. Hence, \( F \) is \( {\mathcal{L}}_{\sigma } \) -measurable. Now apply 5.2.1.
Yes
Proposition 5.2.7 (A. Maitra and B. V. Rao[77]) Let \( T \) be a nonempty set, \( \mathcal{L} \) an algebra on \( T \), and \( X \) a Polish space. Suppose \( F : T \rightarrow X \) is a closed-valued \( {\mathcal{L}}_{\sigma } \) -measurable multifunction. Then there is a sequence \( \left( {f}_{n}\right) \) of \( {\mathcal{L}}_{\sigma } \) -measurable selections of \( F \) such that\n\n\[ F\left( t\right) = \operatorname{cl}\left( \left\{ {{f}_{n}\left( t\right) : n \in \mathbb{N}}\right\} \right) ,\;t \in T. \]
Proof. Fix a countable base \( \left\{ {{U}_{n} : n \in \mathbb{N}}\right\} \) for the topology of \( X \) and fix also an \( {\mathcal{L}}_{\sigma } \) -measurable selection \( f \) for \( F \) . For each \( n,{T}_{n} = {F}^{-1}\left( {U}_{n}\right) \in {\mathcal{L}}_{\sigma } \) . Write \( {T}_{n} = \mathop{\bigcup }\limits_{m}{T}_{nm},{T}_{nm} \in \mathcal{L} \) . Define \( {F}_{nm} : {T}_{nm} \rightarrow X \) by\n\n\[ {F}_{nm}\left( t\right) = \operatorname{cl}\left( {F\left( t\right) \bigcap {U}_{n}}\right) ,\;t \in {T}_{nm}. \]\n\nBy 5.2.1, there is an \( {\mathcal{L}}_{\sigma } \mid {T}_{nm} \) -measurable selection \( {h}_{nm} \) for \( {F}_{nm} \) . Define\n\n\[ {f}_{nm}\left( t\right) = \left\{ \begin{array}{ll} {h}_{nm}\left( t\right) & \text{ if }t \in {T}_{nm}, \\ f\left( t\right) & \text{ otherwise. } \end{array}\right. \]\n\nThen each \( {f}_{nm} \) is \( {\mathcal{L}}_{\sigma } \) -measurable. Further,\n\n\[ F\left( t\right) = \operatorname{cl}\left\{ {{f}_{nm}\left( t\right) : n, m \in \mathbb{N}}\right\} ,\;t \in T. \]
Yes
Theorem 5.2.8 (Srivastava[115]) Let \( T,\mathcal{L}, X \), and \( F \) be as in 5.2.7. Then there is a map \( f : T \times {\mathbb{N}}^{\mathbb{N}} \rightarrow X \) such that\n\n(i) for every \( \alpha \in {\mathbb{N}}^{\mathbb{N}}, t \rightarrow f\left( {t,\alpha }\right) \) is \( {\mathcal{L}}_{\sigma } \) -measurable, and\n\n(ii) for every \( t \in T, f\left( {t, \cdot }\right) \) is a continuous map from \( {\mathbb{N}}^{\mathbb{N}} \) onto \( F\left( t\right) \) .
Proof of 5.2.8 Fix a complete compatible metric \( d \) on \( X \) . Applying 5.2.9 and 5.2.7 repeatedly, for each \( s \in {\mathbb{N}}^{ < \mathbb{N}} \), we get an \( {\mathcal{L}}_{\sigma } \) -measurable selection \( {f}_{s} : T \rightarrow X \) for \( F \) satisfying the following condition: For every \( s \in {\mathbb{N}}^{ < \mathbb{N}} \) and every \( t \in T,\left\{ {{f}_{{s}^{ \frown }n}\left( t\right) : n \in \mathbb{N}}\right\} \) is dense in \( F\left( t\right) \bigcap B\left( {{f}_{s}\left( t\right) ,1/{2}^{-\left| s\right| }}\right) \) ). Note that for every \( \alpha \in {\mathbb{N}}^{\mathbb{N}} \) and every \( t \in T \), the sequence \( \left( {{f}_{\alpha \mid n}\left( t\right) }\right) \) is Cauchy and hence convergent. Take \( f : T \times {\mathbb{N}}^{\mathbb{N}} \rightarrow X \) defined by\n\n\[ f\left( {t,\alpha }\right) = \mathop{\lim }\limits_{n}{f}_{\alpha \mid n}\left( t\right) \]
Yes
Theorem 5.2.10 (S. Bhattacharya and S. M. Srivastava [12]) Let \( F \) : \( X \rightarrow Y \) be closed-valued and strongly \( {\mathcal{L}}_{\sigma } \) -measurable. Suppose \( Z \) is a separable metric space and \( g : Y \rightarrow Z \) a Borel map of class 2 . Then there is an \( {\mathcal{L}}_{\sigma } \) -measurable selection \( f \) of \( F \) such that \( g \circ f \) is \( {\mathcal{L}}_{\sigma } \) -measurable.
Proof. Let \( \left( {U}_{n}\right) \) be a countable base for the topology of \( Z \) . Write \( {g}^{-1}\left( {U}_{n}\right) = \mathop{\bigcup }\limits_{m}{H}_{nm} \), the \( {H}_{nm} \) ’s closed. Also, take a countable base \( \left( {W}_{n}\right) \) for \( Y \) and write \( {W}_{n} = \mathop{\bigcup }\limits_{m}{C}_{nm} \), the \( {C}_{nm} \) ’s closed. Let \( \mathcal{B} \) be the smallest family of subsets of \( Y \) closed under finite intersections, containing each \( {H}_{nm} \) and each \( {C}_{nm} \) . Let \( {\mathcal{T}}^{\prime } \) be the topology on \( Y \) with \( \mathcal{B} \) a base. Note that \( {\mathcal{T}}^{\prime } \) is finer than \( \mathcal{T} \), the original topology of \( Y \) . By Observations 1 and 2 of Section 2, Chapter 3, we see that \( {\mathcal{T}}^{\prime } \) is a Polish topology on \( Y \) . Note that with \( Y \) equipped with the topology \( {\mathcal{T}}^{\prime }, g \) is continuous and \( F{\mathcal{L}}_{\sigma } \) - measurable. By 5.2.1, there is an \( {\mathcal{L}}_{\sigma } \) -measurable selection \( f \) of \( F \) . This \( f \) works.
Yes
Theorem 5.2.11 Let \( X, Y \) be compact metric spaces, \( f : X \rightarrow Y \) a continuous onto map. Suppose \( A \subseteq Y \) and \( 1 \leq \alpha < {\omega }_{1} \). Then\n\n\[ \n{f}^{-1}\left( A\right) \in {\mathbf{\Pi }}_{\alpha }^{0}\left( X\right) \Leftrightarrow A \in {\mathbf{\Pi }}_{\alpha }^{0}\left( Y\right)\n\]
Proof of 5.2.11 We need to prove the \
No
Lemma 5.2.12 Let \( X, Y \), and \( f \) be as in 5.2.11. Suppose \( 1 \leq \alpha < {\omega }_{1}, Z \) is a separable metric space, and \( g : X \rightarrow Z \) is a Borel map of class \( \alpha \) . Then there is a class 2 map \( s : Y \rightarrow X \) such that \( g \circ s \) is of class \( \alpha \) and \( f\left( {s\left( y\right) }\right) = y \) for all \( y \) .
Proof. Let \( F\left( y\right) = {f}^{-1}\left( y\right), y \in Y \) . Then \( F : Y \rightarrow X \) is an upper-semicontinuous closed-valued multifunction. By 5.2.1 there is a selection \( s \) of \( F \) that is Borel of class 2 . This \( s \) works if either \( \alpha = 1 \) (i.e., if \( g \) is continuous) or if \( \alpha \geq {\omega }_{0} \) (in this case \( g \circ s \) is of class \( 1 + \alpha = \alpha \) ). So, we need to prove the result for \( 2 \leq \alpha < {\omega }_{0} \) only. We prove this by induction on \( \alpha \) .\n\nFor \( \alpha = 2 \) we get this by 5.2.10. Let \( n \geq 2 \), and the result is true for \( \alpha = n \) . Let \( g : X \rightarrow Z \) be of class \( n + 1 \) . By 3.6.15, there is a sequence \( \left( {g}_{n}\right) \) of Borel measurable functions from \( X \) to \( Z \) of class \( n \) converging pointwise to \( g \) . By 3.6.5, \( h = \left( {g}_{n}\right) : X \rightarrow {Z}^{\mathbb{N}} \) is of class \( n \) . So, by the induction hypothesis, there is a selection \( s \) of \( F \) of class 2 such that \( h \circ s \) is of class \( n \) . In particular, each \( {g}_{n} \circ s \) is of class \( n \) . As \( {g}_{n} \circ s \rightarrow g \circ s \) pointwise, \( g \circ s \) is of class \( \left( {n + 1}\right) \) by 3.6.5.
Yes
Theorem 5.3.1 (Schäl) Suppose \( \\left( {T,\\mathcal{T}}\\right) \) is a measurable space and let \( Y \) be a separable metric space. Suppose \( G : T \\rightarrow Y \) is a \( \\mathcal{T} \) -measurable compact-valued multifunction. Let \( v \) be a real-valued function on \( \\operatorname{gr}\\left( G\\right) \) , the graph of \( G \), that is the pointwise limit of a nonincreasing sequence \( \\left( {v}_{n}\\right) \) of \( \\mathcal{T} \\otimes {\\mathcal{B}}_{Y} \\mid \\operatorname{gr}\\left( G\\right) \) -measurable functions on \( \\operatorname{gr}\\left( G\\right) \) such that for each \( n \) and each \( t \\in T,{v}_{n}\\left( {t,\\text{.}}\\right) {is}\;{continuous}\;{on}\;G\\left( t\\right) .{Let} \n\n\[ \n{v}^{ * }\\left( t\\right) = \\sup \\{ v\\left( {t, y}\\right) : y \\in G\\left( t\\right) \\} ,\\;t \\in T. \n\] \n\nThen there is a \( \\mathcal{T} \) -measurable selection \( g : T \\rightarrow Y \) for \( G \) such that \n\n\[ \n{v}^{ * }\\left( t\\right) = v\\left( {t, g\\left( t\\right) }\\right) \n\] \n\nfor every \( t \\in T \) .
