Convergence of random variables
In probability theory, there exist several different notions of convergence of random variables. The convergence of sequences of random variables to some limit
random variable is an important concept in probability theory, and its applications to statistics and stochastic processes. The same concepts are known in more
general mathematics as stochastic convergence and they formalize the idea that a sequence of essentially random or unpredictable events can sometimes be
expected to settle down into a behavior that is essentially unchanging when items far enough into the sequence are studied. The different possible notions of
convergence relate to how such a behavior can be characterized: two readily understood behaviors are that the sequence eventually takes a constant value, and that
values in the sequence continue to change but can be described by an unchanging probability distribution.
Background
"Stochastic convergence" formalizes the idea that a sequence of essentially random or unpredictable events can sometimes be expected to settle into a pattern. The
pattern may for instance be
Convergence in the classical sense to a fixed value, perhaps itself coming from a random event
An increasing similarity of outcomes to what a purely deterministic function would produce
An increasing preference towards a certain outcome
An increasing "aversion" against straying far away from a certain outcome
That the probability distribution describing the next outcome may grow increasingly similar to a certain distribution
Some less obvious, more theoretical patterns could be
That the series formed by calculating the expected value of the outcome's distance from a particular value may converge to 0
That the variance of the random variable describing the next event grows smaller and smaller.
These other types of patterns that may arise are reflected in the different types of stochastic convergence that have been studied.
While the above discussion has related to the convergence of a single series to a limiting value, the notion of the convergence of two series towards each other is
also important, but this is easily handled by studying the sequence defined as either the difference or the ratio of the two series.
For example, if the average of n independent random variables Yi, i = 1, ..., n, all having the same finite mean and variance, is given by
then as n tends to infinity, Xn converges in probability (see below) to the common mean, μ, of the random variables Yi. This result is known as the weak law of
large numbers. Other forms of convergence are important in other useful theorems, including the central limit theorem.
Throughout the following, we assume that (Xn ) is a sequence of random variables, and X is a random variable, and all of them are defined on the same probability
space .
Convergence in distribution
With this mode of convergence, we increasingly expect to see the next outcome in a sequence of random experiments Examples of convergence
becoming better and better modeled by a given probability distribution. in distribution
Convergence in distribution is the weakest form of convergence typically discussed, since it is implied by all other types of Dice factory
convergence mentioned in this article. However, convergence in distribution is very frequently used in practice; most often Suppose a new dice factory has just
it arises from application of the central limit theorem. been built. The first few dice come
out quite biased, due to
Definition imperfections in the production
process. The outcome from tossing
A sequence of real-valued random variables, with cumulative distribution functions , is said to any of them will follow a distribution
converge in distribution, or converge weakly, or converge in law to a random variable X with cumulative distribution markedly different from the desired
function F if uniform distribution.
As the factory is improved, the dice
become less and less loaded, and
for every number at which F is continuous. the outcomes from tossing a newly
produced die will follow the uniform
The requirement that only the continuity points of F should be considered is essential. For example, if Xn are distributed
distribution more and more closely.
1
uniformly on intervals (0, n ), then this sequence converges in distribution to the degenerate random variable X = 0 .
Tossing coins
1
Indeed, Fn(x) = 0 for all n when x ≤ 0 , and Fn(x) = 1 for all x ≥ n when n > 0 . However, for this limiting random Let Xn be the fraction of heads after
variable F(0) = 1 , even though Fn(0) = 0 for all n . Thus the convergence of cdfs fails at the point x = 0 where F is tossing up an unbiased coin n
discontinuous. times. Then X1 has the Bernoulli
distribution with expected value
Convergence in distribution may be denoted as μ = 0.5 and variance σ2 = 0.25 .
The subsequent random variables
X2, X3, ... will all be distributed
binomially.
As n grows larger, this distribution
will gradually start to take shape
more and more similar to the bell
curve of the normal distribution. If
we shift and rescale Xn
appropriately, then will
be converging in distribution to
the standard normal, the result that
follows from the celebrated central
limit theorem.
