|
|
-
In theory
Significant contributions to theoretical statistics made by Soumendra
Lahiri.
-
One of the foremost young theoretical statisticians of his time.
That's the reputation Soumendra Lahiri, professor of statistics, has established
in his time on the faculty of the Department of Statistics.
That's the opinion of some of the world’s most renowned theoretical
statisticians including Peter Hall of the Australian National University
who describes Lahiri's work as of "a deep and very sophisticated
theoretical nature." He also says "the breadth of his present
work reflects a marked eclectism such as rarely found in a young statistical
scientist."
Lahiri's research efforts have also been recognized on and off campus.
In 2002 he was named a Fellow of the American Statistical Association.
The previous year he was awarded the same honor from the Institute of
Mathematical Statistics.
Earlier in his Iowa State career, he received the ISU Foundation Award
for Early Achievement in Research and Scholarship.
And just last spring, the College of Liberal Arts and Sciences named Lahiri
a recipient of the college's award for Outstanding Achievement in Research/Artistic
Creativity. This award is given annually to LAS faculty members for their
national or international reputations for contributions in research, and
who have influenced the research activities of students.
Each time Lahiri was recognized for his important basic research contributions
in the areas of resampling methods, long-range dependence, nonparametric
curve estimation, and spatial and environmental statistics.
" My work is mostly theoretical," Lahiri says. "I study
theoretical properties of statistical methods and use them to refine and
develop the tools that others in a wide variety of disciplines can use."
Indeed, Lahiri's work has made a substantial impact on economists, environmental
scientists and other researchers who analyze and model time series data
and data with spatial dependencies. His research indicates that statistical
re-sampling procedures originally developed for sets of independent observations
can be extended to make inferences from data that exhibit dependencies
across time or space.
In particular Lahiri has worked on four fundamental areas:
*The bootstrap - a non-parametric method of estimating any characteristics
of the sampling distribution of a statistic without making any restrictive
structure or model assumptions. One of the most active research areas
in modern statistics, Lahiri gave the first theoretical confirmation of
the superiority of block bootstrap methods. He also developed the first
theoretical result on relative merits of different block bootstrap methods,
and identified the best bootstrap method among the existing ones. He recently
received a $225,000 NSF grant to study the high order accuracy of bootstrap
methods for temporal and spatial processes.
* Asymptotic expansion - an approach that allows researchers to study
the accuracy of an estimator and to construct statistical inference methods
that are more accurate than potential alternatives. Lahiri settled a long-standing
open problem that had been expressed as a conjecture.
*Long range dependence in time series. His recent work provided a unification
and characterization of the “asymptotic independence” property
of Discrete Fourier Transforms for weakly and strongly dependent random
processes.
*Spatial data - Lahiri's development of a new resampling method for spatial
data has immediate applications in environmental and ecological resource
monitoring.
Around LAS
September 22 to October 5, 2003
|
|
