## Thesis

My Ph. D. thesis has been unavailable except by interlibrary loan from the University of Washington library because I stupidly deleted the original PostScript version years ago thinking I could recreate it if I ever wanted to, but after LaTeX2e came in, I no longer could.

There is some good stuff in it that has never been published. Several people have asked me for copies over the years and we had to make do with photocopies. I have published a bit more of it, and hence have finally recreated it. This required installing LaTeX 2.09 and redoing all the figures, which were not valid Encapsulated PostScript.

Unfortunately, LaTeX 2.09 installed on top of a modern TeX with modern computer modern fonts does not paginate exactly like the installation I used in 1990 at UW. Thus the page numbers are not those of the original. Anyone citing anything in the PDF linked below should not cite page numbers. The text, however, is unchanged. Every word or symbol and every section, theorem, equation, figure, or table number is as in the original.

The URL below will be permanent, so this thesis can be cited as follows.

Charles J. Geyer (1990).

Likelihood and Exponential Families.

Ph. D. Thesis, Department of Statistics, University of Washington.

http://www.stat.umn.edu/geyer/thesis/th.pdf

Even more permanent, the thesis has been put on the University of Minnesota Digital Conservancy.

## Papers

Chapter 7 of the thesis (somewhat revised) appeared as

Geyer, Charles J. (1991).

Constrained Maximum Likelihood Exemplified by Isotonic Convex Logistic Regression

Journal of the American Statistical Association,86, 717–724.

JSTOR link

Chapter 6 of the thesis (somewhat revised) appeared as

Geyer, Charles J. and Thompson, Elizabeth A. (1992).

Constrained Monte Carlo Maximum Likelihood for Dependent Data (with discussion).

Journal of the Royal Statistical Society. Series B,54, 657–699.

JSTOR link

The paper above and Chapter 6 of the thesis contained a simulated data example. The real DNA fingerprinting analysis appeared in

Geyer, C. J., Ryder, O. A., Chemnick, L. G. and Thompson, E. A. (1993).

Analysis of Relatedness in the California Condors, from DNA Fingerprints

Molecular Biology and Evolution,10, 571–589.

Oxford Journals link

The Monte Carlo likelihood convergence theory in Chapter 6 of the thesis was a very special case of some of the theory in

Geyer, Charles J. (1994).

On the Convergence of Monte Carlo Maximum Likelihood Calculations

Journal of the Royal Statistical Society. Series B,56, 261–274.

JSTOR link

The theory of maximum likelihood estimation in the Barndorff-Nielsen
completion of a convex exponential family in Chapter 2 of the thesis
has been partially replaced by a simpler theory applicable only to full
exponential families (so the theory in the thesis is still necessary for
non-full, convex families) devised for efficient calculation in R using the
contributed package `rcdd`.

Geyer, Charles J. (2009).

Likelihood inference in exponential families and directions of recession.

Electronic Journal of Statistics,3, 259–289 (electronic).

EJS link