Graphic example
Suppose {Xi} is an iid sequence of
uniform U(−1, 1) random
variables. Let be their
(normalized) sums. Then according
to the central limit theorem, the
distribution of Zn approaches the
1
normal N(0, 3 ) distribution. This
convergence is shown in the
picture: as n grows larger, the
shape of the probability density
function gets closer and closer to
the Gaussian curve.
(1)
where is the law (probability distribution) of X. For example, if X is standard normal we can write .
For random vectors {X1, X2, ...} ⊂ Rk the convergence in distribution is defined similarly. We say that this sequence converges in distribution to a random k-
vector X if
for every A ⊂ Rk which is a continuity set of X.
The definition of convergence in distribution may be extended from random vectors to more general random elements in arbitrary metric spaces, and even to the
“random variables” which are not measurable — a situation which occurs for example in the study of empirical processes. This is the “weak convergence of laws
without laws being defined” — except asymptotically.[1]
In this case the term weak convergence is preferable (see weak convergence of measures), and we say that a sequence of random elements {Xn} converges
weakly to X (denoted as Xn ⇒ X) if
for all continuous bounded functions h .[2] Here E* denotes the outer expectation, that is the expectation of a “smallest measurable function g that dominates
h(Xn)”.
Properties
Since , the convergence in distribution means that the probability for Xn to be in a given range is approximately equal to the
probability that the value of X is in that range, provided n is sufficiently large.
In general, convergence in distribution does not imply that the sequence of corresponding probability density functions will also converge. As
an example one may consider random variables with densities fn(x) = (1 + cos(2πnx))1 (0,1). These random variables converge in
distribution to a uniform U(0, 1), whereas their densities do not converge at all.[3]
However, according to Scheffé’s theorem, convergence of the probability density functions implies convergence in distribution.[4]
The portmanteau lemma provides several equivalent definitions of convergence in distribution. Although these definitions are less intuitive,
they are used to prove a number of statistical theorems. The lemma states that {Xn} converges in distribution to X if and only if any of the
following statements are true:[5]
for all continuity points of ;
for all bounded, continuous functions (where denotes the expected value operator);
for all bounded, Lipschitz functions ;
for all nonnegative, continuous functions ;
for every open set ;
for every closed set ;
for all continuity sets of random variable ;
for every upper semi-continuous function bounded above;
for every lower semi-continuous function bounded below.
The continuous mapping theorem states that for a continuous function g , if the sequence {Xn} converges in distribution to X, then {g(Xn)}
converges in distribution to g(X).
Note however that convergence in distribution of {Xn} to X and {Y n} to Y does in general not imply convergence in distribution of
{Xn + Yn} to X + Y or of {XnYn} to XY.
Lévy’s continuity theorem: The sequence {Xn} converges in distribution to X if and only if the sequence of corresponding characteristic
functions {φ n} converges pointwise to the characteristic function φ of X.
Convergence in distribution is metrizable by the Lévy–Prokhorov metric.
A natural link to convergence in distribution is the Skorokhod's representation theorem.
Convergence in probability
The basic idea behind this type of convergence is that the probability of an “unusual” outcome becomes smaller and Examples of convergence
smaller as the sequence progresses. in probability
The concept of convergence in probability is used very often in statistics. For example, an estimator is called consistent if Height of a person
it converges in probability to the quantity being estimated. Convergence in probability is also the type of convergence Consider the following experiment.
established by the weak law of large numbers. First, pick a random person in the
street. Let X be their height, which
Definition is ex ante a random variable. Then
ask other people to estimate this
A sequence {Xn } of random variables converges in probability towards the random variable X if for all ε > 0 height by eye. Let Xn be the
average of the first n responses.
Then (provided there is no
systematic error) by the law of large
More explicitly, let Pn (ε) be the probability that Xn is outside the ball of radius ε centered at X. Then Xn is said to numbers, the sequence Xn will
converge in probability to X if for any ε > 0 and any δ > 0 there exists a number N (which may depend on ε and δ) such converge in probability to the
that for all n ≥ N, Pn (ε) < δ (the definition of limit). random variable X.
