ITER COPY REPORT DOCUMENTATION PAGE 2L es

_UECSSOFTED

ITER

C

%F=Y7mTY CASSWON OF TMIS FACIE

COPY

FOR REPRODUCTION PURPOSES

-

REPORT DOCUMENTATION PAGE lb. RESTRICTIVE MARKINGS

I. REPORT SECURITY CLASSIFICATION

za.

SECUI[T CLASIICATON A 2b. OECLASSIPICATION/OOWNG

TH

DSTRIlUTKN/NAIULAIIUTY OF REPORT

-. .3. 1DULE

Approved for public release; unlimited. 0 NOV1988distribution

4. PERFORMING ORGANIZATION RES

MIERMS

S. MONITORING ORGANIZATION REPORT NUMBER(S)

C6 NAME OF PERFORMING ORGANIZATION

rARO

&b. OFFICE SYMBOl.

22406.2-MA

7a. NAME OF MONITORING ORGANIZATION

U. S. Army Research Office

Rice University -ADDRESS (Oly, State. ad ZIP Co*)

00

7b. ADDRESS (01y, State, and ZIP Code)

P.O. Box 1892 Houston, TX 77251

P. 0. Box 12211 Research Triangle Park, NC

S,. ORGANIZATION NAME OF FUNDINGISPONSORING U. S.

8b. OFFICE SYMBOL

Of OApkAh)

Army Research Office

" ADDRESS (C ly, State. and ZIP Code)

27709-2211

9. PROCUREMENT INSTRUMENT IDENTIFICATION NUMBER

DAAG29-85-K-0212 10. SOURCE OF FUNDING NUMBERS

P. 0. Box 12211

PROGRAM

Research Triangle Park, NC

ELEMENT NO.

27709-2211

PROJECT

IWORK UNIT

TASK

NO.

I NO.CESSION

1. TITLE (Inchud Selmwy ClaffifCaton)

Some Problems in Data Based Computer Graphics & Simulation 12. PERSONAL AUTHOR($)

James R. Thompson 13a. TYPE OF REPORT

114.

13b. TIME COVERED

I

DATE OF REPORT 1ear, Month Day)

S. PAGE COUNT

I 1 October 21, 1988 SUPPLEMENTARY NOTATION The view, opinions and/or findings contained in this report are those

Final

of the authqr(

FROM 8/1I8

).and should not be constlued as an 9fficial De artment of the Army position,

COSATI CODES

17.

FIELD

If necm wy and klewfly by block numbe) SIMDAT,SIMEST, simulation, density est~mation, time series, statistical computing

18. SUBJECT TERMS (Concri. on revei

SUB-GROUP

GROUP

19. ABSTRACT (Contkwl

TO7/31f8

an revne f necewry and Idendt

by blo¢k nmber)

This has been a three year study in the investigation of topics in the general area of statistical computing. We have developed, jointly with Dr. Malcolm Taylor of BRL the SIMDAT algorithm for using a set of projectile data to generate other data sets which could have happened. Also, we have developed a simulation based technique, SIMEST, to handle problems not tractable via the closed form maximum likelihood approach. We have obtained automated procedures for selecting bandwidths in the kernel density estimation. We have developed graphical techniques for demonstrating density estimates in dimensions as large as 6. We have done extensive work to build algorithms whose function is the creation of multicolored graphical displays of density estimates. We have done work in parameter estimation for time series pr blems using both frequency based and time based approaches. {,