Web Notes and Unpublished Papers
William H. Press
These are unpublished works, including tutorial and unfinished notes on technical topics, but also including recreational papers on various topics. These notes may be cited as "W.H. Press (unpublished paper)", with the appropriate reference to this page. Hopefully, each might be useful to at least someone!
Notes and Papers 2024
When to Bail from an Escalating Expense, What to do when they keep increasing the rent on your self-storage unit while advertising low prices for identical units. (last revised February 26, 2024)
Notes and Papers 2023
Distorted Maps of the United States, Github repository implementing in Python the method discussed in my note U.S. Demographic Quantities... (2010), below (last revised October 9, 2023)The Chatbots Debate (with a human moderator, GPT-4 and Claude-2 debate the future of AI) (last revised October 2, 2023)
Slides for talk: How Does Generative AI Actually Work (a quick introduction to Large Language Models) (last revised May 20, 2023)
Notes and Papers 2019
Principled Derivation of Some Moving Average (Sliding Window) Statistics (last revised February 26, 2019)
Notes and Papers 2016
Science to Action: Thoughts on Convincing a Skeptical Public, condensed and illustrated version of 2015 William D. Carey Lecture, American Association for the Advancement of Science. Also, PowerPoint slides for the lecture. (last revised April 14, 2016)Bayesians: Pay No Attention to Neyman-Scott (last revised March 14, 2016)
Notes and Papers 2013
Analytic Model for Acoustic Propagation in the Deep Sound Channel (last revised November 9, 2013). Also see the movie on YouTube.L'Huilier's Theorem, Spherical Triangles, and the Radius of the Earth (last revised May 25, 2013)
Notes and Papers 2012
C++ Program for Kemeny-Young Preference Aggregation (last revised August 6, 2012)
Notes and Papers 2011
Concise Genome Indexing by Decimated Hashing with Random Collision Resolution (last revised September 4, 2011)Canonical Correlation Clarified by Singular Value Decomposition (last revised May 28, 2011).
Includes J. Hussmann, "Why any orthonormal bases for the data matrices work for finding canonical correlations".Wright-Fisher Models, Approximations, and Minimum Increments of Evolution (last revised January 11, 2011)
U.S. Demographic Quantities Visualized by Cartographic Treemaps (last revised January 2, 2011)
Notes and Papers 2010
Simple Model for U.S. Income Inequality Data (last revised January 5, 2010)
Notes and Papers 2009
Did Dürer Intentionally Show Only His Second-Best Magic Square? (last revised December 25, 2009)Seemingly Remarkable Mathematical Coincidences Are Easy to Generate (last revised June 1, 2009)
Let Us Salute Scientist Citizens, Address to the College of Natural Sciences Convocation, The University of Texas at Austin (May 23, 2009)
Notes and Papers 2008
Confused about accuracy, precision, recall, TPR, FPR, specificity, sensitivity, ROC, and all that stuff? (last revised March 28, 2008)
Notes and Papers 2006
Program for characterizing gene populations by Gene Ontology word count statistics (last revised April 5, 2006)Reconstructing the Full Internal State of a Molecular Phylogenetic Tree with Arbitrary, Branch-Dependent Substitution Probabilities (last revised May 1, 2006)
Maximizing the Information Rate from Single-Molecule, Motion-Based DNA Sequencing Using RNA Polymerase (last revised August 15, 2006)
Notes and Papers 2005
Note on Bayesian Screening of Possibly Rare Features for Predictive Value (last revised February 16, 2005)Working Note on Bilinear Prediction of Relatedness (last revised March 28, 2005). I have since learned that this describes a (new?) generalization of what is called in simpler cases ``canonical correlation."
Notes on a Bayesian Framework for Association Studies with Multiple Hypotheses (last revised March 31, 2005)
Note on the Significance of 2x2 Contingency Tables and Lindley's Paradox (last revised April 13, 2005)
``What Is Better Than Chi-Square?" and Related Koans (last revised September 23, 2005)
Notes and Papers 2004
How to Use Markov Chain Monte Carlo to Do Difficult Integrals (Including Those for Normalizing Constants) (last revised August 9, 2004)