3 Forcing
Definition (Forcing).
π
is called a forcing poset or forcing if it is a partial order and
π
and
π
is the largest element. Elements of
are called condition.
is
interpreted as
β
is stronger than β
β
is weaker than β
Alternative: βJerusalem conventionβ. Interpret
as β is stronger
than β.
We are not following the Jerusalem convention.
Definition (Compatible).
and
are compatible if there is .
Otherwise, we say they are incompatible (which we write as ).
Definition (Antichain).
is an antichain if any two distinct elements of
are incompatible.
Definition (Dense).
is dense if .
Definition (Filter).
is called a filter if
-
(a)
-
(b)
If only
has property (a), we call it a filter base, and then
|
is the filter generated from .
Note.
Filters cannot contain incompatible elements.
Definition (-generic).
If
is a family of dense sets, then
is called -generic
if
is a filter and .
Theorem 3.1.
Assuming that:
Proof.
Let .
Pick
arbitrarily and define by recursion
by picking some
with .
Then
is a filter base, so let
be the filter generated by it. Then it is -generic
by construction. β‘
Main example
Fix any sets
and
|
Define
and π.
What does
mean?
|
Lemma 3.2.
Assuming that:
Then is
a function.
Lemma 3.3.
is a dense set. If ,
and is
-generic,
then .
Proof.
If ,
find ,
then ,
TODO β‘
3.1 Cohen Forcing
Example 1
.
If is
-generic,
then
(by above).
Lemma 3.4.
Assuming that:
-
Fix
and define
|
-
Then .
However: if
is a countable transitive model and you consider
then by TheoremΒ 3.1, there is a -generic.
And for this ,
TODO
Example 2
as
before.
Lemma 3.6.
Assuming that:
Then is
a surjection.
Corollary 3.7.
Assuming that:
However: if
is a countable transitive model and, e.g. TODO
Example 3
.
Assume
is infinite. Consider
|
This is dense for .
|
Lemma 3.8.
Assuming that:
Then there is an injection from
into .
Proof.
Fix
and define
|
guarantees that
is an injection. β‘
Corollary 3.9.
Assuming that:
However: if is a countable
transitive model of ,
is countable
and so a -generic
exists.
Recap: a countable
transitive model of
(or a sufficiently large finite fragment).
: dense / filter
/ -generic.
Example.
produces
a new function .
Cohen forcing.
Example.
produces
a surjection .
COLAPSE
of .
Example.
produces
an injection .
If is a countable
transitive model of ,
then is countable, so by
Theorem, we have a -generic.
GOAL: Build an extension
such that ,
a countable
transitive model of ,
and
is
minimal.
Names
Idea: Think of elements of
as βtruth valuesβ for the von Neumann construction.
Then
|
is the proper class of all names.
Consider: .
Then .
Since this is a recursive definition using absolute concepts, being a name is absolute for transitive models:
|
TODO
Examples
,
|
βThe name for a set that contains
with value .β
βThe name for whatever
describes with value β.
Interpretation
If , we interpret
a -name
as follows:
|
Important: This is a recursive definition.
Thus: the valuation is absolute for transitive models containing
and
.
Example.
-
(1)
:
clearly .
-
(2)
:
|
-
(3)
:
|
The relationship between
and
affects these possibilities.
Example: if
and
is a filter, then
is impossible.
Example: if π
and
is a non-empty filter, then
is impossible.
Example: if
and
is a filter, then
is impossible.
Definition 3.11 (Generic extension).
The (generic) extension for any countable transitive model
and any
where
is
|
Obviously, is a
countable set with
(Example (1)).
Also, by definition,
is transitive.
Note:
|
Need to show
-
(1)
.
-
(2)
.
-
(3)
.
-
(4)
is minimal.
Canonical Names
Definition 3.12 (Canonical name).
Let .
Define by recursion the canonical name for
by
π |
Lemma 3.13.
if π.
Corollary 3.14.
if π.
Alternative construction of canonical names without
π is on
Example Sheet 2.
Lemma 3.15.
.
Proof.
Calculate:
β‘
Corollary 3.16.
.
Warm-up: Suppose .
Define
ππ |
Pairing: Then
|
(βupβ stands for unordered pair).
Corollary 3.17.
(if π).
Union: If
is a name, define
|
Claim:
if is a
filter.
Proof.
Suppose .
So
for some
with .