Proof of 5.3.1. (Burgess and Maitra[24]) Without any loss of generality we assume that \( Y \) is Polish. Fix a complete metric \( d \) on \( Y \) compatible with its topology. By 5.2.7, we get \( \\mathcal{T} \) -measurable selections \( {g}_{n} : T \\rightarrow Y \) of \( G \) such that \n\n\[ \nG\\left( t\\right) = \\operatorname{cl}\\left( \\left\\{ {{g}_{n}\\left( t\\right) : n \\in \\mathbb{N}}\\right\\} \\right) ,\\;t \\in T. \n\] \n\nThen \( {v}^{ * }\\left( t\\right) = \\sup \\left\\{ {v\\left( {t,{g}_{n}\\left( t\\right) }\\right) : n \\in \\mathbb{N}}\\right\\} \) . Hence, \( {v}^{ * } \) is \( \\mathcal{T} \) -measurable.\n\nWe first prove the result when \( v \) is \( \\mathcal{T} \\otimes {\\mathcal{B}}_{Y} \\mid \\operatorname{gr}\\left( G\\right) \) -measurable with \( v\\left( {t,\\text{.}}\\right) \) continuous for all \( t \) . Set \n\n\[ \nH\\left( t\\right) = \\left\\{ {y \\in G\\left( t\\right) : v\\left( {t, y}\\right) = {v}^{ * }\\left( t\\right) }\\right\\} ,\\;t \\in T. \n\] \n\nClearly, \( H \) is a compact-valued multifunction. Let \( C \) be any closed set in \( Y \) and let \n\n\[ \n{C}_{n} = \\{ y \\in Y : d\\left( {y, C}\\right) < 1/n\\} ,\\;n \\geq 1. \n\] \n\nWe easily check that \n\n\[ \n\\{ t : H\\left( t\\right) \\cap C \\neq \\varnothing \\} = \\mathop{\\bigcap }\\limits_{n}\\mathop{\\bigcup }\\limits_{i}\\left\\{ {t : v\\left( {t,{g}_{i}\\left( t\\right) }\\right) > {v}^{ * }\\left( t\\right) - 1/n\\text{ and }{g}_{i}\\left( t\\right) \\in {C}_{n}}\\right\\} . \n\] \n\nIt follows that \( H \) is \( \\mathcal{T} \) -measurable. To complete the proof in the special case, apply the Kuratowski and Ryll-Nardzewski selection theorem (5.2.1) and take any \( \\mathcal{T} \) -measurable selection \( g \) for \( H \) .\n\nWe now turn to the general case. By the above case, for each \( n \) there is a \( \\mathcal{T} \) -measurable selection \( {g}_{n} : T \\rightarrow Y \) of \( G \) such that \n\n\[ \n{v}_{n}\\left( {t,{g}_{n}\\left( t\\right) }\\right) = \\sup \\left\\{ {{v}_{n}\\left( {t, y}\\right) : y \\in G\\left( t\\right) }\\right\\} \n\] \n\nfor every \( t \\in T \) . For \( t \\in T \), set \n\n\( H\\left( t\\right) = \\left\\{ {y \\in G\\left( t\\right) : \\text{ there is a subsequence }\\left( {{g}_{{n}_{i}}\\left( t\\right) }\\right) \\text{ such that }{g}_{{n}_{i}}\\left( t\\right) \\rightarrow y}\\right\\} . \n\nSince \( G\\left( t\\right) \) is nonempty and compact, so is \( H\\left( t\\right) \) . We now show that \( H \) is \( \\mathcal{T} \) -measurable. Let \( C \) be closed in \( Y \) . Then \n\n\[ \n\\{ t \\in T : H\\left( t\\right) \\bigcap C \\neq \\varnothing \\} = \\mathop{\\bigcap }\\limits_{{k \\geq 1}}\\mathop{\\bigcup }\\limits_{{m > k}}\\left\\{ {t \\in T : d\\left( {{g}_{m}\\left( t\\right), C}\\right) < 1/k}\\right\\} . \n\] \n\nIt follows that \( H \) is \( \\mathcal{T} \) -measurable. By the Kur
Yes
It is not unreasonable to conjecture that 5.3.1 remains true even for \( v \) that are \( \mathcal{T}\bigotimes {\mathcal{B}}_{Y} \mid {gr}\left( G\right) \) -measurable such that \( v\left( {t,\text{.}}\right) {isupper} \) semicontinuous for every \( t \) . However, this is not true.
Recall that in the last chapter, using Solovay's coding of Borel sets, we showed that there is a coanalytic set \( T \) and a function \( g : T \rightarrow {2}^{\mathbb{N}} \) whose graph is relatively Borel in \( T \times {2}^{\mathbb{N}} \) but that is not Borel measurable. Take \( \mathcal{T} = {\mathcal{B}}_{T}, G\left( t\right) = {2}^{\mathbb{N}} \) \( \left( {t \in T}\right) \), and \( v : T \times {2}^{\mathbb{N}} \rightarrow \mathbb{R} \) the characteristic function of \( \operatorname{graph}\left( g\right) \) .
Yes
Theorem 5.4.1 (Effros [40]) Every lower-semicontinuous or upper-semicontinuous partition \( \mathbf{\Pi } \) of a Polish space \( X \) into closed sets admits a Borel measurable section \( f : X \rightarrow X \) of class 2. In particular, they admit a \( {G}_{\delta } \) cross section.
Proof. In 5.2.1, take \( Y = X,\mathcal{L} \) the family of invariant sets that are simultaneously \( {F}_{\sigma } \) and \( {G}_{\delta } \), and \( F\left( x\right) = \left\lbrack x\right\rbrack \), the equivalence class containing \( x \) . So, there is an \( {\mathcal{L}}_{\sigma } \) -measurable selection \( f : X \rightarrow X \) of \( F \) . This means that \( f \) is a Borel measurable section of \( \mathbf{\Pi } \) of class 2 . The corresponding cross section \( S = \{ x \in X : x = f\left( x\right) \} \) is a \( {G}_{\delta } \) cross section of \( \Pi \) .
Yes
Theorem 5.4.2 (Effros - Mackey cross section theorem) Suppose \( H \) is a closed subgroup of a Polish group \( G \) and \( \mathbf{\Pi } \) the partition of \( G \) consisting of all the right cosets of \( H \) . Then \( \mathbf{\Pi } \) admits a Borel measurable section of class 2. In particular, it admits a \( {G}_{\delta } \) cross section.
Proof. Note that for any open set \( U \) in \( G \) ,\n\n\[ \n{U}^{ * } = \bigcup \{ g \cdot U : g \in H\} . \n\]\n\nSo, \( {U}^{ * } \) is open. Thus \( \mathbf{\Pi } \) is lower semicontinuous. The result follows from Effros's cross section theorem (5.4.1).
No
Theorem 5.4.3 Every Borel measurable partition \( \mathbf{\Pi } \) of a Polish space \( X \) into closed sets admits a Borel measurable section \( f : X \rightarrow X \) . In particular, it admits a Borel cross section.
Proof. Let \( \mathcal{A} \) be the \( \sigma \) -algebra of all invariant Borel subsets of \( X \) and \( F : X \rightarrow X \) the multifunction that assigns to each \( x \in X \) the member of \( \mathbf{\Pi } \) containing \( x \) . By our assumptions, \( F \) is \( \mathcal{A} \) -measurable. By 5.2.1, we get a measurable selection \( f \) for \( F \) . Note that \( f \) is a section of \( \mathbf{\Pi } \) . The corresponding cross section \( S = \{ x \in X : x = f\left( x\right) \} \) of \( \mathbf{\Pi } \) is clearly a Borel cross section of \( \Pi \) .
Yes
Theorem 5.4.4 The classification space \( \operatorname{irr}\left( n\right) / \sim \) is standard Borel.
Proof. Fix any irreducible \( A \) . Then the \( \sim \) -equivalence class \( \left\lbrack A\right\rbrack \) containing \( A \) equals\n\n\[ \n{\pi }_{1}\left\{ {\left( {B, U}\right) \in \operatorname{irr}\left( n\right) \times U\left( n\right) : A = {UB}{U}^{ * }}\right\} , \n\]\n\nwhere \( {\pi }_{1} : \operatorname{irr}\left( n\right) \times U\left( n\right) \rightarrow \operatorname{irr}\left( n\right) \) is the projection map to the first coordinate space. (Recall that \( U\left( n\right) \) denotes the set of all \( n \times n \) unitary matrices.) As the set\n\n\[ \n\left\{ {\left( {B, U}\right) \in \operatorname{irr}\left( n\right) \times U\left( n\right) : A = {UB}{U}^{ * }}\right\} \n\]\n\nis closed and \( U\left( n\right) \) compact, \( \left\lbrack A\right\rbrack \) is closed by 2.3.24.\n\nNow let \( \mathcal{O} \) be any open set in \( \operatorname{irr}\left( n\right) \) . Its saturation is\n\n\[ \n\mathop{\bigcup }\limits_{{U \in U\left( n\right) }}\left\{ {A \in \operatorname{irr}\left( n\right) : {UA}{U}^{ * } \in \mathcal{O}}\right\} \n\]\n\nwhich is open. Thus \( \sim \) is a lower-semicontinuous partition of \( \operatorname{irr}\left( n\right) \) into closed sets. By 5.4.3, let \( C \) be a Borel cross section of \( \sim \) . Then \( q \mid C \) is a one-to-one Borel map from \( C \) onto \( \operatorname{irr}\left( n\right) / \sim \), where \( q : \operatorname{irr}\left( n\right) \rightarrow \operatorname{irr}\left( n\right) / \sim \) is the canonical quotient map. By the Borel isomorphism theorem (3.3.13), \( q \) is a Borel isomorphism, and our result is proved.
Yes
Theorem 5.4.5 (Miller[84]) Let \( \\left( {G, \\cdot }\\right) \) be a Polish group, \( X \) a Polish space, and \( a\\left( {g, x}\\right) = g \\cdot x \) an action of \( G \) on \( X \) . Suppose for a given \( x \\in X \) that \( g \\rightarrow g \\cdot x \) is Borel. Then the orbit\n\n\\[ \n\\{ g \\cdot x : g \\in G\\} \n\\]\n\nof \( x \) is Borel.
Proof. Let \( H = {G}_{x} \) be the stabilizer of \( x \) . By 4.8.4, \( H \) is closed in \( G \) . Let \( S \) be a Borel cross section of the partition \( \\mathbf{\\Pi } \) consisting of the left cosets of \( H \) . The map \( g \\rightarrow g \\cdot x \) restricted to \( S \) is one-to-one, Borel, and onto the orbit of \( x \) . The result follows from 4.5.4.
Yes
Proposition 5.5.1 Let \( X, Y \) be Polish spaces, \( B \subseteq X \times Y \) Borel, and \( C \) an analytic uniformization of \( B \) . Then \( C \) is Borel.
Proof. We show that \( C \) is also coanalytic. The result will then follow from Souslin’s theorem. That \( C \) is coanalytic follows from the following relation:\n\n\[ \left( {x, y}\right) \in C \Leftrightarrow \left( {x, y}\right) \in B\& \forall z\left( {\left( {x, z}\right) \in C \Rightarrow y = z}\right) . \]
No
Theorem 5.5.2 (Von Neumann[124]) Let \( X \) and \( Y \) be Polish spaces, \( A \subseteq \) \( X \times Y \) analytic, and \( \mathcal{A} = \sigma \left( {{\mathbf{\sum }}_{1}^{1}\left( X\right) }\right) \), the \( \sigma \) -algebra generated by the analytic subsets of \( X \) . Then there is an \( \mathcal{A} \) -measurable section \( u : {\pi }_{X}\left( A\right) \rightarrow Y \) of A.
Proof. Let \( f : {\mathbb{N}}^{\mathbb{N}} \rightarrow A \) be a continuous surjection. Consider\n\n\[ B = \left\{ {\left( {x,\alpha }\right) \in X \times {\mathbb{N}}^{\mathbb{N}} : {\pi }_{X}\left( {f\left( \alpha \right) }\right) = x}\right\} . \]\n\nThen \( B \) is a closed set with \( {\pi }_{X}\left( B\right) = {\pi }_{X}\left( A\right) \) . For \( x \in {\pi }_{X}\left( A\right) \), define \( g\left( x\right) \) to be the lexicographic minimum of \( {B}_{x} \) ; i.e.,\n\n\[ g\left( x\right) = \alpha \Leftrightarrow \left( {x,\alpha }\right) \in B \]\n\n\[ \& \forall \beta \{ \left( {x,\beta }\right) \in B \Rightarrow \]\n\n\[ \exists n\left\lbrack {\alpha \left( n\right) < \beta \left( n\right) \text{ and }\forall m < n\left( {\alpha \left( m\right) = \beta \left( m\right) }\right) }\right\rbrack \} . \]\n\nBy induction on \( \left| s\right| \), we prove that \( {g}^{-1}\left( {\sum \left( s\right) }\right) \in \mathcal{A} \) for every \( s \in {\mathbb{N}}^{ < \mathbb{N}} \) . Since \( \left\{ {\sum \left( s\right) : s \in {\mathbb{N}}^{ < \mathbb{N}}}\right\} \) is a base for \( {\mathbb{N}}^{\mathbb{N}} \), it follows that \( g \) is \( \mathcal{A} \) -measurable. Suppose \( {g}^{-1}\left( {\sum \left( t\right) }\right) \in \mathcal{A} \) and \( s = t \hat{} k, k \in \mathbb{N} \) . Then for any \( x \),\n\n\[ x \in {g}^{-1}\left( {\sum \left( s\right) }\right) \Leftrightarrow x \in {g}^{-1}\left( {\sum \left( t\right) }\right) \]\n\n\[ \& \exists \alpha \left( {\left( {x,\alpha }\right) \in B\& s \prec \alpha }\right) \]\n\n\[ \text{&}\forall l < k\neg \exists \beta \left( {\left( {x,\beta }\right) \in B\& t \hat{} l \prec \beta }\right) \text{.} \]\n\nHence, \( {g}^{-1}\left( {\sum \left( s\right) }\right) \in \mathcal{A} \) . Now, define \( u\left( x\right) = {\pi }_{Y}\left( {f\left( {g\left( x\right) }\right) }\right), x \in {\pi }_{X}\left( A\right) \) . Then \( u \) is an \( \mathcal{A} \) -measurable section of \( A \) .