Predicting random number
Notice that for the condition to be satisfied, it is not possible that for each n the random variables X and Xn are
generation
independent (and thus convergence in probability is a condition on the joint cdf's, as opposed to convergence in
Suppose that a random number
distribution, which is a condition on the individual cdf's), unless X is deterministic like for the weak law of large numbers.
At the same time, the case of a deterministic X cannot, whenever the deterministic value is a discontinuity point (not generator generates a
isolated), be handled by convergence in distribution, where discontinuity points have to be explicitly excluded. pseudorandom floating point
number between 0 and 1. Let
Convergence in probability is denoted by adding the letter p over an arrow indicating convergence, or using the "plim" random variable X represent the
probability limit operator: distribution of possible outputs by
the algorithm. Because the
pseudorandom number is
generated deterministically, its next
value is not truly random. Suppose
that as you observe a sequence of
randomly generated numbers, you
can deduce a pattern and make
increasingly accurate predictions as
to what the next randomly
generated number will be. Let Xn
be your guess of the value of the
next random number after
observing the first n random
numbers. As you learn the pattern
and your guesses become more
accurate, not only will the
distribution of Xn converge to the
distribution of X, but the outcomes
of Xn will converge to the outcomes
of X.
(2)
For random elements {Xn } on a separable metric space (S, d), convergence in probability is defined similarly by[6]
Properties
Convergence in probability implies convergence in distribution.[proof]
In the opposite direction, convergence in distribution implies convergence in probability when the limiting random variable X is a constant.
[proof]
Convergence in probability does not imply almost sure convergence.[proof]
The continuous mapping theorem states that for every continuous function , if , then also .
Convergence in probability defines a topology on the space of random variables over a fixed probability space. This topology is metrizable by
the Ky Fan metric:[7]
or alternately by this metric
Almost sure convergence
This is the type of stochastic convergence that is most similar to pointwise convergence known from elementary real Examples of almost sure
analysis. convergence
Example 1
Definition Consider an animal of some short-
lived species. We record the
To say that the sequence Xn converges almost surely or almost everywhere or with probability 1 or strongly towards amount of food that this animal
X means that consumes per day. This sequence
of numbers will be unpredictable,
but we may be quite certain that
one day the number will become
zero, and will stay zero forever
after.
This means that the values of Xn approach the value of X, in the sense that events for which Xn does not converge to X
Example 2
have probability 0 (see Almost surely). Using the probability space and the concept of the random variable as a
function from Ω to R, this is equivalent to the statement Consider a man who tosses seven
coins every morning. Each
afternoon, he donates one pound to
a charity for each head that
appeared. The first time the result
is all tails, however, he will stop
Using the notion of the limit superior of a sequence of sets, almost sure convergence can also be defined as follows: permanently.
Let X1, X2, … be the daily amounts
the charity received from him.
We may be almost sure that one
Almost sure convergence is often denoted by adding the letters a.s. over an arrow indicating convergence: day this amount will be zero, and
stay zero forever after that.
(3)
However, when we consider any
For generic random elements {Xn } on a metric space , convergence almost surely is defined similarly: finite number of days, there is a
nonzero probability the terminating
condition will not occur.
Properties
Almost sure convergence implies convergence in probability (by Fatou's lemma), and hence implies convergence in distribution. It is the
notion of convergence used in the strong law of large numbers.
The concept of almost sure convergence does not come from a topology on the space of random variables. This means there is no topology
on the space of random variables such that the almost surely convergent sequences are exactly the converging sequences with respect to
that topology. In particular, there is no metric of almost sure convergence.
Sure convergence or pointwise convergence
To say that the sequence of random variables (Xn ) defined over the same probability space (i.e., a random process) converges surely or everywhere or pointwise
towards X means
where Ω is the sample space of the underlying probability space over which the random variables are defined.
This is the notion of pointwise convergence of a sequence of functions extended to a sequence of random variables. (Note that random variables themselves are
functions).
Sure convergence of a random variable implies all the other kinds of convergence stated above, but there is no payoff in probability theory by using sure
convergence compared to using almost sure convergence. The difference between the two only exists on sets with probability zero. This is why the concept of sure
convergence of random variables is very rarely used.
