You are on page 1of 3

Disintegration theorem

In mathematics, the disintegration theorem is a result in


measure theory and probability theory. It rigorously denes the idea of a non-trivial restriction of a measure to
a measure zero subset of the measure space in question. It
is related to the existence of conditional probability measures. In a sense, disintegration is the opposite process
to the construction of a product measure.

measure). Let P(Y), let : Y X be a Borelmeasurable function, and let P(X) be the pushforward
measure = () = 1 . Then there exists a almost everywhere uniquely determined family of probability measures {x}xX P(Y) such that
the function x 7 x is Borel measurable, in the
sense that x 7 x (B) is a Borel-measurable function for each Borel-measurable set B Y;

Motivation

x lives on the ber 1 (x): for -almost all x


X,

Consider the unit square in the Euclidean plane R2 , S =


[0, 1] [0, 1]. Consider the probability measure dened on S by the restriction of two-dimensional Lebesgue
measure 2 to S. That is, the probability of an event E
S is simply the area of E. We assume E is a measurable
subset of S.

(
)
x Y \ 1 (x) = 0,
and so x(E) = x(E 1 (x));

Consider a one-dimensional subset of S such as the line


segment Lx = {x} [0, 1]. Lx has -measure zero; every
subset of Lx is a -null set; since the Lebesgue measure
space is a complete measure space,

for every Borel-measurable function f : Y [0, ],

E Lx = (E) = 0.


f (y) d(y) =
Y

While true, this is somewhat unsatisfying. It would be


nice to say that restricted to Lx is the one-dimensional
Lebesgue measure 1 , rather than the zero measure. The
probability of a two-dimensional event E could then be
obtained as an integral of the one-dimensional probabilities of the vertical slices E Lx: more formally, if x
denotes one-dimensional Lebesgue measure on Lx, then

1 (x)

f (y) dx (y)d(x).

In particular, for any event E Y, taking f to


be the indicator function of E,[1]

x (E) d(x).

(E) =
X

x (E Lx ) dx

(E) =

3 Applications

[0,1]

for any nice E S. The disintegration theorem makes


this argument rigorous in the context of measures on 3.1 Product spaces
metric spaces.
The original example was a special case of the problem
of product spaces, to which the disintegration theorem
applies.

Statement of the theorem

When Y is written as a Cartesian product Y = X1 X2


and i : Y Xi is the natural projection, then each bre
(Hereafter, P(X) will denote the collection of Borel prob- 1 (x ) can be canonically identied with X and there
2
1
1
ability measures on a metric space (X, d).)
exists a Borel family of probability measures {x1 }x1 X1
Let Y and X be two Radon spaces (i.e. separable met- in P(X2 ) (which is (1 )()-almost everywhere uniquely
ric spaces on which every probability measure is a Radon determined) such that
1

(
)
x1 11 (dx1 ) =

=
X1

x1 d(1 ) ()(x1 ),
X1

which is in particular

X1 X2

f (x1 , x2 ) (dx1 , dx2 ) =


X1

and

(
)
(B|x1 ) 11 (dx1 ) .

(A B) =
A

The relation to conditional expectation is given by the


identities

E(f |1 )(x1 ) =

f (x1 , x2 )(dx2 |x1 ),


X2

(A B|1 )(x1 ) = 1A (x1 ) (B|x1 ).

3.2

Vector calculus

The disintegration theorem can also be seen as justifying


the use of a restricted measure in vector calculus. For
instance, in Stokes theorem as applied to a vector eld
owing through a compact surface R3 , it is implicit
that the correct measure on is the disintegration of
three-dimensional Lebesgue measure 3 on , and that
the disintegration of this measure on is the same as
the disintegration of 3 on .[2]

3.3

Conditional distributions

The disintegration theorem can be applied to give a rigorous treatment of conditional probability distributions in
statistics, while avoiding purely abstract formulations of
conditional probability.[3]

See also
Joint probability distribution
Copula (statistics)
Conditional expectation

[2] Ambrosio, L., Gigli, N. & Savar, G. (2005). Gradient


Flows in Metric Spaces and in the Space of Probability
Measures. ETH Zrich, Birkhuser Verlag, Basel. ISBN
3-7643-2428-7.
[3] Chang, J.T.; Pollard, D. (1997). Conditioning as disintegration (PDF). Statistica Neerlandica 51 (3): 287.
doi:10.1111/1467-9574.00056.
)

(
)
f (x1 , x2 )(dx2 |x1 ) 11 (dx1 )

X2

References

[1] Dellacherie, C. & Meyer, P.-A. (1978). Probabilities and


potential. North-Holland Mathematics Studies, NorthHolland Publishing Co., Amsterdam.

REFERENCES

Text and image sources, contributors, and licenses

6.1

Text

Disintegration theorem Source: https://en.wikipedia.org/wiki/Disintegration_theorem?oldid=708140736 Contributors: Giftlite, Tsirel,


Grafen, SmackBot, RDBury, A. Pichler, Sullivan.t.j, David Eppstein, Pjoef, Melcombe, Citation bot, Andr Caldas, Peter baum, Loopy48,
Dman445, Brad7777, Marc.coram, BattyBot, Illia Connell, Monkbot and Anonymous: 6

6.2

Images

File:Text_document_with_red_question_mark.svg Source: https://upload.wikimedia.org/wikipedia/commons/a/a4/Text_document_


with_red_question_mark.svg License: Public domain Contributors: Created by bdesham with Inkscape; based upon Text-x-generic.svg
from the Tango project. Original artist: Benjamin D. Esham (bdesham)

6.3

Content license

Creative Commons Attribution-Share Alike 3.0