Yes
Theorem 5.5.3 Every analytic subset \( A \) of the product of Polish spaces \( X, Y \) admits a section \( u \) that is universally measurable as well as Baire measurable.
Proof. The result follows from 5.5.2, 4.3.1, and 4.3.2.
No
Proposition 5.5.4 In 5.5.3, further assume that \( A \) is Borel. Then the graph of the section \( u \) is coanalytic.
Proof. Note that\n\n\[ \begin{matrix} u\left( x\right) = y & \Leftrightarrow & \left( {x, y}\right) \in A\;\& \;\left( {\forall \alpha \in {\mathbb{N}}^{\mathbb{N}}}\right) \left( {\forall \beta \in {\mathbb{N}}^{\mathbb{N}}}\right) (\lbrack \left( {x,\alpha }\right) \in B \end{matrix}\n\n\[ \left. {\& \left( {x,\beta }\right) \in B\& f\left( \alpha \right) = \left( {x, y}\right) \rbrack \; \Rightarrow \alpha { \leq }_{\text{lex }}\beta }\right) ,\n\nwhere \( { \leq }_{\text{lex }} \) is the lexicographic ordering on \( B \) .
No
Theorem 5.5.7 Let \( \left( {X,\mathcal{E}}\right) \) be a measurable space with \( \mathcal{E} \) closed under the Souslin operation, \( Y \) a Polish space, and \( A \in \mathcal{E}\bigotimes {\mathcal{B}}_{Y} \) . Then \( {\pi }_{X}\left( A\right) \in \mathcal{E} \) , and there is an \( \mathcal{E} \) -measurable section of \( A \) .
Proof. By 3.1.7, there exists a countable sub \( \sigma \) -algebra \( \mathcal{D} \) of \( \mathcal{E} \) such that \( A \in \mathcal{D}\bigotimes {\mathcal{B}}_{Y} \) . Let \( \left( {B}_{n}\right) \) be a countable generator of \( \mathcal{D} \) and \( \chi : X \rightarrow \mathcal{C} \) the map defined by\n\n\[ \chi \left( x\right) = \left( {{\chi }_{{B}_{0}}\left( x\right) ,{\chi }_{{B}_{1}}\left( x\right) ,{\chi }_{{B}_{2}}\left( x\right) ,\ldots }\right) ,\;x \in X. \]\n\nLet \( Z = \chi \left( X\right) \) . Then \( \chi \) is a bimeasurable map from \( \left( {X,\mathcal{D}}\right) \) onto \( \left( {Z,{\mathcal{B}}_{Z}}\right) \) .\n\nLet\n\n\[ B = \{ \left( {\chi \left( x\right), y}\right) \in Z \times Y : \left( {x, y}\right) \in A\} . \]\n\n\( B \) is Borel in \( Z \times Y \) . Take a Borel set \( C \) in \( \mathcal{C} \times Y \) such that \( B = C\bigcap \left( {Z \times Y}\right) \) .\n\nLet \( E = {\pi }_{\mathcal{C}}\left( C\right) \) . Then \( E \) is analytic, and therefore it is the result of the Souslin operation on a system \( \left\{ {{E}_{s} : s \in {\mathbb{N}}^{ < \mathbb{N}}}\right\} \) of Borel subsets of \( \mathcal{C} \) . Note that\n\n\[ {\pi }_{X}\left( A\right) = {\chi }^{-1}\left( E\right) = \mathcal{A}\left( {{\chi }^{-1}\left( \left\{ {E}_{s}\right\} \right) }\right) . \]\n\nSince \( \mathcal{E} \) is closed under the Souslin operation, \( {\pi }_{X}\left( A\right) \in \mathcal{E} \) .\n\nBy 5.5.2, there is a \( \sigma \left( {{\mathbf{\sum }}_{1}^{1}\left( \mathcal{C}\right) }\right) \) -measurable section \( v : E \rightarrow Y \) of \( C \) . Take \( f = v \circ \chi \) . Then \( f \) is an \( \mathcal{E} \) -measurable section of \( A \) .
Yes
Theorem 5.5.7 Let \( \\left( {X,\\mathcal{E}}\\right) \) be a measurable space with \( \\mathcal{E} \) closed under the Souslin operation, \( Y \) a Polish space, and \( A \\in \\mathcal{E}\\bigotimes {\\mathcal{B}}_{Y} \) . Then \( {\\pi }_{X}\\left( A\\right) \\in \\mathcal{E} \) , and there is an \( \\mathcal{E} \) -measurable section of \( A \) .
Proof. By 3.1.7, there exists a countable sub \( \\sigma \) -algebra \( \\mathcal{D} \) of \( \\mathcal{E} \) such that \( A \\in \\mathcal{D}\\bigotimes {\\mathcal{B}}_{Y} \) . Let \( \\left( {B}_{n}\\right) \) be a countable generator of \( \\mathcal{D} \) and \( \\chi : X \\rightarrow \\mathcal{C} \) the map defined by\n\n\[ \n\\chi \\left( x\\right) = \\left( {{\\chi }_{{B}_{0}}\\left( x\\right) ,{\\chi }_{{B}_{1}}\\left( x\\right) ,{\\chi }_{{B}_{2}}\\left( x\\right) ,\\ldots }\\right) ,\\;x \\in X.\n\]\n\nLet \( Z = \\chi \\left( X\\right) \) . Then \( \\chi \) is a bimeasurable map from \( \\left( {X,\\mathcal{D}}\\right) \) onto \( \\left( {Z,{\\mathcal{B}}_{Z}}\\right) \) .\n\nLet\n\n\[ \nB = \\{ \\left( {\\chi \\left( x\\right), y}\\right) \\in Z \\times Y : \\left( {x, y}\\right) \\in A\\} .\n\]\n\n\( B \) is Borel in \( Z \\times Y \) . Take a Borel set \( C \) in \( \\mathcal{C} \\times Y \) such that \( B = C\\bigcap \\left( {Z \\times Y}\\right) \) .\n\nLet \( E = {\\pi }_{\\mathcal{C}}\\left( C\\right) \) . Then \( E \) is analytic, and therefore it is the result of the Souslin operation on a system \( \\left\\{ {{E}_{s} : s \\in {\\mathbb{N}}^{ < \\mathbb{N}}}\\right\\} \) of Borel subsets of \( \\mathcal{C} \) . Note that\n\n\[ \n{\\pi }_{X}\\left( A\\right) = {\\chi }^{-1}\\left( E\\right) = \\mathcal{A}\\left( {{\\chi }^{-1}\\left( \\left\\{ {E}_{s}\\right\\} \\right) }\\right) .\n\]\n\nSince \( \\mathcal{E} \) is closed under the Souslin operation, \( {\\pi }_{X}\\left( A\\right) \\in \\mathcal{E} \) .\n\nBy 5.5.2, there is a \( \\sigma \\left( {{\\mathbf{\\sum }}_{1}^{1}\\left( \\mathcal{C}\\right) }\\right) \) -measurable section \( v : E \\rightarrow Y \) of \( C \) . Take \( f = v \\circ \\chi \) . Then \( f \) is an \( \\mathcal{E} \) -measurable section of \( A \) .
Yes
Corollary 5.5.8 Let \( \left( {X,\mathcal{A}, P}\right) \) be a complete probability space, \( Y \) a Polish space, and \( B \in \mathcal{A}\bigotimes {\mathcal{B}}_{Y} \) . Then \( {\pi }_{X}\left( B\right) \in \mathcal{A} \), and \( B \) admits an \( \mathcal{A} \) -measurable section.
Proof. Since \( \mathcal{A} \) is closed under the Souslin operation, the result follows from 5.5.7.
No
Theorem 5.7.1 (Novikov [90]) Let \( X, Y \) be Polish spaces and \( \mathcal{A} \) a countably generated sub \( \sigma \) -algebra of \( {\mathcal{B}}_{X} \) . Suppose \( B \in \mathcal{A}\bigotimes {\mathcal{B}}_{Y} \) is such that the sections \( {B}_{x} \) are compact. Then \( {\pi }_{X}\left( B\right) \in \mathcal{A} \), and \( B \) admits an \( \mathcal{A} \) -measurable section.
Proof. Since the projection of a Borel set with compact sections is Borel (4.7.11), \( {\pi }_{X}\left( B\right) \) is Borel. Since \( {\pi }_{X}\left( B\right) \) is a union of atoms of \( \mathcal{A} \), by the Blackwell - Mackey theorem (4.5.7), it is in \( \mathcal{A} \) .\n\nLet \( U \) be an open set in \( Y \) . Write \( U = \mathop{\bigcup }\limits_{n}{F}_{n} \), the \( {F}_{n} \) ’s closed. Then\n\n\[{\pi }_{X}\left( {B\bigcap \left( {X \times U}\right) }\right) = \mathop{\bigcup }\limits_{n}{\pi }_{X}\left( {B\bigcap \left( {X \times {F}_{n}}\right) }\right) .\n\]\n\nHence, by 4.7.11 and 4.5.7, \( {\pi }_{X}\left( {B\bigcap \left( {X \times U}\right) }\right) \in \mathcal{A} \) . It follows that the multifunction \( x \rightarrow {B}_{x} \) defined on \( {\pi }_{X}\left( B\right) \) is \( \mathcal{A} \) -measurable. The result follows from the selection theorem of Kuratowski and Ryll-Nardzewski (5.2.1). ∎
Yes
Theorem 5.7.2 (Lusin) Let \( X, Y \) be Polish spaces and \( B \subseteq X \times Y \) Borel with sections \( {B}_{x} \) countable. Then \( B \) admits a Borel uniformization.
Proof. By 3.3.17, there is a closed set \( E \) in \( {\mathbb{N}}^{\mathbb{N}} \) and a one-to-one continuous map \( f : E \rightarrow X \times Y \) with range \( B \) . Set\n\n\[ H = \left\{ {\left( {x,\alpha }\right) \in X \times E : {\pi }_{X}\left( {f\left( \alpha \right) }\right) = x}\right\} . \]\n\nThen \( H \) is a closed set in \( X \times {\mathbb{N}}^{\mathbb{N}} \) with sections \( {H}_{x} \) countable. Further, \( {\pi }_{X}\left( B\right) = {\pi }_{X}\left( H\right) \) . Fix a countable base \( \left( {V}_{n}\right) \) for \( {\mathbb{N}}^{\mathbb{N}} \) . Let\n\n\[ {Z}_{n} = \left\{ {x \in X : {H}_{x}\bigcap {V}_{n}\text{ is a singleton }}\right\} . \]\n\nBy 4.12.2, \( {Z}_{n} \) is coanalytic. Each \( {H}_{x} \) is countable and closed, and so if nonempty must have an isolated point. Therefore,\n\n\[ \mathop{\bigcup }\limits_{n}{Z}_{n} = {\pi }_{X}\left( H\right) = {\pi }_{X}\left( B\right) \]\n\nHence, \( {\pi }_{X}\left( B\right) \) is both coanalytic and analytic, and so by Souslin’s theorem, Borel. By the weak reduction principle for coanalytic sets (4.6.5), there exist pairwise disjoint Borel sets \( {B}_{n} \subseteq {Z}_{n} \) such that \( \mathop{\bigcup }\limits_{n}{B}_{n} = \mathop{\bigcup }\limits_{n}{Z}_{n} \) . Let\n\n\[ D = \mathop{\bigcup }\limits_{n}\left\lbrack {\left( {{B}_{n} \times {V}_{n}}\right) \bigcap H}\right\rbrack \]\n\nThen \( D \) is a Borel uniformization of \( H \) . Let \( g : D \rightarrow X \times X \) be the map defined by \( g\left( {x,\alpha }\right) = f\left( \alpha \right) \) . Since \( g \) is one-to-one, the set\n\n\[ C = \{ f\left( \alpha \right) : \left( {x,\alpha }\right) \in D\} \]\n\n is Borel (4.5.4). It clearly uniformizes \( B \) .