So
with ,
,
.
So .
Hence ,
.
So .
Conversely, suppose .
Then
(
with ,
with ).
Hence since
a filter, find
with .
Then ,
so .
β‘
TODO
Further Recap
We proved
|
Homework was: Think about why power set is not easy.
βUnionβ proof was:
collect all natural candidates of names for elements and assign the natural values.
Problem: If you try to do this for power set, neither the βnatural candidates for namesβ nor the
βnatural valuesβ are obvious. Itβll turn out that they are obvious in the end, but that requires some
assistance.
Note: Separation and Replacement are even worse.
One remaining easy axiom: .
By well-ordering theorem,
holds if and only if
injection. If ,
then there is
such that .
Notation.
I write .
Consider .
In , I have
for some
ordinal .
In ,
define
|
Call this .
Then is an
injection. So,
holds in .
Forcing
Definition (Forcing language).
Fix
a countable transitive model of ,
a forcing poset.
We call the language
|
the forcing language (over ).
Definition (Interpretation of forcing language).
If
is a
-generic
over
and is
in the forcing language, we say
if and only if
where .
Definition (Semantic forcing predicate).
:
Forall
-generic
over
with ,
.
β
forces β
We often omit .
Two theorems at the heart of forcing:
-
(1)
Forcing Theorem: If
is a -generic
over ,
then
|
-
(2)
Definability Theorem: ββ
is absolutely definable for transitive models containing .
We are going to do the following:
-
(a)
Define the syntactic forcing predicate
that is absolutely definable.
-
(b)
Prove the Forcing Theorem for .
-
(c)
Derive that .
Lectures 11 and 12.
Observations (under the assumption of the Forcing Theorem):
-
(1)
If
and ,
then .
-
(2)
If ,
then there is a name
and
such that .
[if
and ,
then by definition ,
so there is a name
such that .
By Forcing Theorem, find
such that .
Find ,
.
By (1), .
]
Proof of Separation in .
Let be an
-formula (not in the
forcing language) and .
Need a name for
|
[For readability, we now drop parameters ].
Fix such
that .
Define
|
By the Definability Theorem,
is a name in .
Claim: .
-
ββ
If , then
there is
such that ,
. Since
, we
get ,
. Together
with ,
we get
|
Hence .
-
ββ
If ,
then
and .
So there is ,
.
By the Forcing THeorem, ,
.
Also, there is
and .
Find ,
such that .
Hence ,
so .
β‘
Proof of Replacement.
Since we already have Separation, itβs enough to show the following:
if is a functional
formula and ,
then there is
such that
|
(as before, we suppress parameters for notational ease).
We work in and
identify a name
for . Fix
such that
. Find
large enough
such that .
Consider
(,
a name). By Definability
Theorem, this is an
formula.
By Levy Reflection Theorem, find
such that is
absolute between
and .
Define
and .
Now we verify ()
holds: Fix ,
. Since
is functional,
we know that
holds in for some
. By Forcing
Theorem, there is
such that .
So .
By absoluteness, .
This means
such that .
Thus: .
β‘
Definition (Dense below ).
is called dense below
if .
Lemma 3.18.
Assuming that:
Then
Definition of the syntactic forcing relation
.
Two recursions:
First ββ by
recursion on .
Then ββ
without recursion.
Then the rest by recursion on formula complexity.
: if and
only if
and
Remark.
This is a recursion on .
: if and
only if
|
is dense below .
Remark.
No recursion involved, just ββ.
Recursion on complexity of formulas:
-
:
if and only if
and .
-
:
if and only if ,
.
-
:
if and only if .
Remark.
These definitions remind us of Kripke semantics for intuitionistic logic.
Lemma 3.19.
The following are equivalent:
-
(i)
.
-
(ii)
.
-
(iii)
is dense below .
Proof.
If this is true for
of the form
and of the form ,
then itβs true for all formulas.
For
atomic, we get that (ii)
(i) and (ii)
(iii) are trivial.
(i)
(ii) follows from LemmaΒ 3.18(i).
(iii)
(i) follows from LemmaΒ 3.18(ii). β‘
Strategy:
Theorem 3.20 (Syntactic Forcing Theorem).
if and only if .
Corollary 3.21 (Definability Theorem).
if and only if .
()
Corollary 3.22 (Forcing Theorem).
if and only if .
This corollary is immediate from combining the two previously stated results.
