On Large Deviations, I
The paper presents a new point of view on the Cramér type large deviations \[ P(\xi >x)=(1-\Phi (x))Q\exp(L) \] for the ``right tail of the distribution of a random variable \(\xi\). Here \(\Phi (x)\) is a standard normal distribution function, \(Q\) and \(L\) are some quantities depending on \(x\) and \(\xi\). A simple proof of a known representation of large deviations is given under the Cramér condition \(E e^{\lambda\xi}<\infty\). The explicit expressions for \(Q\) and \(L\) are written using the auxiliary parameter \(\lambda \) instead of \(x\): \(Q=Q(\lambda)\), \(L=L(\lambda)\), \(x=x(\lambda)\). The obtained expressions for large deviations do not contain the cumulants of the random variable \(\xi\) as usual. It is shown, that the traditional form of writing the exponent \(L\) by the Cramér series essentially narrows its natural domain of definition. The ``canonical form of the Cramér series is suggested. The relationships between the introduced functions and the deviation function are discussed.
- On Large Deviations, II
- scientific article; zbMATH DE number 3907466
- Cramer Large Deviations when the Extreme Conjugate Distribution is Heavy-tailed
- Большие уклонения случайной величины с конечным числом семиинвариантов, значения которых вычислены приближенно
- Cramér type large deviations for a class of statistics
- Large deviation estimates involving deformed exponential functions
- Comment on ``Towards a large deviation theory for strongly correlated systems
- Large deviations. The simplest case.
- Optimal investment and asymmetric risk: a large deviations approach
- Large deviations theorems for optimal investment problems with large portfolios
- Cramér Asymptotics in a System with Slow and Fast Markovian Motions
- On Large Deviations, II
- On the large deviation principle for the almost sure CLT.
This page was built for publication: On Large Deviations, I
Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q4510005)