Yes
Proposition 5.7.3 Let \( X \) be a Polish space and \( \Pi \) a countably separated partition of \( X \) with all equivalence classes countable. Then \( \mathbf{\Pi } \) admits a Borel cross section.
Proof. Let \( Y \) be a Polish space and \( f : X \rightarrow Y \) a Borel map such that\n\n\[ \n{x\Pi }{x}^{\prime } \Leftrightarrow f\left( x\right) = f\left( {x}^{\prime }\right) .\n\]\n\nDefine\n\n\[ \nB = \{ \left( {y, x}\right) \in Y \times X : f\left( x\right) = y\} .\n\]\n\nThen \( B \) is a Borel set with sections \( {B}_{y} \) countable. By 5.7.2, \( {\pi }_{Y}\left( B\right) \) is Borel and there is a Borel section \( g : {\pi }_{Y}\left( B\right) \rightarrow X \) of \( B \) . Note that \( g \) is one-toone. Take \( S \) to be the range of \( g \) . Then \( S \) is Borel by 4.5.4. Evidently, it is a cross section of \( \mathbf{\Pi } \) .
Yes
Theorem 5.8.4 (Kechris [52]) Let \( X, Y \) be Polish spaces. Assume that \( x \rightarrow {\mathcal{I}}_{x} \) is a Borel on Borel map assigning to each \( x \in X \) a \( \sigma \) -ideal \( {\mathcal{I}}_{x} \) of subsets of \( Y \) . Suppose \( B \subseteq X \times Y \) is a Borel set such that for every \( x \in {\pi }_{X}\left( B\right) ,{B}_{x} \notin {\mathcal{I}}_{x} \) . Then \( {\pi }_{X}\left( B\right) \) is Borel, and \( B \) admits a Borel section.
Proof. Since \( x \rightarrow {\mathcal{I}}_{x} \) is Borel on Borel,\n\n\[{\pi }_{X}\left( B\right) = {\left\{ x : {B}_{x} \in {\mathcal{I}}_{x}\right\} }^{c}\]\n\nis Borel.\n\nIt remains to prove that \( B \) admits a Borel section. Fix a closed subset \( F \) of \( {\mathbb{N}}^{\mathbb{N}} \) and a continuous bijection \( f : F \rightarrow B \) . For each \( s \in {\mathbb{N}}^{ < \mathbb{N}} \) we define a Borel subset \( {B}_{s} \) of \( X \) such that for every \( s, t \in {\mathbb{N}}^{ < \mathbb{N}} \),\n\n(i) \( {B}_{e} = {\pi }_{X}\left( B\right) \) ;\n\n(ii) \( \left| s\right| = \left| t\right| \& s \neq t \Rightarrow {B}_{s} \cap {B}_{t} = \varnothing \) ;\n\n(iii) \( {B}_{s} = \mathop{\bigcup }\limits_{n}{B}_{{s}^{ \frown }n} \) ; and\n\n(iv) \( {B}_{s} \subseteq \left\{ {x \in X : {\left( f\left( \sum \left( s\right) \cap F\right) \right) }_{x} \notin {\mathcal{I}}_{x}}\right\} \) .\n\nWe define such a system of sets by induction on \( \left| s\right| \) . Suppose \( {B}_{t} \) have been defined for every \( t \in {\mathbb{N}}^{ < \mathbb{N}} \) of length \( < n \), and \( s \in {\mathbb{N}}^{ < \mathbb{N}} \) is of length \( n - 1 \) . For any \( k \in \mathbb{N} \), let\n\n\[{D}_{k} = \left\{ {x \in {B}_{s} : {\left( f\left( \sum \left( s \hat{} k\right) \bigcap F\right) \right) }_{x} \notin {\mathcal{I}}_{x}}\right\} .\n\nSince \( f \) is one-to-one and continuous, \( f\left( {\sum \left( {s\widehat{}k}\right) \bigcap F}\right) \) is Borel (4.5.4). Hence, as \( x \rightarrow {\mathcal{I}}_{x} \) is Borel on Borel, each \( {D}_{k} \) is Borel. By (iv), \( {B}_{s} = \mathop{\bigcup }\limits_{k}{D}_{k} \) . Take\n\n\[{B}_{s \land k} = {D}_{k} \smallsetminus \mathop{\bigcup }\limits_{{l < k}}{D}_{l}\]\n\nWe define \( u : {\pi }_{X}\left( B\right) \rightarrow Y \) as follows. Given any \( x \in {\pi }_{X}\left( B\right) \) there is a unique \( \alpha \in F \) (call it \( p\left( x\right) \) ) such that \( x \in {B}_{\alpha \mid k} \) for every \( k \) . Define \( u \) by\n\n\[u\left( x\right) = {\pi }_{Y}\left( {f\left( {p\left( x\right) }\right) }
Yes
Example 5.8.3 Let \( X, Y \) be Polish spaces and \( G : X \rightarrow Y \) a closed-valued Borel measurable multifunction. Define \( \mathcal{I} : X \rightarrow \mathcal{P}\left( {\mathcal{P}\left( Y\right) }\right) \) by \[ \mathcal{I}\left( x\right) = \{ I \subseteq Y : I\text{ is meager in }G\left( x\right) \} .
By imitating the proof of 3.5.18 we can show the following: For every open set \( U \) in \( Y \) and every Borel set \( B \) in \( X \times Y \), the sets \[ {B}^{*U} = \left\{ {x \in X : G\left( x\right) \bigcap U \neq \varnothing }\right. \] \[ \text{&}{B}_{x}\bigcap G\left( x\right) \bigcap U\text{is comeager in}G\left( x\right) \bigcap U\} \] and \[ {B}^{\Delta U} = \left\{ {x \in X : G\left( x\right) \bigcap U \neq \varnothing }\right. \] \[ \text{&}{B}_{x}\bigcap G\left( x\right) \bigcap U\text{is nonmeager in}G\left( x\right) \bigcap U\} \] are Borel. It follows that \( \mathcal{I} \) is Borel on Borel.
No
Theorem 5.8.5 (Kechris [52] and Sarbadhikari [100]) If B is a Borel subset of the product of two Polish spaces \( X \) and \( Y \) such that \( {B}_{x} \) is nonmeager in \( Y \) for every \( x \in {\pi }_{X}\left( B\right) \), then \( B \) admits a Borel uniformization.
Proof. Apply 5.8.4 with \( {\mathcal{I}}_{x} \) as in example 5.8.2.
No
Example 5.8.6 As a special case of 5.8.5 we see that every Borel set \( B \subseteq X \times Y \) with \( {B}_{x} \) a dense \( {G}_{\delta } \) set admits a Borel uniformization. However, there is an \( {F}_{\sigma } \) subset \( E \) of \( \left\lbrack {0,1}\right\rbrack \times {\mathbb{N}}^{\mathbb{N}} \) with sections \( {E}_{x} \) dense and that does not admit a Borel uniformization. Here is an example.
Let \( C \subseteq \left\lbrack {0,1}\right\rbrack \times {\mathbb{N}}^{\mathbb{N}} \) be a closed set with projection to the first coordinate space \( \left\lbrack {0,1}\right\rbrack \), that does not admit a Borel uniformization. Such a set exists by 5.1.7. For each \( s \in {\mathbb{N}}^{ < \mathbb{N}} \), fix a homeomorphism \( {f}_{s} : \sum \rightarrow \sum \left( s\right) \) . Take\n\n\[ E = \mathop{\bigcup }\limits_{{s \in {\mathbb{N}}^{ < \mathbb{N}}}}\left\{ {\left( {x,{f}_{s}\left( \alpha \right) }\right) : \left( {x,\alpha }\right) \in B}\right\} .\n\]\n\nThis \( E \) works.
No
Theorem 5.8.7 (Blackwell and Ryll-Nardzewski [17]) Let \( X, Y \) be Polish spaces, \( P \) a transition probability on \( X \times Y \), and \( B \) a Borel subset of \( X \times Y \) such that \( P\left( {x,{B}_{x}}\right) > 0 \) for all \( x \in {\pi }_{X}\left( B\right) \) . Then \( {\pi }_{X}\left( B\right) \) is Borel, and \( B \) admits a Borel uniformization.
Proof. Apply 5.8.4 with \( {\mathcal{I}}_{x} \) as in Example 5.8.1.
No
Theorem 5.8.8 (Blackwell and Ryll-Nardzewski) Let \( X, Y \) be Polish spaces, \( \mathcal{A} \) a countably generated sub \( \sigma \) algebra of \( {\mathcal{B}}_{X} \), and \( P \) a transition probability on \( X \times Y \) such that for every \( B \in {\mathcal{B}}_{Y}, x \rightarrow P\left( {x, B}\right) \) is \( \mathcal{A} \) -measurable. Suppose \( B \in \mathcal{A}\bigotimes {\mathcal{B}}_{Y} \) is such that \( P\left( {x,{B}_{x}}\right) > 0 \) for all \( x \in {\pi }_{X}\left( B\right) \) . Then \( {\pi }_{X}\left( B\right) \in \mathcal{A} \), and \( B \) admits an \( \mathcal{A} \) -measurable section.
Proof of 5.8.8. By a slight modification of the argument contained in the proof of 3.4.24 we see that for every \( E \in \mathcal{A} \otimes {\mathcal{B}}_{Y}, x \rightarrow P\left( {x,{E}_{x}}\right) \) is \( \mathcal{A} \) -measurable. As \( {\pi }_{X}\left( B\right) = \left\{ {x \in X : P\left( {x,{B}_{x}}\right) > 0}\right\} \), it follows that \( {\pi }_{X}\left( B\right) \in \mathcal{A} \) . By 5.8.9, there is a \( C \subseteq B \) in \( \mathcal{A}\bigotimes {\mathcal{B}}_{Y} \) with compact \( x \) -sections such that \( P\left( {x,{C}_{x}}\right) > 0 \) for every \( x \in {\pi }_{X}\left( B\right) \) . In particular, \( {\pi }_{X}\left( B\right) = {\pi }_{X}\left( C\right) \) . The result follows from Novikov's uniformization theorem (5.7.1).
Yes
Lemma 5.8.9 Let \( X, Y,\mathcal{A} \), and \( P \) be as above. For every \( E \in \mathcal{A}\bigotimes {\mathcal{B}}_{Y} \) and every \( \epsilon > 0 \), there is an \( F \in \mathcal{A}\bigotimes {\mathcal{B}}_{Y} \) contained in \( E \) such that \( {F}_{x} \) is compact and \( P\left( {x,{F}_{x}}\right) \geq \epsilon \cdot P\left( {x,{E}_{x}}\right) \) .
Proof. Let \( \mathcal{M} \) be the class of all sets in \( \mathcal{A}\bigotimes {\mathcal{B}}_{Y} \) such that the conclusion of the lemma holds for every \( P \) and every \( \epsilon > 0 \) . By 3.4.20, \( \mathcal{M} \) contains all rectangles \( A \times B \), where \( A \in \mathcal{A} \) and \( B \) Borel in \( Y \) . So, \( \mathcal{M} \) contains all finite disjoint unions of such rectangles. It is fairly routine to check that \( \mathcal{M} \) is a monotone class. Therefore, the result follows from the monotone class theorem.
Yes
Proposition 5.8.10 Let \( X, f \), and \( \mathcal{A} \) be as above. An everywhere proper conditional distribution given \( f \) exists if and only if there is an \( \mathcal{A} \) - measurable \( g : X \rightarrow X \) such that \( f\left( {g\left( x\right) }\right) = f\left( x\right) \) for all \( x \) .
