Colloquium Schedule         Guidelines         Directions        

Colloquium
Thursday, December 4, 2014


The fundamental problem of forensic statistics

Richard Gill (Leiden)


Abstract: The Netherlands Forensic Institute has a data-base of Y-chromosome DNA profiles from a random sample of 2085 Dutch males. Within this sample, some men have the same profile as others, but most have different profiles. The sample generates the following (number theoretic) partition of the sample size: 2085 = 13 x 1 + 7 x 1+ 6 x 2 + 5 x 5 + 4 x 7 + 3 x 30 + 2 x 130 + 1 x 1650 = 13 + 7 + 6 + 6 + 5 (repeated 5 times) + .... + 2 (repeated 130 times) + 1 (repeated 1650 times).

From the body of an otherwise unidentified MH17 victim a Y-chromosome profile has been recovered. There is a missing person who we know has the same Y-chromosome profile. That profile is not present in our sample. So it is a rare profile and the match we have observed presumably constitutes pretty strong evidence that the missing person and the unidentified MH17 victim are the same. Can we quantify this?

For a preview of my talk see

http://www.slideshare.net/gill1109/a-walk-in-the-black-forest-during-which-i-explain-the-fundamental-problem-of-forensic-statistics

and for some mathematical statistical theorems

http://arxiv.org/abs/1312.1200

Hardy's theorem on the asymptotic number of partitions of an integer will turn out to play a crucial role in proving consistency of the "pattern maximum likelihood estimator" of the probability distribution of profiles.

<<< previous talk
next talk >>>



Location :  Room 611 of the Hans Freudenthal building, formerly known as the Wiskunde/Maths building, (campus De Uithof) Budapestlaan 6, Utrecht.
Date and time : Thursday, December 4, 2014 15:30-16:30. The lecture is preceded by coffee, tea, and cookies from 15.00 to 15.30 hour in the same room.