If a probability density p(x) (x ∈ ℝk) is bounded and R(t) := ∫e〈x, tu〉p(x)dx < ∞ for some linear functional u and all t ∈ (0,1), then, for each t ∈ (0,1) and all large enough n, the n-fold convolution of the t-tilted density ˜pt := e〈x, tu〉p(x)/R(t) is bounded. This is a corollary of a general, “non-i.i.d.” result, which is also shown to enjoy a certain optimality property. Such results and their corollaries stated in terms of the absolute integrability of the corresponding characteristic functions are useful for saddle-point approximations.
Mots-clés : probability density, saddle-point approximation, sums of independent random variables/vectors, convolution, exponential integrability, boundedness, exponential tilting, exponential families, absolute integrability, characteristic functions
@article{PS_2012__16__86_0, author = {Pinelis, Iosif}, title = {Exponential deficiency of convolutions of densities}, journal = {ESAIM: Probability and Statistics}, pages = {86--96}, publisher = {EDP-Sciences}, volume = {16}, year = {2012}, doi = {10.1051/ps/2010010}, mrnumber = {2946121}, zbl = {1266.60021}, language = {en}, url = {http://archive.numdam.org/articles/10.1051/ps/2010010/} }
Pinelis, Iosif. Exponential deficiency of convolutions of densities. ESAIM: Probability and Statistics, Tome 16 (2012), pp. 86-96. doi : 10.1051/ps/2010010. http://archive.numdam.org/articles/10.1051/ps/2010010/
[1] Edgeworth and saddle-point approximations with statistical applications. J. R. Stat. Soc., Ser. B 41 (1979) 279-312. With discussion. | MR | Zbl
and ,[2] Normal approximation and asymptotic expansions. Robert E. Krieger Publishing Co. Inc., Melbourne, FL (1986). Reprint of the 1976 original. | MR | Zbl
and ,[3] Tail probability approximations. Int. Stat. Rev. 55 (1987) 37-48. | MR | Zbl
,[4] On convolution tails. Stoch. Proc. Appl. 13 (1982) 263-278. | MR | Zbl
and ,[5] Saddlepoint approximation for Student's t-statistic with no moment conditions. Ann. Stat. 32 (2004) 2679-2711. | MR | Zbl
, and ,[6] Subexponential distributions and characterizations of related classes. Probab. Theory Relat. Fields 82 (1989) 259-269. | MR | Zbl
,[7] Saddle point approximation for the distribution of the sum of independent random variables. Adv. Appl. Probab. 12 (1980) 475-490. | MR | Zbl
and ,[8] Asymptotic equivalence of the probabilities of large deviations for sums and maximum of independent random variables, in Limit theorems of probability theory. “Nauka” Sibirsk. Otdel., Novosibirsk. Trudy Inst. Mat. 5 (1985) 144-173, 176. | MR | Zbl
,[9] Saddlepoint methods and statistical inference. Stat. Sci. 3 (1988) 213-238. With comments and a rejoinder by the author. | MR | Zbl
,[10] Recent progress on self-normalized limit theorems, in Probability, finance and insurance. World Sci. Publ., River Edge, NJ (2004) 50-68. | MR
,Cité par Sources :