Proof. Suppose an \( \mathcal{A} \) -measurable \( g : X \rightarrow X \) such that \( f \circ g \) is the identity exists. Define\n\n\[ Q\left( {x, B}\right) = \left\{ \begin{array}{ll} 1 & \text{ if }g\left( x\right) \in B \\ 0 & \text{ otherwise. } \end{array}\right. \]\n\nIt is easy to verify that \( Q \) has the desired properties.\n\nConversely, let an everywhere proper conditional distribution \( Q \) given \( f \) exist. Let\n\n\[ S = \{ \left( {x, y}\right) \in X \times X : f\left( x\right) = f\left( y\right) \} .\n\nThen \( S \in \mathcal{A}\bigotimes {\mathcal{B}}_{Y} \) and \( Q\left( {x,{S}_{x}}\right) = 1 \) . By 5.8.8, there is an \( \mathcal{A} \) -measurable section \( g \) of \( S \), which is what we are looking for.\n\nSince \( g \) is \( \mathcal{A} \) -measurable, \( g\left( x\right) = g\left( y\right) \) whenever \( f\left( x\right) = f\left( y\right) \) . It follows that there is a Borel function \( h : \mathbb{R} \rightarrow X \) such that \( g\left( x\right) = h\left( {f\left( x\right) }\right) \) for all \( x \) . Then the range of \( f \) equals \( \{ y \in \mathbb{R} : f\left( {h\left( y\right) }\right) = y\} \), which is a Borel set. It follows from the above proposition that whenever the range of \( f \) is not a Borel set, everywhere proper conditional distributions given \( f \) cannot exist.
Yes
Proposition 5.8.13 (Feldman and Moore [41]) Every Borel equivalence relation on a Polish space \( X \) with equivalence classes countable is induced by a countable group of Borel automorphisms.
Proof. Let \( \Pi \) be a Borel equivalence relation on \( X \) with equivalence classes countable. By 5.8.11, write\n\n\[ \Pi = \mathop{\bigcup }\limits_{n}{G}_{n} \]\n\nwhere \( {\pi }_{1} \mid {G}_{n} \) is one-to-one, \( {\pi }_{1}\left( {x, y}\right) = x \) ; i.e., the \( {G}_{n} \) ’s are graphs of Borel functions. Let\n\n\[ {H}_{n} = \varphi \left( {G}_{n}\right) \]\n\nwhere \( \varphi \left( {x, y}\right) = \left( {y, x}\right) \) . Then \( {\pi }_{2} \mid {H}_{n} \) is one-to-one, where \( {\pi }_{2}\left( {x, y}\right) = y \) . Let\n\n\[ X \times X \smallsetminus \Delta = \mathop{\bigcup }\limits_{k}\left( {{U}_{k} \times {V}_{k}}\right) \]\n\n\( {U}_{k},{V}_{k} \) open, where \( \Delta = \{ \left( {x, x}\right) : x \in X\} \) . Note that \( {U}_{k} \cap {V}_{k} = \varnothing \) . Put\n\n\[ {D}_{nmk} = \left( {{G}_{n}\bigcap {H}_{m}}\right) \bigcap \left( {{U}_{k} \times {V}_{k}}\right) . \]\n\nNote that \( {\pi }_{1}\left| {{D}_{nmk}\text{and}{\pi }_{2}}\right| {D}_{nmk} \) are one-to-one, and\n\n\[ {\pi }_{1}\left( {D}_{nmk}\right) \bigcap {\pi }_{2}\left( {D}_{nmk}\right) = \varnothing . \]\n\nSo, there is a Borel automorphism \( {g}_{nmk} \) of \( X \) given by\n\n\[ {g}_{nmk}\left( x\right) = \left\{ \begin{array}{ll} y & \text{ if }\left( {x, y}\right) \in {D}_{nmk}\text{ or }\left( {y, x}\right) \in {D}_{nmk}, \\ x & \text{ otherwise. } \end{array}\right. \]\n\nClearly,\n\n\[ \mathbf{\Pi } = \Delta \bigcup \mathop{\bigcup }\limits_{{nmk}}\operatorname{graph}\left( {g}_{nmk}\right) \]\n\nNow take \( G \) to be the group of automorphisms generated by \( \left\{ {g}_{nmk}\right. \) : \( n, m, k \in \mathbb{N}\} \) .
Yes
Theorem 5.9.1 (Miller [85]) Every partition \( \Pi \) of a Polish space \( X \) into \( {G}_{\delta } \) sets such that the saturation of every basic open set is simultaneously \( {F}_{\sigma } \) and \( {G}_{\delta } \) admits a section \( s : X \rightarrow X \) that is Borel measurable of class 2. In particular, such partitions admit a \( {G}_{\delta } \) cross section.
Proof. Let \( \left( {U}_{n}\right) \) be a countable base for the topology of \( X \) . Let \( \left( {V}_{n}\right) \) be an enumeration of \( \left\{ {{U}_{n}^{ * } : n \in \mathbb{N}}\right\} \bigcup \left\{ {{\left( {U}_{n}^{ * }\right) }^{c} : n \in \mathbb{N}}\right\} \) . Let \( {\mathcal{T}}^{\prime } \) be the topology on \( X \) generated by \( \left\{ {{U}_{n} : n \in \mathbb{N}}\right\} \bigcup \left\{ {{V}_{n} : n \in \mathbb{N}}\right\} \) . Note that every \( {\mathcal{T}}^{\prime } \) open set is an \( {F}_{\sigma } \) set in \( X \) relative to the original topology of \( X \) . Consider the map \( f : X \rightarrow X \times {2}^{\mathbb{N}} \) defined by\n\n\[ f\left( x\right) = \left( {x,{\chi }_{{V}_{0}}\left( x\right) ,{\chi }_{{V}_{1}}\left( x\right) ,{\chi }_{{V}_{2}}\left( x\right) ,\ldots }\right) ,\;x \in X.\]\n\nThe map \( f \) is one-to-one and of class 2 . Let \( G \) be the range of \( f \) . It is quite easy to see that\n\n\[ {\mathcal{T}}^{\prime } = \left\{ {{f}^{-1}\left( W\right) : W\text{ open in }G}\right\}\]\n\nArguing as in the proof of 3.2.5, it is easily seen that \( G \) is a \( {G}_{\delta } \) set in \( X \times {2}^{\mathbb{N}} \) . Therefore, by 2.2.1, \( \left( {X,{\mathcal{T}}^{\prime }}\right) \) is Polish. As argued in 5.1.13,\n\n\[ \left\lbrack x\right\rbrack = \bigcap \left\{ {{U}_{n}^{ * } : {U}_{n}^{ * } \supseteq \left\lbrack x\right\rbrack }\right\}\]\n\nSo, each \( \Pi \) -equivalence class is closed relative to \( {\mathcal{T}}^{\prime } \) .\n\nLet \( \mathcal{L} \) be the set of all invariant subsets of \( X \) that are clopen relative to \( {\mathcal{T}}^{\prime } \) . We claim that the multifunction \( x \rightarrow \left\lbrack x\right\rbrack \) is \( {\mathcal{L}}_{\sigma } \) -measurable. Let \( \mathcal{S} = {\left\{ {V}_{n} : n \in \mathbb{N}\right\} }_{d} \), the set of all finite intersections of sets in \( \left\{ {{V}_{n} : n \in \mathbb{N}}\right\} \) . Any \( {\mathcal{T}}^{\prime } \) -open \( W \) is of the form \( U\bigcup V, U \) open relative to the original toplogy of \( X \) and \( V \) a union of sets in \( \mathcal{S} \) . Then \( {W}^{ * } = {U}^{ * }\bigcup V \), which proves our claim.\n\nBy the selection theorem of Kuratowski and Ryll-Nardzewski, there exists an \( {\mathcal{L}}_{\sigma } \) -measurable selection \( s \) for \( x \rightarrow \left\lbrack x\right\rbrack \) . In particular, \( s \) is continuous with respect to \( {\mathcal{T}}^{\prime } \) . The associated cross section \( S = \{ x \in s\left( x\right) = x\} \) is \( {\mathcal{T}}^{\prime } \) - closed and so is a \( {G}_{\delta } \) set relative to the original toplogy of \( X \) .
Yes
Theorem 5.9.2 (Srivastava [114]) Every Borel measurable partition \( \mathbf{\Pi } \) of a Polish space \( X \) into \( {G}_{\delta } \) sets admits a Borel cross section.
Proof. (Kechris) For \( x \in X \) let \( \left\lbrack x\right\rbrack \) denote the member of \( \mathbf{\Pi } \) containing \( x \) . Consider the multifunction \( p : X \rightarrow X \) defined by\n\n\[ p\left( x\right) = \operatorname{cl}\left( \left\lbrack x\right\rbrack \right) \]\n\nThen \( p : X \rightarrow X \) is a closed-valued measurable multifunction. Further, for every \( x, y \in X, x \equiv y \Leftrightarrow p\left( x\right) = p\left( y\right) \) (5.9.1).\n\nNow consider \( F\left( X\right) \), the set of nonempty closed subsets of \( X \) with Effros Borel structure. By 3.3.10, it is standard Borel. Note that \( p \) considered as a map from \( X \) to \( F\left( X\right) \) is measurable. Let\n\n\[ P = \{ \left( {F, x}\right) \in F\left( X\right) \times X : p\left( x\right) = F\} . \]\n\nThe set \( P \) is Borel. For \( F \in F\left( X\right) \), let \( {\mathcal{I}}_{F} \) be the \( \sigma \) -ideal of subsets of \( X \) that are meager in \( F \) . As the multifunction \( F \rightarrow F \) from \( F\left( x\right) \) to \( X \) is measurable, by \( {5.8.3}, F \rightarrow {\mathcal{I}}_{F} \) is Borel on Borel. By the Baire category theorem, \( {P}_{F} \notin {\mathcal{I}}_{F} \) for each \( F \) . Therefore, by 5.8.4, \( D = {\pi }_{F\left( X\right) }\left( P\right) \) is Borel, and there is a Borel section \( q : D \rightarrow X \) of \( P \) . Let\n\n\[ S = \{ x \in X : x = q\left( {p\left( x\right) }\right) \} . \]\n\nClearly \( S \) is a Borel cross section of \( \mathbf{\Pi } \) .
Yes
Theorem 5.9.5 \( \operatorname{irr}\left( A\right) / \sim \) is standard Borel if and only if \( A \) is GCR.
Its proof makes crucial uses of 5.4.3 and 4.5.4. We refer the interested reader to [4] and [43] for a proof.
No
Theorem 5.10.1 (The reflection theorem) Let \( X \) be a Polish space and \( \Phi \subseteq \mathcal{P}\left( X\right) {\mathbf{\Pi }}_{1}^{1} \) on \( {\mathbf{\Pi }}_{1}^{1} \) . For every \( {\mathbf{\Pi }}_{1}^{1} \) set \( A \in \Phi \) there is a Borel \( B \subseteq A \) in \( \Phi \) .
Proof. Suppose there is a \( {\mathbf{\Pi }}_{1}^{1} \) set \( A \subseteq X \) in \( \Phi \) that does not contain a Borel set belonging to \( \Phi \) . We shall get a contradiction. Let \( \varphi \) be a \( {\mathbf{\Pi }}_{1}^{1} \) -norm on \( A \) and\n\n\[ C = \left\{ {\left( {x, y}\right) : y{ < }_{\varphi }^{ * }x}\right\} \]\n\nWe claim that\n\n\[ x \notin A \Leftrightarrow {C}_{x} \in \Phi . \]\n\n\( \left( *\right) \)\n\nSuppose \( x \notin A \) . Then \( {C}_{x} = A \in \Phi \) . Conversely, if \( x \in A \), then \( {C}_{x} \) is a Borel subset of \( A \) . So by our assumptions, \( {C}_{x} \notin \Phi \) .\n\nSince \( \Phi \) is \( {\mathbf{\Pi }}_{1}^{1} \) on \( {\mathbf{\Pi }}_{1}^{1},{A}^{c} \) is \( {\mathbf{\Pi }}_{1}^{1} \) by \( \left( \star \right) \) . Hence, by Souslin’s theorem, it is Borel, contradicting our assumption again.
Yes
Theorem 5.10.2 Let \( X, Y \) be Polish spaces and \( A \subseteq X \times Y \) analytic with sections \( {A}_{x} \) countable. Then every coanalytic set \( B \) containing \( A \) contains a Borel set \( E \supseteq A \) with all sections countable.