Proof of Definability Theorem from Syntactic Forcing Theorem.
-
This is just Syntactic Forcing Theorem.
-
Suppose .
By LemmaΒ 3.18, we need to show
is dense below .
Suppose not. Then I find
such that ,
.
So .
If ,
then .
Since ,
.
Since ,
.
Contradiction! β‘
Proof of the Syntactic Forcing Theorem.
The Theorem for
can be proved by induction on .
The Theorem for
can be proved once the Theorem has been established for .
Rest is induction on formula complexity.
Letβs do
and leave the rest to Example Sheet 3. Assume we have proved it for .
We will prove it for .
-
.
Consider
|
By definition of ,
this is dense. Find .
By assumption, ,
so .
-
Let
such that .
By definition, ,
.
If ,
thus .
By induction hypothesis, find ,
.
So .
By Lemma, ,
contradiction to .
β‘
Note.
The Forcing Theorem is very useful! The inner details of the proof are not very important though. We
wonβt really discuss these details once we have proved the theorem.
Continuing the proof of Syntactic Forcing Theorem. Assume Syntactic Forcing Theorem for
ββ and prove if
for ββ.
Want to show:
if and
only if
|
is dense below .
Proof.
-
Assume that ,
i.e. .
Thus, there is
such that
()
and .
So by Syntactic Forcing Theorem for ,
find
such that .
Find
such that .
Claim that
is dense below .
Let
and pick the above .
Then we have
and
(since ).
-
Assume ,
,
i.e.
is dense below .
By LemmaΒ 3.18(iv), find ,
thus there is
such that ,
.
Since ,
,
we have .
By Syntactic Forcing Theorem for ββ,
we have .
So .
β‘
Now we prove it for ββ.
We prove this by induction on name rank, so assume that Syntactic Forcing Theorem for
ββ is true for
names
with smaller name rank.
Reminder: we need to show
if and only if ,
.
TODO: double check the below
Proof.
-
Assume
such that .
Need to prove .
Weβll show that the fact that
is dense below
implies .
The other direction is the same proof but with
and
flipped.
Proof of this: Let ,
so find
such that
and .
So, the corresponding
is dense below .
Find
such that .
Then
is dense below .
So find
such that .
(Note that ,
so ).
Since
and ,
find
such that .
By induction hypothesis,
and ,
which implies
(since
and ).
-
Assume .
Consider the set
Claim: is
dense. If , then
, so nothing to show.
If , then there is
some that fails to
be dense below .
Weβll show: if is
not dense below ,
then we can find
such that
holds.
(Other proof:
not dense below
is
flipping
and ).
Suppose and
there is such that
is not dense
below . This
means: there is
such that
|
So, we get
|
for any that satisfies
. It canβt be compatible
with (finishes the
proof of the claim that
is dense).
Summary: We now have that
is dense.
|
Claim 2: If ,
then neither
nor holds
(again, just do
and then flip
and
for ).
If holds, then
find such that
and the incompatibility
statement holds. ,
so .
If
such that TODO
Put everything together: since
is dense by Claim 1, find .
Therefore, by Claim 2, ,
do
not hold.
Thus .
β‘
Summary: If
is -generic, then
there is an
injection from
into .
This implies if
we have ,
. This is
the goal for todayβs lecture.
Note that our proof above is not good enough for the statement
() from
Lecture 8:
For all
finite, there exists
finite such that if
is a countable transitive model of ,
then there is
countable transitive model of .
For this, we need to look more carefully at the proof of the GMT:
The proof proceeds AXIOM BY AXIOM and thus for each
, we find
finite such
that
implies .
Let finite
such that
proves that all relevant notions (name, value, β¦) are well-defined and absolute. Then for
finite,
define
Then, the proof shows ().
Letβs prove
in .
Definition (Preserves cardinals).
We say
preserves cardinals if
-generic
over ,
β
is a cardinalβ is absolute between
and .
Definition 3.23 (Countable chain condition).
We say
has the countable chain condition (c.c.c.) if every antichain (note the anti!) in
is countable.
Theorem 3.24.
Assuming that:
Lemma 3.27.
Assuming that:
-
-
-
-
,
Then there is
such that ,
and
.
Proof of TheoremΒ 3.24.
Suppose ,
,
so there is
and ,
,
is a surjection. Apply the lemma to get .
Define .
Since ,
.