Convergence in mean
Given a real number r ≥ 1 , we say that the sequence Xn converges in the r-th mean (or in the Lr-norm) towards the random variable X, if the r-th absolute
moments E(|Xn |r ) and E(|X|r ) of Xn and X exist, and
where the operator E denotes the expected value. Convergence in r-th mean tells us that the expectation of the r-th power of the difference between and
converges to zero.
This type of convergence is often denoted by adding the letter Lr over an arrow indicating convergence:
(4)
The most important cases of convergence in r-th mean are:
When Xn converges in r-th mean to X for r = 1, we say that Xn converges in mean to X.
When Xn converges in r-th mean to X for r = 2, we say that Xn converges in mean square (or in quadratic mean) to X.
Convergence in the r-th mean, for r ≥ 1, implies convergence in probability (by Markov's inequality). Furthermore, if r > s ≥ 1, convergence in r-th mean implies
convergence in s-th mean. Hence, convergence in mean square implies convergence in mean.
It is also worth noticing that if
,
(4)
then
Properties
Provided the probability space is complete:
If and , then almost surely.
If and , then almost surely.
If and , then almost surely.
If and , then (for any real numbers a and b ) and .
If and , then (for any real numbers a and b ) and .
If and , then (for any real numbers a and b ).
None of the above statements are true for convergence in distribution.
The chain of implications between the various notions of convergence are noted in their respective sections. They are, using the arrow notation:
These properties, together with a number of other special cases, are summarized in the following list:
Almost sure convergence implies convergence in probability:[8][proof]
Convergence in probability implies there exists a sub-sequence which almost surely converges:[9]
Convergence in probability implies convergence in distribution:[8][proof]
Convergence in r-th order mean implies convergence in probability:
Convergence in r-th order mean implies convergence in lower order mean, assuming that both orders are greater than or equal to one:
provided r ≥ s ≥ 1.
If Xn converges in distribution to a constant c, then Xn converges in probability to c:[8][proof]
provided c is a constant.
If Xn converges in distribution to X and the difference between Xn and Yn converges in probability to zero, then Yn also converges in
distribution to X:[8][proof]
If Xn converges in distribution to X and Yn converges in distribution to a constant c, then the joint vector (Xn, Y n) converges in distribution to
:[8][proof]
provided c is a constant.
Note that the condition that Y n converges to a constant is important, if it were to converge to a random variable Y then we wouldn't be
able to conclude that (Xn, Y n) converges to .
If Xn converges in probability to X and Yn converges in probability to Y, then the joint vector (Xn, Y n) converges in probability to (X, Y):[8][proof]
If Xn converges in probability to X, and if P(| Xn | ≤ b) = 1 for all n and some b, then Xn converges in rth mean to X for all r ≥ 1 . In other
words, if Xn converges in probability to X and all random variables Xn are almost surely bounded above and below, then Xn converges to X
also in any rth mean.[10]
Almost sure representation. Usually, convergence in distribution does not imply convergence almost surely. However, for a given sequence
{Xn} which converges in distribution to X0 it is always possible to find a new probability space (Ω, F, P) and random variables {Yn, n = 0, 1, ...}
defined on it such that Yn is equal in distribution to Xn for each n ≥ 0 , and Yn converges to Y0 almost surely.[11][12]
If for all ε > 0,
then we say that Xn converges almost completely, or almost in probability towards X. When Xn converges almost completely towards X
then it also converges almost surely to X. In other words, if Xn converges in probability to X sufficiently quickly (i.e. the above sequence
of tail probabilities is summable for all ε > 0 ), then Xn also converges almost surely to X. This is a direct implication from the Borel–
Cantelli lemma.
If S n is a sum of n real independent random variables:
then S n converges almost surely if and only if S n converges in probability.
The dominated convergence theorem gives sufficient conditions for almost sure convergence to imply L1-convergence:
(5)
A necessary and sufficient condition for L1 convergence is and the sequence (Xn) is uniformly integrable.
If are discrete and independent, then implies that . This is a consequence of the second Borel–Cantelli lemma.
See also
Proofs of convergence of random variables
Convergence of measures
Convergence in measure
Continuous stochastic process: the question of continuity of a stochastic process is essentially a question of convergence, and many of the
same concepts and relationships used above apply to the continuity question.
