The cute-as-a-jingle phrase “a chicken is a Dirac limit of a tyrannosaur” emerges in the final section of a study called “Diagonalizing the genome II: toward possible applications,” by Satyan L. Devadoss and Jack Morava.
Here’s the full passage in which that phrase crops up:
“Branching in descent diagrams can be modeled by specialization in the sense of algebraic geometry, defined (for example) by fixing some parameter. In the language of stratified spaces this corresponds to moving from the interior of some region to its boundary: something like a phase change (like water to ice). In such a cartoon description, a chicken is a Dirac limit of a tyrannosaur, in which many of its genetic parameters tend to zero. At this level of vagueness, there is reason to work with codimension than with probability: evolutionary events are highly unlikely, and in reasonable models will have effective probability zero; but in geometry any subspace of positive codimension has measure, and hence probability, zero. The modern theory of phase change in condensed matter physics has developed powerful tools for the study of such transitions (viewed as moving towards a stratum boundary, eg of some phenotype), but in current work there is usually only one such event in focus at a given moment. Evolution forces us to consider long concatenations of such events, and trees are a natural tool for their book-keeping.”
(Thanks to Mason Porter for bringing this to our attention.)
BONUS (distantly related, as most things are): “Pretty Good Gravity, by Jack Morava.