Thursday, 10 March 2011

Mapping the Subsurface

Last night I went to an excellent talk by Henning Omre on Combining Geological Understanding with Multiple Types of Measurements. I particularly enjoyed the way that he combined statistics, geology, and physics, to yield quantitative outcomes with real-world applications.

After introducing the foundations of probability, through the correspondence between Pascal and Fermat about the equitable division of the stakes of an interrupted game of chance, the Duckworth-Lewis method was presented as a modern application. Then Carl Friedrich Gauss—along with the equation for the Normal curve—on the old 10 Deutsch Mark bill was mentioned. Aside: my favourite mathematician, Leonard Euler, is on the old Swiss 10 Franc note.

As a fan of Edward Tufte, it was nice to see Charles Joseph Minard's graphic of Napoleon's march discussed. Omre referred to the Economist; the article worth a thousand words, which highlights the contributions of Florence Nightingale, makes interesting reading.

Omre mentioned Percy Diaconis and coin-tossing, which resonated with me. He then discussed Bayes' rule and gave some very nice examples its practical application, linked to the use of (stationary) Markov chains.

Finally, he indicated how using a non-stationary Markov chain leads to a problem which can be solved analytically using forward-backward algorithm. Omre concluded by presented some quite convincing 2D and 3D data sets analysed using this technique.

No comments:

Post a Comment