Asymptotic distribution
Big O in probability notation
Skorokhod's representation theorem
The Tweedie convergence theorem
Slutsky's theorem
Continuous mapping theorem
Notes
1. Bickel et al. 1998, A.8, page 475 7. Dudley 2002, p. 289
2. van der Vaart & Wellner 1996, p. 4 8. van der Vaart 1998, Theorem 2.7
3. Romano & Siegel 1985, Example 5.26 9. Gut, Allan (2005). Probability: A graduate course. Theorem 3.4:
4. Durrett, Rick (2010). Probability: Theory and Examples. p. 84. Springer. ISBN 978-0-387-22833-4.
5. van der Vaart 1998, Lemma 2.2 10. Grimmett & Stirzaker 2020, p. 354
6. Dudley 2002, Chapter 9.2, page 287 11. van der Vaart 1998, Th.2.19
12. Fristedt & Gray 1997, Theorem 14.5
References
Bickel, Peter J.; Klaassen, Chris A.J.; Ritov, Ya’acov; Wellner, Jon Ledoux, Michel; Talagrand, Michel (1991). Probability in Banach
A. (1998). Efficient and adaptive estimation for semiparametric spaces. Berlin: Springer-Verlag. pp. xii+480. ISBN 978-3-540-
models. New York: Springer-Verlag. ISBN 978-0-387-98473-5. 52013-9. MR 1102015 (https://mathscinet.ams.org/mathscinet-getite
Billingsley, Patrick (1986). Probability and Measure. Wiley Series in m?mr=1102015).
Probability and Mathematical Statistics (2nd ed.). Wiley. Romano, Joseph P.; Siegel, Andrew F. (1985). Counterexamples in
Billingsley, Patrick (1999). Convergence of probability measures (ht probability and statistics. Great Britain: Chapman & Hall. ISBN 978-
tps://archive.org/details/convergenceofpro0000bill) (2nd ed.). John 0-412-98901-8.
Wiley & Sons. pp. 1–28 (https://archive.org/details/convergenceofpr Grimmett, Geoffrey R.; Stirzaker, David R. (2020). Probability and
o0000bill/page/1). ISBN 978-0-471-19745-4. Random Processes (4th ed.). Oxford University Press. ISBN 978-0-
Dudley, R.M. (2002). Real analysis and probability. Cambridge, UK: 198-84760-1.
Cambridge University Press. ISBN 978-0-521-80972-6. van der Vaart, Aad W.; Wellner, Jon A. (1996). Weak convergence
Fristedt, Bert; Gray, Lawrence (1997). A Modern Approach to and empirical processes. New York: Springer-Verlag. ISBN 978-0-
Probability Theory. New York: Springer Science+Business Media. 387-94640-5.
doi:10.1007/978-1-4899-2837-5 (https://doi.org/10.1007%2F978-1- van der Vaart, Aad W. (1998). Asymptotic statistics. New York:
4899-2837-5). ISBN 978-1-4899-2837-5. Cambridge University Press. ISBN 978-0-521-49603-2.
Grimmett, G.R.; Stirzaker, D.R. (1992). Probability and random Williams, D. (1991). Probability with Martingales. Cambridge
processes (2nd ed.). Clarendon Press, Oxford. pp. 271–285. University Press. ISBN 978-0-521-40605-5.
ISBN 978-0-19-853665-9. Wong, E.; Hájek, B. (1985). Stochastic Processes in Engineering
Jacobsen, M. (1992). Videregående Sandsynlighedsregning Systems. New York: Springer–Verlag.
(Advanced Probability Theory) (3rd ed.). HCØ-tryk, Copenhagen.
pp. 18–20. ISBN 978-87-91180-71-2.
https://www.ma.utexas.edu/users/gordanz/notes/weak.pdf
This article incorporates material from the Citizendium article "Stochastic convergence", which is licensed under the Creative Commons Attribution-ShareAlike
3.0 Unported License but not under the GFDL.
Retrieved from "https://en.wikipedia.org/w/index.php?title=Convergence_of_random_variables&oldid=1158031313"