But .
But then
thinks that
is not a cardinal, contradiction. β‘
Now we prove the lemma:
Proof.
Let .
Fix
a name for .
By Definability Theorem , .
Then
follows from definition.
follows from Forcing Theorem.
Letβs look at
is countable: If ,
let
be such that .
If , then
. Thus
is an antichain.
By countable chain condition, itβs countalbe. So
was countable. β‘
TODO
Proof of TheoremΒ 3.25.
Actually, we prove
has countable chain condition whenever
is countable.
-systems
form Example Sheet 3 Q33 are also called quasi-disjoint families.
(A family of finite sets
is called a -system
if there is a finite set
(called the root of the -system)
such that for all ,
if
then ).
-system
lemma: Any countable family of finite sets contains an uncountable -system.
Take any
uncountable and prove that itβs not an antichain.
If , then
finite.
Consider
Thatβs an uncountable family of finite sets, so by -system
lemma, find -system
uncountable.
Let
finite be the root of .
Since
is countable, there are only countably many functions .
Since
is uncountable, by pigeonhole principle, there are
such that
and .
But since ,
and
are compatible.
So
is not an antichain. β‘
We got .
Next time: What is the size of
in ?
Remember: LST Example Sheet 4: .
What if
was one of the forbidden values?
Remark.
Obtaining cannot be
quite as general as this proof: if ,
then this will remain true in .
Previous lecture: Suppose a
countable transitive model of .
TODO
Question: () What are
the possible values for ?
Mentioned last lecture: not all values are possible. In particular,
.
Definition (Cofinal).
is cofinal (
unbounded) if .
We can then define:
|
Example 3.28.
,
.
Lemma 3.29 (KΕnigβs Lemma).
.
Then LST Example Sheet 4 Q10 is the special case
,
of
KΕnigβs Lemma.
Consequence for :
, thus
.
Preview of answer to ():
Every value not prohibited by KΕnigβs Lemma is possible.
Now: !
Definition.
If
is any forcing, let
be any -sequence
of antichains in .
Let
Ε |
We call these nice names.
If and
has countable chain condition,
then there are at most many
antichains and thus at most
many -sequences
of antichains and thus nice names.
Theorem 3.30.
Assuming that:
Then there is a
nice name
such that
.
Note.
The Theorem does not need any assumptions about
.
Proof.
Start with
such that
(possibly not nice). Fix .
Fix either a well-ordering of
(possibly using
to get one) and build a maximal antichain
such that ,
Ε.
TODO.
Claim: .
:
If ,
then there is
such that Ε,
and then ,
so Ε.
So .
:
If ,
then by Forcing Theorem, get
such that Ε.
Subclaim: .
Indeed, by our lemma on compatibility ( Example Sheet 3), get
such that
for all .
Find .
Then Ε.
But that is in contradiction to
being maximal.
So find .
By definition, Ε
and .
So .
β‘
Corollary 3.31.
Assuming that:
Then .
Proof.
Follows directly from:
-
(a)
Theorem.
-
(b)
Calculation of the number of nice names.
β‘
Main Application
If , then
.
Calculate in ,
.
Hausdorffβs Formula:
|
So in particular:
So .
By this calculation, if ,
then .
Corollary 3.32.
If ,
then .
Remark.
This proof also shows that if
and is
-generic
over
where ,
then
Corollary: If ,
then .
Remark.
What happens at ?
|
By general theory, ,
but KΕnigβs Lemma gives .
What about the lower bound?
Our theorem and counting of nice names yields
|
If , then
|
So by KΕnigβs Lemma, .
Therefore, if ,
then .
TODO
First limit cardinal that is a possible value of
is .
Clearly, adds
injection from
into . So
.
Count nice names: .
If for all ,
(), then
.
(since has no
cofinal -sequence).
Thus (by
assumption ()).
Thus .
TODO
What about ?
More on nice names:
Definition (-nice names).
Generalise βnice namesβ to -nice
names:
Let
be a family of many
maximal -antichains:
|
for .
These are names for subsets of .
Observe that our theorem βevery
in has a
-nice
nameβ still goes through.
If is an
-cardinal such that every
antichain of has size
. [On Example Sheet 3,
this is called the -chain
condition.]
Let (in
). Then
is an upper bound on the number of -nice
names.
Thus
|
Question: Forcing with
and calculate .
Assume .
