Archive for University of Warwick

One World ABC seminar [season 2]

Posted in Books, Statistics, University life with tags , , , , , , on March 23, 2021 by xi'an

The One World ABC seminar will resume its talks on ABC methods with a talk on Thursday, 25 March, 12:30CET, by Mijung Park, from the Max Planck Institute for Intelligent Systems, on the exciting topic of producing differential privacy by ABC. (Talks will take place on a monthly basis.)

professor position at Warwick

Posted in Statistics with tags , , , , , , on January 21, 2021 by xi'an

corporate smokescreen

Posted in University life with tags , , , , , on January 8, 2021 by xi'an

Warwick U had all its staff going through an endless list of compulsory (under threat of being cut from the local system) “training” modules that droned through fairly elementary notions of computer safety and data protection for hours. With a compulsory quiz associated with each module that cannot be taken without “listening” to the entire thing. Given the absence of any real training provided by this enormous waste of staff time (since I presume the modules can be completed during working hours!), I wonder at the purpose of this corporate exercise, beside checking all the boxes to achieve legal protection in case of a data breach or another disaster… (For certain, nothing meaningful can be inferred from the policy pages attached to the exercise, as they are written in newspeak with enough acronyms to incapacitate a brand new typewriter!)

Don Fraser (1925-2020)

Posted in Books, Statistics, University life with tags , , , , , , , , , , , , , , , on December 24, 2020 by xi'an

I just received the very sad news that Don Fraser, emeritus professor of statistics at the University of Toronto, passed away this Monday, 21 December 2020. He was a giant of the field, with a unique ability for abstract modelling and he certainly pushed fiducial statistics much further than Fisher ever did. He also developed a theory of structural  inference that came close to objective Bayesian statistics, although he remained quite critical of the Bayesian approach (always in a most gentle manner, as he was a very nice man!). And most significantly contributed to high order asymptotics, to the critical analysis of ancilarity and sufficiency principles, and more beyond. (Statistical Science published a conversation with Don, in 2004, providing more personal views on his career till then.) I met with Don and Nancy rather regularly over the years, as they often attended and talked at (objective) Bayesian meetings, from the 1999 edition in Granada, to the last one in Warwick in 2019. I also remember a most enjoyable barbecue together, along with Ivar Ekeland and his family, during JSM 2018, on Jericho Park Beach, with a magnificent sunset over the Burrard Inlet. Farewell, Don!

Simulating a coin with irrational bias using rational arithmetic

Posted in Books, Statistics with tags , , , , , , on December 17, 2020 by xi'an

An arXived paper by Luis Mendo adresses the issue of simulating coins with irrational probabilities from a fair coin, somehow connected with one of the latest riddles. Where I realised only irrational coins could be simulated in a fixed and finite number of throws! The setting of the paper is however similar to the one of a Bernoulli factory in that an unlimited number of coins can be generated (but it relies on a  fair coin). And the starting point is a series representation of the irrational ζ as a sum of positive and rational terms. As well as an earlier paper by the Warwick team of Łatuszyński et al. (2011). The solution is somewhat anticlimactic in that the successive draws of the fair coin lead to a sequence of intervals with length divided by 2 at each step. And stopping when a certain condition is met, requiring some knowledge on the tail error of the series. The paper shows further that the number of inputs used by its algorithm has an exponential tail. The examples provided therein are Euler’s constant

\gamma =\frac{1}{2} + \sum_{i=1}^\infty \frac{B(i)}{2i(2i+1)(2i+2)}

where B(j) is the number of binary digits of j, and π/4 which can be written as an alternating series. An idle question that came to me while reading this paper is the influence of the series chosen to represent the irrational ζ as it seems that a faster decrease in the series should lead to fewer terms being used. However, the number of iterations is a geometric random variable with parameter 1/2, therefore the choice of the series curiously does not matter.