| |
 |
|
|
Science Forum Index » Mathematics Forum » Comparative sigma-fields
Page 1 of 1
|
| Author |
Message |
| Stephen J. Herschkorn |
Posted: Thu Dec 18, 2003 6:26 am |
|
|
|
Guest
|
In general topology, one sometimes compares different topologies on the
same set. In particular, I am thinking of the various topologies on
function spaces, but there are several toplogies (mainly for examples,
it seems) even on the reals (e.g., the Sorgenfrey line).
In what I have seen of analysis (not much beyond first-year graduate
study), it seems that one compares measures on the same measurable space
(e.g., in the context of absolute continuity), but I have rarely seen
topics where one compares different sigma-fields on the same set. The
only example I can think of is conditional expectation in probability,
which is defined with respect to a sub-sigma-field of the underlying
probability space.
Are there any interesting and/or important examples where one compares
implication of different sigma-fields on the same set? I suspect one of
the issues here is that with unrelated sigma-fields, one cannot compare
measures.
--
Stephen J. Herschkorn herschko@rutcor.rutgers.edu |
|
|
| Back to top |
|
| ArtflDodgr |
Posted: Thu Dec 18, 2003 11:02 am |
|
|
|
Guest
|
In article <O9gEb.266296$655.48928492@news4.srv.hcvlny.cv.net>,
"Stephen J. Herschkorn" <herschko@rutcor.rutgers.edu> wrote:
Quote: In general topology, one sometimes compares different topologies on the
same set. In particular, I am thinking of the various topologies on
function spaces, but there are several toplogies (mainly for examples,
it seems) even on the reals (e.g., the Sorgenfrey line).
In what I have seen of analysis (not much beyond first-year graduate
study), it seems that one compares measures on the same measurable space
(e.g., in the context of absolute continuity), but I have rarely seen
topics where one compares different sigma-fields on the same set. The
only example I can think of is conditional expectation in probability,
which is defined with respect to a sub-sigma-field of the underlying
probability space.
Are there any interesting and/or important examples where one compares
implication of different sigma-fields on the same set? I suspect one of
the issues here is that with unrelated sigma-fields, one cannot compare
measures.
Here's one (Blackwell's theorem): Let (Omega, F) be a countably
generated measurable space with the property that for each F-measurable
map f: Omega -> R, and each B in F, the image f(B) is an analytic subset
of R. Let G and H be sub sigma fields of F, H assumed to be countably
generated. Then G is contained in H if and only if each G-atom is a
union of H-atoms. [The G-atom (for example) containing x (a point of
Omega) is the intersection of all sets in G that contain x.]
Corollary: A random variable X (aka, an F-measurable function from Omega
to R) is H-measurable if and only if X is constant on each H-atom.
--
A. |
|
|
| Back to top |
|
| Stephen J. Herschkorn |
Posted: Fri Dec 19, 2003 2:38 am |
|
|
|
Guest
|
ArtflDodgr wrote:
Quote: [... = my original post]
Here's one (Blackwell's theorem): Let (Omega, F) be a countably
generated measurable space with the property that for each F-measurable
map f: Omega -> R, and each B in F, the image f(B) is an analytic subset
of R. Let G and H be sub sigma fields of F, H assumed to be countably
generated. Then G is contained in H if and only if each G-atom is a
union of H-atoms. [The G-atom (for example) containing x (a point of
Omega) is the intersection of all sets in G that contain x.]
Sorry for my ignorance, but what is an analytic subset? (I know what an
analytic *function* is.)
Quote:
Corollary: A random variable X (aka, an F-measurable function from Omega
to R) is H-measurable if and only if X is constant on each H-atom.
BTW, I thought of another example, again relating to conditional
expectaion: In the study of martigales, one refers to filtrations,
i.e., increasing sequences of sub-sigma-fields of a given probability space.
--
Stephen J. Herschkorn herschko@rutcor.rutgers.edu |
|
|
| Back to top |
|
| G. A. Edgar |
Posted: Fri Dec 19, 2003 8:12 am |
|
|
|
Guest
|
In article <3FE2AB51.2060901@rutcor.rutgers.edu>, Stephen J. Herschkorn
<herschko@rutcor.rutgers.edu> wrote:
Quote: ArtflDodgr wrote:
[... = my original post]
Here's one (Blackwell's theorem): Let (Omega, F) be a countably
generated measurable space with the property that for each F-measurable
map f: Omega -> R, and each B in F, the image f(B) is an analytic subset
of R. Let G and H be sub sigma fields of F, H assumed to be countably
generated. Then G is contained in H if and only if each G-atom is a
union of H-atoms. [The G-atom (for example) containing x (a point of
Omega) is the intersection of all sets in G that contain x.]
Sorry for my ignorance, but what is an analytic subset? (I know what an
analytic *function* is.)
An analytic subset of R is a continuous image of a Borel set.
Quote:
Corollary: A random variable X (aka, an F-measurable function from Omega
to R) is H-measurable if and only if X is constant on each H-atom.
--
G. A. Edgar http://www.math.ohio-state.edu/~edgar/ |
|
|
| Back to top |
|
| |
|
Page 1 of 1
All times are GMT - 5 Hours
The time now is Wed Oct 15, 2008 3:23 pm
|
|