Proof. Let \( C = {B}^{c} \) . Define \( \Phi \subseteq \mathcal{P}\left( {X \times Y}\right) \) by\n\n\[ D \in \Phi \Leftrightarrow {D}^{c} \subseteq B\& \forall x\left( {\left( {D}^{c}\right) }_{x}\right. \text{is countable})\text{.}\]\n\nUsing 4.3.7 we can easily check that \( \Phi \) is \( {\mathbf{\Pi }}_{1}^{1} \) on \( {\mathbf{\Pi }}_{1}^{1} \) . Since \( {A}^{c} \in \Phi \), by 5.10.1 there is a Borel set \( D \) in \( \Phi \) contained in \( {A}^{c} \) . Take \( E = {D}^{c} \) .
Yes
Theorem 5.10.3 (Lusin) Every analytic set with countable sections, in the product of two Polish spaces, can be covered by countably many Borel graphs.
Proof. The result immediately follows from 5.10.2 and 5.8.11.
No
Proposition 5.10.4 (Burgess) Let \( X \) be Polish, \( E \) an analytic equivalence relation on \( X \), and \( C \subseteq X \times X \) a coanalytic set containing \( E \) . Then there is a Borel equivalence relation \( B \) such that \( E \subseteq B \subseteq C \) .
Proof of 5.10.4. Applying 5.10.5 repeatedly, by induction on \( n \) we can define a sequence of Borel sets \( \left( {B}_{n}\right) \) such that\n\n\[ E \subseteq {B}_{n} \subseteq \mathcal{E}\left( {B}_{n}\right) \subseteq {B}_{n + 1} \subseteq C \]\n\nfor all \( n \) . Take \( B = \mathop{\bigcup }\limits_{n}{B}_{n} \) .
No
Lemma 5.10.5 Let \( X \) be a Polish space, \( P \) analytic, \( C \) coanalytic, and \( \mathcal{E}\left( P\right) \subseteq C \) . Then there is a Borel set \( B \) containing \( P \) such that\n\n\[ \mathcal{E}\left( B\right) \subseteq C\text{.} \]
Proof. Define \( \Phi \subseteq \mathcal{P}\left( {X \times X}\right) \) by\n\n\[ D \in \Phi \Leftrightarrow \mathcal{E}\left( {D}^{c}\right) \subseteq C. \]\n\n\( \Phi \) is \( {\mathbf{\Pi }}_{1}^{1} \) on \( {\mathbf{\Pi }}_{1}^{1} \) . Further, \( {P}^{c} \in \Phi \) . By the reflection theorem (5.10.1), there is a Borel set \( D \) in \( \Phi \) that is contained in \( {P}^{c} \) . Take \( B = {D}^{c} \) .
Yes
Corollary 5.10.6 For every analytic equivalence relation \( E \) on a Polish space \( X \) there exist Borel equivalence relations \( {B}_{\alpha },\alpha < {\omega }_{1} \), such that \( E = \) \( \mathop{\bigcap }\limits_{{\alpha < {\omega }_{1}}}{B}_{\alpha }. \)
Proof. By 4.3.17, write \( E = \mathop{\bigcap }\limits_{{\alpha < {\omega }_{1}}}{C}_{\alpha },{C}_{\alpha } \) coanalytic. By 5.10.4, for each \( \alpha \) there exists a Borel equivalence relation \( {B}_{\alpha } \) such that \( E \subseteq {B}_{\alpha } \subseteq {C}_{\alpha } \) .
Yes
Theorem 5.11.4 Every countably generated sub \( \sigma \) -algebra of the Borel \( \sigma \) - algebra of a Polish space has a minimal complement.
Proof of 5.11.4. Let \( X \) be Polish and \( \mathcal{C} \) a countably generated sub \( \sigma \) -algebra of \( {\mathcal{B}}_{X} \) . \n\nCase 1. There is a cocountable atom \( A \) of \( \mathcal{C} \) .\n\nLet \( f : X \smallsetminus A \rightarrow A \) be a one-to-one map. Take\n\n\[ \mathcal{D} = \sigma \left( {\{ \{ x, f\left( x\right) \} : x \in X \smallsetminus A\} \bigcup {\mathcal{B}}_{A \smallsetminus f\left( {A}^{c}\right) }}\right) . \]\n\nBy 5.11.5, \( \mathcal{D} \) is a minimal complement of \( \mathcal{C} \) .\n\nCase 2. There is an uncountable atom \( A \) of \( \mathcal{C} \) such that \( X \smallsetminus A \) is also uncountable.\n\nLet \( f : A \rightarrow {A}^{c} \) be a Borel isomorphism and \( g : X \rightarrow X \) the map that equals \( f \) on \( A \) and the identity on \( {A}^{c} \) . Take\n\n\[ \mathcal{D} = {g}^{-1}\left( {\mathcal{B}}_{X}\right) \]\n\nBy 5.11.5, \( \mathcal{D} \) is a minimal complement of \( \mathcal{C} \) .\n\nCase 3. All atoms of \( \mathcal{C} \) are countable. Since \( \mathcal{C} \) is countably generated with all atoms countable, by 5.8.12 there exists a countable partition \( {G}_{n} \) of \( X \) such that each \( {G}_{n} \) is a partial cross section of the set of atoms of \( \mathcal{C} \) . It is easy to choose the \( {G}_{n} \) ’s in such a way that for distinct \( {G}_{n} \) and \( {G}_{m} \) , \( {G}_{n}\bigcup {G}_{m} \) is not a partial cross section of the set of atoms of \( \mathcal{C} \) . The result follows by 5.11.5 by taking\n\n\[ \mathcal{D} = \sigma \left( \left\{ {{G}_{n} : n \in \mathbb{N}}\right\} \right) . \]
Yes
Lemma 5.11.5 Let \( X \) be Polish and \( \mathcal{C} \) a countably generated sub \( \sigma \) -algebra of \( {\mathcal{B}}_{X} \) . Suppose \( \mathcal{D} \) is a countably generated sub \( \sigma \) -algebra of \( {\mathcal{B}}_{X} \) such that every atom \( A \) of \( \mathcal{D} \) is a partial cross section of the atoms of \( \mathcal{C} \) . Further, assume that for any two distinct atoms \( {C}_{1},{C}_{2} \) of \( \mathcal{D},{C}_{1}\bigcup {C}_{2} \) is not a partial cross section of the set of atoms of \( \mathcal{C} \) . Then \( \mathcal{D} \) is a minimal complement of \( \mathcal{C} \).
Proof. Under the hypothesis, \( \mathcal{C} \vee \mathcal{D} \) is a countably generated sub \( \sigma \) - algebra of \( {\mathcal{B}}_{X} \) with atoms singletons. Hence, by 4.5.7, \( \mathcal{C} \vee \mathcal{D} = {\mathcal{B}}_{X} \) . Let \( {\mathcal{D}}^{ * } \) be a proper countably generated sub \( \sigma \) -algebra of \( \mathcal{D} \) . By the corollary to 4.5.7, there is an atom \( A \) of \( {\mathcal{D}}^{ * } \) that is not an atom of \( \mathcal{D} \) . Hence, it is a union of more than one atom of \( \mathcal{D} \) . Hence, there exist two distinct points \( x, y \) of \( A \) that belong to the same atom of \( \mathcal{C} \) . This implies that there is no \( E \in \mathcal{C}\bigvee {\mathcal{D}}^{ * } \) containing exactly one of \( x, y \) . So, \( \mathcal{C}\bigvee {\mathcal{D}}^{ * } \neq {\mathcal{B}}_{X} \) . The result now follows from 5.11.1 and 5.11.2.
Yes
Theorem 5.12.1 (Arsenin, Kunugui [60]) Let \( B \subseteq X \times Y \) be a Borel set, \( X, Y \) Polish, such that \( {B}_{x} \) is \( \sigma \) -compact for every \( x \) . Then \( {\pi }_{X}\left( B\right) \) is Borel, and \( B \) admits a Borel uniformization.
Proof of 5.12.1. Write \( B = \mathop{\bigcup }\limits_{n}{B}_{n} \), the \( {B}_{n} \) ’s Borel with compact sections. That this can be done follows from 5.12.3. Then\n\n\[{\pi }_{X}\left( B\right) = \mathop{\bigcup }\limits_{n}{\pi }_{X}\left( {B}_{n}\right)\]\n\nSince the projection of a Borel set with compact sections is Borel (4.7.11), each \( {\pi }_{X}\left( {B}_{n}\right) \), and hence \( {\pi }_{X}\left( B\right) \), is Borel. Let\n\n\[{D}_{n} = {\pi }_{X}\left( {B}_{n}\right) \smallsetminus \mathop{\bigcup }\limits_{{m < n}}{\pi }_{X}\left( {B}_{m}\right)\]\n\nThen the \( {D}_{n} \) ’s are Borel and pairwise disjoint. Further, the set\n\n\[C = \mathop{\bigcup }\limits_{n}\left( {{B}_{n}\bigcap \left( {{D}_{n} \times Y}\right) }\right)\]\n\n is a Borel subset of \( B \) with compact sections such that \( {\pi }_{X}\left( C\right) = {\pi }_{X}\left( B\right) \) . By Novikov’s uniformization theorem (5.7.1), \( C \) admits a Borel uniformization, and our result follows.
Yes
Theorem 5.12.3 Let \( X, Y \) be Polish spaces and \( A \subseteq X \times Y \) a Borel set with sections \( {A}_{x} \) \(\sigma\) -compact. Then \( A = \mathop{\bigcup }\limits_{n}{B}_{n} \), where each \( {B}_{n} \) is Borel with \( {\left( {B}_{n}\right) }_{x} \) compact for all \( x \) and all \( n \) .
Proof. The result trivially follows from 5.12.2 by taking \( B = {A}^{c} \) .
No
Proposition 5.12.4 Let \( B \subseteq X \times Y \) be a Borel set with sections \( {B}_{x} \) that are \( {G}_{\delta } \) sets in \( Y \) . Then there exist Borel sets \( {B}_{n} \) with open sections such that \( B = \mathop{\bigcap }\limits_{n}{B}_{n} \) .
Proof. Let \( Z \) be a compact metric space containing (a homeomorph of) \( Y \) . Then \( B \) is Borel in \( X \times Z \) with sections \( {G}_{\delta } \) sets (2.2.7). By 5.12.3, there exist Borel sets \( {C}_{n} \) in \( X \times Z \) with sections compact such that \( \left( {X \times Z}\right) \smallsetminus B = \) \( \mathop{\bigcup }\limits_{n}{C}_{n} \) . Take \( {B}_{n} = \left( {X \times Y}\right) \smallsetminus {C}_{n} \) .
Yes
Proposition 5.12.6 Let \( X \) be a Polish space and \( \mathcal{B} \subseteq F\left( X\right) \) hereditary. Then \( {\Omega }_{{D}_{\mathcal{B}}} = {\mathcal{B}}_{\sigma } \cap F\left( X\right) \) .
Proof. Fix a closed set \( A \subseteq X \) and a countable base \( \left( {U}_{n}\right) \) for \( X \) .\n\nLet \( {D}^{\infty }\left( A\right) = \varnothing \) . Then\n\n\[ A = \mathop{\bigcup }\limits_{{\alpha < {\left| A\right| }_{D}}}\left( {{D}^{\alpha }\left( A\right) \smallsetminus {D}^{\alpha + 1}\left( A\right) }\right) \]\n\n\[ = \mathop{\bigcup }\limits_{{\alpha < {\left| A\right| }_{D}}}\mathop{\bigcup }\limits_{n}\left\{ {{U}_{n} \cap {D}^{\alpha }\left( A\right) : \operatorname{cl}\left( {{U}_{n} \cap {D}^{\alpha }\left( A\right) }\right) \in \mathcal{B}}\right\} \]\n\n\[ = \;\mathop{\bigcup }\limits_{{\alpha < {\left| A\right| }_{D}}}\mathop{\bigcup }\limits_{n}\{ \operatorname{cl}\left( {{U}_{n}\bigcap {D}^{\alpha }\left( A\right) }\right) : \operatorname{cl}\left( {{U}_{n}\bigcap {D}^{\alpha }\left( A\right) }\right) \in \mathcal{B}\} . \]\n\nThe last equality holds because \( A \) is closed. Thus, \( A \in {\mathcal{B}}_{\sigma } \).\n\nTo prove the converse, take an \( A \in {\mathcal{B}}_{\sigma } \cap F\left( X\right) \) . Suppose \( {D}^{\infty }\left( A\right) \neq \varnothing \) . We shall get a contradiction. Write \( A = \mathop{\bigcup }\limits_{m}{B}_{m},{B}_{m} \in \mathcal{B} \) . By the Baire category theorem, there exist \( n \) and \( m \) such that\n\n\[ \varnothing \neq {D}^{\infty }\left( A\right) \bigcap {U}_{n} \subseteq {D}^{\infty }\left( A\right) \bigcap {B}_{m} \]\n\nThis implies that\n\n\[ {D}^{{\left| A\right| }_{D} + 1}\left( A\right) \neq {D}^{{\left| A\right| }_{D}}\left( A\right) \]\n\nWe have arrived at a contradiction.
