Kolmogorov inequality

In probability theory, Kolmogorov's inequality is a so-called "maximal inequality" that gives a bound on the probability that the partial sums of a finite collection of independent random variables exceed some specified bound.

Statement of the inequality

Let ''X''₁, ..., ''X''_''n'' : Ω → R be independent random variables defined on a common probability space (Ω, ''F'', Pr), with expected value E 'X''_''k''nbsp;= 0 and variance Var 'X''_''k''nbsp;< +∞ for ''k'' = 1, ..., ''n''. Then, for each λ > 0,

:\Pr \left(\max_ , S_k , \geq\lambda\right)\leq \frac \operatorname _n\equiv \frac\sum_^n \operatorname _k\frac\sum_^\text _k^2

where ''S''_''k'' = ''X''₁ + ... + ''X''_''k''.

The convenience of this result is that we can bound the worst case deviation of a random walk at any point of time using its value at the end of time interval.

Proof

The following argument employs discrete martingales.
As argued in the discussion of Doob's martingale inequality, the sequence $S_1, S_2, \dots, S_n$ is a martingale.
Define $(Z_i)_^n$ as follows. Let $Z_0 = 0$, and
:$Z_ = \left\{ \begin{array}{ll} S_{i+1} & \text{ if } \displaystyle \max_{1 \leq j \leq i} , S_j , < \lambda \\ Z_i & \text{ otherwise} \end{array} \right.$
for all $i$.
Then $(Z_i)_{i=0}^n$ is also a martingale.

For any martingale $M_i$ with $M_0 = 0$, we have that

\begin{align}
\sum_{i=1}^n \text{E} (M_i - M_{i-1})^2&= \sum_{i=1}^n \text{E} M_i^2 - 2 M_i M_{i-1} + M_{i-1}^2 \\
&= \sum_{i=1}^n \text{E}\left M_i^2 - 2 (M_{i-1} + M_{i} - M_{i-1}) M_{i-1} + M_{i-1}^2 \right\\
&= \sum_{i=1}^n \text{E}\left M_i^2 - M_{i-1}^2 \right- 2\text{E}\left M_{i-1} (M_{i}-M_{i-1})\right\
&= \text{E} _n^2- \text{E} _0^2= \text{E} _n^2
\end{align}

Applying this result to the martingale $(S_i)_{i=0}^n$, we have

\begin{align}
\text{Pr}\left( \max_{1 \leq i \leq n} , S_i, \geq \lambda\right) &=
\text{Pr} _n^2=\frac{1}{\lambda^2}_\sum_{i=1}^n_\text{E} Z_i_-_Z_{i-1})^2\\
&\leq_\frac{1}{\lambda^2}_\sum_{i=1}^n_\text{E} S_i_-_S_{i-1})^2=\frac{1}{\lambda^2}_\text{E} _n^2=_\frac{1}{\lambda^2}_\text{Var} _n\end{align}

where_the_first_inequality_follows_by_Chebyshev's_inequality.

This_inequality_was_generalized_by_Hájek_and_Rényi_in_1955.

_See_also

*_Chebyshev's_inequality
*_Etemadi's_inequality
*_Landau–Kolmogorov_inequality
*_Markov's_inequality
*_Bernstein_inequalities_(probability_theory)

_References

*__(Theorem_22.4)
*_
*_

{{PlanetMath_attribution, id=3687, title=Kolmogorov's_inequality

Stochastic_processes
Probabilistic_inequalities
Articles_containing_proofshtml" ;"title="Z_n, \geq \lambda] \\
&\leq \frac{1}{\lambda^2} \text{E} _n^2=\frac{1}{\lambda^2} \sum_{i=1}^n \text{E} Z_i - Z_{i-1})^2\\
&\leq \frac{1}{\lambda^2} \sum_{i=1}^n \text{E} S_i - S_{i-1})^2=\frac{1}{\lambda^2} \text{E} _n^2= \frac{1}{\lambda^2} \text{Var} _n\end{align}

where the first inequality follows by Chebyshev's inequality.

This inequality was generalized by Hájek and Rényi in 1955.

See also

* Chebyshev's inequality
* Etemadi's inequality
* Landau–Kolmogorov inequality
* Markov's inequality
* Bernstein inequalities (probability theory)

References

* (Theorem 22.4)
*
*

{{PlanetMath attribution, id=3687, title=Kolmogorov's inequality

Stochastic processes
Probabilistic inequalities
Articles containing proofs>Z_n, \geq \lambda\\
&\leq \frac{1}{\lambda^2} \text{E} _n^2=\frac{1}{\lambda^2} \sum_{i=1}^n \text{E} Z_i - Z_{i-1})^2\\
&\leq \frac{1}{\lambda^2} \sum_{i=1}^n \text{E} S_i - S_{i-1})^2=\frac{1}{\lambda^2} \text{E} _n^2= \frac{1}{\lambda^2} \text{Var} _n\end{align}

where the first inequality follows by Chebyshev's inequality.

This inequality was generalized by Hájek and Rényi in 1955.

See also

* Chebyshev's inequality
* Etemadi's inequality
* Landau–Kolmogorov inequality
* Markov's inequality
* Bernstein inequalities (probability theory)

References

* (Theorem 22.4)
*
*

{{PlanetMath attribution, id=3687, title=Kolmogorov's inequality

Stochastic processes
Probabilistic inequalities
Articles containing proofs