Then:
(since
has countable chain condition). So
|
Calculate :
|
Together:
Question: Is it possible to get PSA, i.e.
|
without .
In particular, can we get
First idea: Force with .
-
(1)
Yields
many subsets of .
-
(2)
Still has countable chain condition, so all cardinals are preserved.
-
(3)
How many -nice
names are there:
|
Get:
in .
Unfortunately,
Interpret the generic object
as for
.
Define .
Claim: For ,
. This is
since
|
is still dense.
So, forcing with gives the
same situation as forcing with
for ,
.
Idea:
|
Thus .
Consider
|
Properties:
-
(1)
We still have .
-
(2)
Not clear that this forcing is preserving cardinals!
First goal: What about preserving cardinals?
Clearly,
does not have the countable chain condition anymore.
With example (38) (on Example Sheet 3), we need to figure out the chain condition of
.
We need a -system
lemma for this: If
is regular () and
is a family of
sets of size of
size . Then there
is a -system
of size
.
This -system
lemma gives with the same proof as before:
has the
-chain
condition.
So: preserves
cardinals .
Closure
Definition (-closed).
A forcing is
called -closed
if any family
for that
is a descending chain:
there is
such that
for all .
Example.
is -cloesd.
[If is a descending
chain, then is a
condition in .]
Theorem 3.33.
Assuming that:
Corollary 3.34.
Forcing with
-
(a)
Does not change .
-
(b)
Therefore preserves
(see
Example Sheet 1 and the relation between codes for countable well-orders and preserving
).
Summary: Forcing with
over a model of
gives
with:
-
(1)
The same cardinals (cardinals
preserved by -chain
condition;
preserved by -closure).
-
(2)
(standard).
-
(3)
(by corollary 1 to the closure theorem).
-
(4)
Calculate number of nice names:
|
Hence .
|
|
|
|
Forcing |
Property |
Preservation |
Arithmetic |
|
|
|
|
|
countable chain
condition |
all cardinals |
,
|
|
|
|
|
|
countable chain
condition |
all cardinals |
,
,
|
|
|
|
|
| -closed.
If
,
then
-chain
condition |
preserved.
(Closure lemma
[not yet proved])
preserved | If
,
then
,
. |
|
|
|
|
|
Note , but our
model of
fails .
Question: Can we have ?
Closure lemma:
Theorem.
If
is -closed
and ,
then .
-closed: every descending
sequence of length
has a lower bound.
Proof.
Let ,
and assume towards
contradiction that .
.
Let
be a name for .
By Forcing Theorem, there is
such that .
Construct a -sequence
of conditions
TODO
TODO
The sequence
is defined in
(by Definability Theorem), so we can define
|
Then .
But now
or
for all .
uso Η§.
Hence .
But .
Contradiction. β‘
Note that while is always
-closed, the chain condition
depended on the value of .
The partial order has
in general the -chain
condition.
implies
, so all
cardinals are preserved.
However, if ,
then there is a gap and we do not know whether TODO.
If , does
|
preserve ?
Answer: always adds
a surjection from
to .
()
Application: If
and , then
adds a
surjection from
onto ,
i.e. . So
.
[Proof of ().
] A generic for
βisβ a map .
Define
by
|
Claim:
is a surjection onto .
If ,
,
consider
|
This is dense, and thus .
β‘
Back to our question: Can we get ?
Start with .
Consider
|
and let be
-generic
over .
Consider
|
and let be
-generic
over .
Claim: .
( is a
forcing iteration).
-
(1)
is cardinal preserving over
since .
So .
-
(2)
has countable chain condition,
so is cardinal preserving:
|
-
(3)
(lecture 15; since ).
-
(4)
Then
(using a nice name analysis, we could calculate ).
This proves the claim.
Important: The order of forcings matters!
Suppose .
|
-generic
over .
|
-generic
over .
Consider .
Then
But that means that forcing with
will collapse .
Since is
-closed,
|
In particular, .
So, this order does not achieve what we want.
Final Remark on Forcing CH
Assume .
Question: Can you obtain ?
The natural forcing would be
This collapses ;
it does not have the countable chain condition, but since it has size
, it has the
-chain condition,
so all cardinals
are preserved.
Clearly therefore:
|
But: is ?
Nice names: gives upper bound of
|
Thatβs not surprising, since any
in becomes a
new subset of
in via the new
bijection between
and .