Yes
Proposition 5.12.7 Let \( X \) be Polish and \( D \) a derivative on \( X \) such that\n\n\[ \n\\{ \\left( {A, B}\\right) \\in F\\left( X\\right) \\times F\\left( X\\right) : A \\subseteq D\\left( B\\right) \\} \n\]\n\nis analytic. Then\n\n(i) \( {\\Omega }_{D} \) is coanalytic, and\n\n(ii) for every analytic \( \\mathcal{A} \\subseteq {\\Omega }_{D} \), \n\n\[ \n\\sup \\left\\{ {{\\left| A\\right| }_{D} : A \\in \\mathcal{A}}\\right\\} < {\\omega }_{1} \n\]
Proof. Assertion (i) follows from the following equivalence:\n\n\[ \nA \\notin {\\Omega }_{D} \\Leftrightarrow \\exists B\\left( {B \\neq \\varnothing \\& B \\subseteq A\\& B \\subseteq D\\left( B\\right) }\\right) .\n\]\n\n(The sets \( A \) and \( B \) are closed in \( X \).)\n\nSuppose (ii) is false for some analytic \( \\mathcal{A} \\subseteq {\\Omega }_{D} \). Then,\n\n\[ \n\\sup \\left\\{ {{\\left| A\\right| }_{D} : A \\in \\mathcal{A}}\\right\\} = {\\omega }_{1} \n\]\n\nDefine \( R \\subseteq {2}^{\\mathbb{N} \\times \\mathbb{N}} \\times F\\left( X\\right) \) as follows:\n\n\[ \n\\left( {x, A}\\right) \\in R \\Leftrightarrow x \\in L{O}^{ * }\\& \n\]\n\n\[ \n\\exists f \\in F{\\left( X\\right) }^{\\mathbb{N}}\\lbrack f\\left( 0\\right) = A\\& \n\]\n\n\[ \n\\forall m \\in D\\left( x\\right) \\{ f\\left( m\\right) \\neq \\varnothing \\& \n\]\n\n\[ \n\\left( {m \\neq 0 \\Rightarrow \\forall n{ < }_{x}^{ * }m\\left( {f\\left( m\\right) \\subseteq D\\left( {f\\left( n\\right) }\\right) }\\right) }\\right) \\} \\rbrack .\n\]\n\nIt is fairly easy to check that \( R \) is analytic and that for \( \\varnothing \\neq A \\in {\\Omega }_{D} \),\n\n\[ \nR\\left( {x, A}\\right) \\Leftrightarrow x \\in W{O}^{ * }\\& \\left| x\\right| \\leq {\\left| A\\right| }_{D}.\n\]\n\nBy our assumptions,\n\n\[ \nx \\in W{O}^{ * } \\Leftrightarrow \\exists A \\in \\mathcal{A}\\left( {R\\left( {x, A}\\right) }\\right) .\n\]\n\nThis implies that \( W{O}^{ * } \) is analytic, which is not the case, and our result is proved.
Yes
Theorem 5.13.1 (Lopez-Escobar) A subset \( A \) of \( {X}_{L} \) is invariant (with respect to the logic action) and Borel, if and only if there is a sentence \( \sigma \) of \( {L}_{{\omega }_{1}\omega } \) such that \( A = {A}_{\sigma } \) .
Proof. The sufficient part of this result is proved by induction on formulae of \( {L}_{{\omega }_{1}\omega } \) as follows:\n\nFor every formula \( \phi \left\lbrack {{v}_{0},{v}_{1},\ldots ,{v}_{k - 1}}\right\rbrack \), the set\n\n\[ \n{A}_{\phi, k} = \left\{ {\left( {x,{n}_{0},{n}_{1},\ldots ,{n}_{k - 1}}\right) : {\mathcal{A}}_{x} \vDash \phi \left\lbrack {{n}_{0},{n}_{1},\ldots ,{n}_{k - 1}}\right\rbrack }\right\} \n\]\n\nis Borel.\n\nThe necessary part is also proved by induction, but the induction in this case is a bit subtle. We proceed as follows. Let \( {\left( \mathbb{N}\right) }^{k} \) denote the set of all one-to-one finite sequences in \( \mathbb{N} \) of length \( k \) and for any \( s \in {\left( \mathbb{N}\right) }^{k} \) ,\n\n\[ \n\left\lbrack s\right\rbrack = \left\{ {g \in {S}_{\infty } : s \prec {g}^{-1}}\right\} \n\]\n\nClearly, \( \left\{ {\left\lbrack s\right\rbrack : s \in {\left( \mathbb{N}\right) }^{k}}\right\} \) form a base for the topology of \( {S}_{\infty } \) .\n\nSuppose \( A \) is a Borel set in \( {X}_{L} \) . Then, for every \( k \) there is a formula \( \phi \left\lbrack {{v}_{0},{v}_{1},\ldots ,{v}_{k - 1}}\right\rbrack \) of \( {L}_{{\omega }_{1}\omega } \) such that\n\n\[ \n{A}_{\phi, k} = \left\{ {\left( {x, s}\right) : s \in {\left( \mathbb{N}\right) }^{k}\& x \in {A}^{*\left\lbrack s\right\rbrack }}\right\} .\n\]\n\nThis is proved by induction on \( A \) using basic identities on Vaught transforms. We invite readers to complete the proof themselves. (Otherwise consult \( \left\lbrack {\left\lbrack {53}\right\rbrack \text{, p.97}}\right\rbrack \) .)\n\nNow, if \( A \subseteq {X}_{L} \) is an invariant Borel set, then \( {A}^{ * } = A \) and the result follows from the above assertion by taking \( k = 0 \) .
No
Theorem 5.13.8 Topological Vaught conjecture holds if \( G \) is a locally compact Polish group.
Assuming 5.13.9, we prove 5.13.8 as follows: Let \( G \) be a locally compact Polish group acting continuously on a Polish space \( X \) . Write \( G = \mathop{\bigcup }\limits_{n}{K}_{n} \) , \( {K}_{n} \) compact. Then, for \( x, y \in X \) ,\n\n\[ \exists g \in G\left( {y = g \cdot x}\right) \Leftrightarrow \exists n\exists g \in {K}_{n}\left( {y = g \cdot x}\right) . \]\n\nSince \( {K}_{n} \) is compact and the set \( \left\{ {\left( {x, y, g}\right) \in X \times X \times {K}_{n} : y = g \cdot x}\right\} \) is closed, the equivalence relation induced by the group action is an \( {F}_{\sigma } \) set. Our result now follows from 5.13.9.
Yes
Theorem 5.13.9 Let \( E \) be an analytic equivalence relation on a Polish space \( X \) with all equivalence classes \( {F}_{\sigma } \) . Then the number of equivalence classes is \( \leq {\aleph }_{0} \) or perfectly many.
Proof of 5.13.9. Let \( X \) be a Polish space and \( E \) an analytic equivalence relation on \( X \) with all its equivalence classes \( {F}_{\sigma } \) sets. Further assume that there are uncountably many \( E \) -equivalence classes. Fix a countable base \( \left( {V}_{n}\right) \) for the topology of \( X \) . Let \( P \) be the union of all basic open sets which is contained in countably many equivalence classes and \( Q \) its saturation; i.e., \( Q = \operatorname{proj}\left( {E\bigcap \left( {P \times P}\right) }\right) \) . Thus
No
Proposition 5.13.10 Suppose \( X \) is a Polish space and \( E \) an equivalence relation on \( X \) which is meager in \( {X}^{2} \) . Then \( E \) has perfectly many equivalence classes.
Proof. Let \( E \subseteq \mathop{\bigcup }\limits_{n}{F}_{n},{F}_{n} \) closed and nowhere dense in \( {X}^{2} \) . Without any loss of generality, we further assume that the diagonal \( \left\{ {\left( {x, y}\right) \in {X}^{2} : x = y}\right\} \) is contained in each of \( {F}_{n} \) .\n\nFor each \( s \in {2}^{ < \mathbb{N}} \), we define a nonempty open set \( U\left( s\right) \) in \( X \) satisfying the following properties.\n\n(i) \( \operatorname{diameter}\left( {U\left( s\right) }\right) \leq {2}^{-\left| s\right| } \) .\n\n(ii) \( s \prec t \Rightarrow \operatorname{cl}\left( {U\left( t\right) }\right) \subseteq U\left( s\right) \) .\n\n(iii) If \( s \neq {s}^{\prime } \) and \( \left| s\right| = \left| {s}^{\prime }\right| \), then \( \left( {U\left( s\right) \times U\left( {s}^{\prime }\right) }\right) \cap {F}_{\left| s\right| } = \varnothing \) . In particular, \( {U}_{s} \) and \( {U}_{s} \) ’s are disjoint.\n\nWe define \( \left\{ {U\left( s\right) : s \in {2}^{ < \mathbb{N}}}\right\} \) by induction on \( \left| s\right| \) . Take \( U\left( e\right) \) to be any nonempty open set of diameter less than 1 disjoint from \( {F}_{0} \) . Since \( {F}_{0} \) is closed nowhere dense, such a set exists. Suppose \( n \) is a positive integer and \( U\left( s\right) \) has been defined for every sequence \( s \) of length less than \( n \) . Consider the set \( {F}_{n}^{{2}^{n}} \) . It is closed and nowhere dense in \( {X}^{{2}^{n + 1}} \) . Hence, there is an open set of the form \( \mathop{\prod }\limits_{{s \in {2}^{n - 1}}}\left( {U\left( {s \hat{} 0}\right) \times U\left( {s \hat{} 1}\right) }\right) \) contained in \( \mathop{\prod }\limits_{{s \in {2}^{n - 1}}}\left( {U\left( s\right) \times U\left( s\right) }\right) \) disjoint from \( {F}_{n}^{{2}^{n}} \) . We can further assume that the diameter of \( U\left( {s \hat{} \epsilon }\right) ,\left| s\right| = n - 1 \) and \( \epsilon = 0 \) or 1, is less than \( {2}^{-n} \), and that its closure is contained in \( U\left( s\right) \) .\n\nFor \( \alpha \in {2}^{\omega } \), let \( f\left( \alpha \right) \) be the unique element of \( X \) that belongs to each of \( U\left( {\alpha \mid n}\right) \) . It is easy to see that the range of \( f \) is a perfect set of pairwise \( E \) -inequivalent elements.
Yes
Theorem 5.13.12 (Sami) Topological Vaught conjecture holds if \( G \) is abelian.
