Colin LaMont Colin LaMont

Fishing for the right Wright-Fisher

What’s the best way to simulate the competition between individuals in a population? Can we avoid the computational costs of sampling from a big multinomial distribution?

Read More
Colin LaMont Colin LaMont

Classifiers for Likelihoods: Why Logistic?

If you train a binary classifier to minimize the logistic-loss in distinguishing two datasets X and Y, then, as if by magic, your classifier statistic is an approximation for the likelihood ratio of the two distributions p[X]/p[Y]. Where does this magic come from?

Read More
Colin LaMont Colin LaMont

Redesigning the Periodic Table

The periodic table is iconic. A pure scientific breakthrough finding order in chaos. But in some ways it hides the quantum chemistry. Can we redesign the Table to better reflect the quantum mechanical story?

Read More