skip to main content
OSTI.GOV title logo U.S. Department of Energy
Office of Scientific and Technical Information

Title: Recursive recovery of Markov transition probabilities from boundary value data

Thesis/Dissertation ·
DOI:https://doi.org/10.2172/10150977· OSTI ID:10150977
 [1]
  1. Univ. of California, Berkeley, CA (United States)

In an effort to mathematically describe the anisotropic diffusion of infrared radiation in biological tissue Gruenbaum posed an anisotropic diffusion boundary value problem in 1989. In order to accommodate anisotropy, he discretized the temporal as well as the spatial domain. The probabilistic interpretation of the diffusion equation is retained; radiation is assumed to travel according to a random walk (of sorts). In this random walk the probabilities with which photons change direction depend upon their previous as well as present location. The forward problem gives boundary value data as a function of the Markov transition probabilities. The inverse problem requires finding the transition probabilities from boundary value data. Problems in the plane are studied carefully in this thesis. Consistency conditions amongst the data are derived. These conditions have two effects: they prohibit inversion of the forward map but permit smoothing of noisy data. Next, a recursive algorithm which yields a family of solutions to the inverse problem is detailed. This algorithm takes advantage of all independent data and generates a system of highly nonlinear algebraic equations. Pluecker-Grassmann relations are instrumental in simplifying the equations. The algorithm is used to solve the 4 x 4 problem. Finally, the smallest nontrivial problem in three dimensions, the 2 x 2 x 2 problem, is solved.

Research Organization:
Lawrence Berkeley National Lab. (LBNL), Berkeley, CA (United States)
Sponsoring Organization:
USDOE; Department of Defense (DOD); National Aeronautics and Space Administration (NASA); National Science Foundation (NSF)
DOE Contract Number:
AC03-76SF00098
OSTI ID:
10150977
Report Number(s):
LBL-35495; ON: DE94011914; CNN: Contract FDF-49620-92-J-0067-11792; Grant NAG3-1143; Grant DMS89-02831
Resource Relation:
Other Information: TH: Thesis (Ph.D.); PBD: Apr 1994
Country of Publication:
United States
Language:
English