Proof. Assume that the number of orbits is uncountable. We shall show that there is a perfect set of inequivalent elements.\n\nLet \( E \) be the equivalence relation on \( X \) defined by\n\n\[ \n{xEy} \Leftrightarrow {G}_{x} = {G}_{y} \n\]\n\nwhere \( {G}_{x} \) is the stabilizer of \( x \) . Let \( y = g \cdot x \) for some \( g \in G \) . Then \( {G}_{x} = {g}^{-1} \cdot {G}_{y} \cdot g = {G}_{y} \), as \( G \) is abelian. Thus,\n\n\[ \nx{E}_{a}y \Rightarrow {xEy} \n\]\n\nwhere \( {E}_{a} \) is the equivalence relation induced by the action. Now note that\n\n\[ \n{xEy} \Leftrightarrow \forall g\left( {g \cdot x = x \Leftrightarrow g \cdot y = y}\right) . \n\]\n\nHence, \( E \) is coanalytic.\n\nSuppose there are uncountably many \( E \) -equivalence classes. Then by Silver’s theorem, there is a perfect set of \( E \) -inequivalent elements. In particular, there is a perfect set of \( {E}_{a} \) -inequivalent elements.\n\nNow assume that the set of \( E \) -equivalence classes is countable. We shall show that \( {E}_{a} \) is Borel. Our proof will then follow from Silver’s theorem.\n\nLet \( Y \subseteq X \) be an \( E \) -equivalence class. It is sufficient to show that \( {E}_{a} \cap \left( {Y \times Y}\right) \) is Borel. Let \( x \in Y \) and \( H = {G}_{x} \) . The partition of \( G \) by the cosets of \( H \) is lower-semicontinuous. Hence, there is a Borel cross-section \( S \) for this partition. For \( x, y \in Y \), we have the following:\n\n\[ \nx{E}_{a}y \Leftrightarrow \left( {\exists \text{ a unique }g \in S}\right) \left( {y = g \cdot x}\right) ; \n\]\n\n i.e., \( {E}_{a} \cap \left( {Y \times Y}\right) \) is a one-to-one projection of the Borel set\n\n\[ \n\{ \left( {x, y, g}\right) : g \in S\text{ and }y = g \cdot x\} . \n\]\n\nHence, \( {E}_{a} \) is Borel.
Yes
Lemma 5.13.14 Suppose \( \left\{ {{A}_{\alpha } : \alpha < {\omega }_{1}}\right\} \) is a family of Borel subsets of a Polish space \( X \) and \( E \) the equivalence relation on \( X \) defined by\n\n\[ \n{xEy} \Leftrightarrow \forall \alpha \left( {x \in {A}_{\alpha } \Leftrightarrow y \in {A}_{\alpha }}\right), x, y \in X.\n\]\n\n\( \left( *\right) \)\n\nThen the number of \( E \) -equivalence classes is \( \leq {\aleph }_{1} \) or perfectly many.
Proof of 5.13.14. Although the proof of the lemma is messy looking, ideawise it is quite simple. Assume that the number of \( E \) -equivalence classes is \( > {\aleph }_{1} \) . We shall then show that there are perfectly many \( E \) -equivalence classes. The following fact will be used repeatedly in the proof of the lemma.\n\nFact. Suppose \( Z \) is a subset of \( X \) of cardinality \( > {\aleph }_{1} \) such that no two distinct elements \( Z \) are \( E \) -equivalent. Then there is an \( \alpha < {\omega }_{1} \) such that both \( Z \cap {A}_{\alpha } \) and \( Z \cap {A}_{\alpha }^{c} \) are of cardinality \( > {\aleph }_{1} \) .\n\nWe prove this fact by contradiction. If possible, let for every \( \alpha < {\omega }_{1} \) at least one of \( Z\bigcap {A}_{\alpha } \) and \( Z\bigcap {A}_{\alpha }^{c} \) be of cardinality \( \leq {\aleph }_{1} \) . Denote one such set by \( {M}_{\alpha } \) . We claim that \( Z \smallsetminus \mathop{\bigcup }\limits_{\alpha }{M}_{\alpha } \) is a singleton. Suppose not. Let \( x \) , \( y \) be two distinct elements of \( Z \smallsetminus \mathop{\bigcup }\limits_{\alpha }{M}_{\alpha } \) . Since \( x, y \) are \( E \) -inequivalent, by \( \left( \star \right) \) there exists an \( \alpha < {\omega }_{1} \) such that exactly one of \( x \) and \( y \) belong to \( {A}_{\alpha } \) . It follows that at least one of \( x, y \) belong to \( {M}_{\alpha } \) . But this is not the case. Hence, \( Z \smallsetminus \mathop{\bigcup }\limits_{\alpha }{M}_{\alpha } \) contains at most one point. It follows that the cardinality of \( Z \) is at most \( {\aleph }_{1} \), and we have arrived at a contradiction.\n\nFix a compatible complete metric on \( X \) . Following our usual notation, for \( \epsilon = 0 \) or 1, we set\n\n\[ \n{A}_{\alpha }^{\epsilon } = \left\{ \begin{array}{ll} {A}_{\alpha } & \text{ if }\epsilon = 0 \\ {A}_{\alpha }^{c} & \text{ if }\epsilon = 1 \end{array}\right.\n\]\n\nSince \( {A}_{\alpha }^{\epsilon } \) analytic, there is a continuous map \( {f}_{\alpha }^{\epsilon } : {\mathbb{N}}^{\mathbb{N}} \rightarrow X \) whose range is \( {A}_{\alpha }^{\epsilon } \) . We can arrange matters so that for every \( s \in {\mathbb{N}}^{ < \mathbb{N}} \), the diameter of \( {f}_{\alpha }^{\epsilon }\left( {\sum \left( s\right) }\right) \) is at most \( {2}^{-\left| s\right| } \) .\n\nFix any subset \( Z \) of \( X \) of cardinality \( > {\aleph }_{1} \) consisting of pairwise \( E \) -
Yes
Example 5.13.15 Let \( L \) be a first order language whose non-logical symbols consists of exactly one binary relation symbol. So, \( {X}_{L} = {2}^{\omega \times \omega } \) . We claim that in this case the equivalence relation \( {E}_{a} \) induced by the logic action is not Borel. Suppose not. Then \( {E}_{a} \in {\mathbf{\sum }}_{\beta }^{0} \) for some \( \beta < {\omega }_{1} \) .
It follows that \( W{O}^{\alpha } = \{ x \in {WO} : \left| x\right| \leq \alpha \} \in {\mathbf{\sum }}_{\beta }^{0} \) for every \( \alpha < {\omega }_{1} \) . Now take any Borel set \( A \) in \( {\mathbb{N}}^{\mathbb{N}} \) which is not of additive class \( \beta \) . Since \( {WO} \) is \( {\mathbf{\Pi }}_{1}^{1} \) -complete, there is a continuous function \( f : {\mathbb{N}}^{\mathbb{N}} \rightarrow {LO} \) such that \( A = {f}^{-1}\left( {WO}\right) \) . But by the boundedness theorem, \( A = {f}^{-1}\left( {W{O}^{\alpha }}\right) \) for some \( \alpha \) . It follows that \( A \in {\mathbf{\sum }}_{\beta }^{0} \), and we have arrived at a contradiction.
Yes
Theorem 5.13.19 (Stern) Assume analytic determinacy. Let \( E \) be an analytic equivalence relation on a Polish space \( X \) such that all but countably many equivalence classes are of bounded Borel rank. Then the number of equivalence classes is \( \leq {\aleph }_{0} \) or perfectly many.
The proof this result is beyond the scope of this book.
No
Theorem 5.14.1 (Kondô’s theorem) Let \( X, Y \) be Polish spaces. Every coanalytic set \( C \subseteq X \times Y \) admits a coanalytic uniformization.
We shall show that there is a sequence of coanalytic norms on a given co-analytic set with certain \
No
Corollary 5.14.5 Let \( X \) be a Polish space and \( A \subseteq X \) coanalytic. Then A admits a very good \( {\mathbf{\Pi }}_{1}^{1} \) -scale.
Proof. By 2.6.9 there is a closed set \( D \subseteq {\mathbb{N}}^{\mathbb{N}} \) and a continuous bijection \( f : D \rightarrow X \) . Now, \( {f}^{-1}\left( A\right) \cap D \) is a \( {\mathbf{\Pi }}_{1}^{1} \) subset of \( {\mathbb{N}}^{\mathbb{N}} \) and hence admits a very good \( {\mathbf{\Pi }}_{1}^{1} \) -scale by 5.14.4. The scale on \( A \) is now obtained by transfer via the function \( f \) .
Yes
We consider the discretization of the boundary value problem for the ordinary differential equation\n\n\[ - {u}^{\prime \prime }\left( x\right) = f\left( {x, u\left( x\right) }\right) ,\;x \in \left\lbrack {0,1}\right\rbrack \]\n\n(2.1)\n\nwith boundary condition\n\n\[ u\left( 0\right) = u\left( 1\right) = 0. \]\n\n(2.2)\n\nHere, \( f : \left\lbrack {0,1}\right\rbrack \times \mathbb{R} \rightarrow \mathbb{R} \) is a given continuous function, and we are looking for a twice continuously differentiable solution \( u : \left\lbrack {0,1}\right\rbrack \rightarrow \mathbb{R} \).
For the approximate solution we choose an equidistant subdivision of the interval \( \left\lbrack {0,1}\right\rbrack \) by setting\n\n\[ {x}_{j} = {jh},\;j = 0,\ldots, n + 1, \]\n\nwhere the step size is given by \( h = 1/\left( {n + 1}\right) \) with \( n \in \mathbb{N} \). At the internal grid points \( {x}_{j}, j = 1,\ldots, n \), we replace the differential quotient in the differential equation (2.1) by the difference quotient\n\n\[ {u}^{\prime \prime }\left( {x}_{j}\right) \approx \frac{1}{{h}^{2}}\left\lbrack {u\left( {x}_{j + 1}\right) - {2u}\left( {x}_{j}\right) + u\left( {x}_{j - 1}\right) }\right\rbrack \]\nto obtain the system of equations\n\n\[ - \frac{1}{{h}^{2}}\left\lbrack {{u}_{j - 1} - 2{u}_{j} + {u}_{j + 1}}\right\rbrack = f\left( {{x}_{j},{u}_{j}}\right) ,\;j = 1,\ldots, n, \]\n\nfor approximate values \( {u}_{j} \) to the exact solution \( u\left( {x}_{j}\right) \). This system has to be complemented by the two boundary conditions \( {u}_{0} = {u}_{n + 1} = 0 \). For an abbreviated notation we introduce the \( n \times n \) matrix\n\n\[ A = \frac{1}{{h}^{2}}\left( \begin{array}{rrrrrr} 2 & - 1 & & & & \\ - 1 & 2 & - 1 & & & \\ & - 1 & 2 & - 1 & & \\ \cdot & \cdot & \cdot & \cdot & \cdot & \cdot \\ & & & - 1 & 2 & - 1 \\ & & & & - 1 & 2 \end{array}\right) \]\n\nand the vectors \( U = {\left( {u}_{1},\ldots ,{u}_{n}\right) }^{T} \) and \( F\left( U\right) = {\left( f\left( {x}_{1},{u}_{1}\right) ,\ldots, f\left( {x}_{n},{u}_{n}\right) \right) }^{T} \). Then our system of equations, including the boundary conditions, reads\n\n\[ {AU} = F\left( U\right) \]\n\n(2.3)\n\nFor obvious reasons, the above matrix \( A \) is called a tridiagonal matrix, and the vector \( F \) is diagonal; i.e., the \( j \) th component of \( F \) depends only on the \( j \) th component of \( u \). If (2.1) is a linear differential equation, i.e., if \( f \) depends linearly on the second variable \( u \), then the tridiagonal system of equations (2.3) also is linear.
Yes
Consider the linear integral equation\n\n\[ \varphi \left( x\right) - {\int }_{0}^{1}K\left( {x, y}\right) \varphi \left( y\right) {dy} = f\left( x\right) ,\;x \in \left\lbrack {0,1}\right\rbrack ,\]
For the numerical approximation we replace the integral by the rectangular sum\n\n\[ {\int }_{0}^{1}K\left( {x, y}\right) \varphi \left( y\right) {dy} \approx \frac{1}{n}\mathop{\sum }\limits_{{k = 1}}^{n}K\left( {x,{x}_{k}}\right) \varphi \left( {x}_{k}\right) \]\n\nwith equidistant grid points \( {x}_{k} = k/n, k = 1,\ldots, n \) . If we require the approximated equation to be satisfied only at the grid points, we arrive at the system of linear equations\n\n\[ {\varphi }_{j} - \frac{1}{n}\mathop{\sum }\limits_{{k = 1}}^{n}K\left( {{x}_{j},{x}_{k}}\right) {\varphi }_{k} = f\left( {x}_{j}\right) ,\;j = 1,\ldots, n, \]\n\nfor approximate values \( {\varphi }_{j} \) to the exact solution \( \varphi \left( {x}_{j}\right) \) .
Yes