## The Whitefire Crossing

Posted in Books, Mountains, Travel with tags , , , , on January 15, 2012 by xi'an

I grabbed The Whitefire Crossing (by Courtney Schafer) in the Barnes-and-Nobles of Provo, Utah, after one great day of ice-climbing and because of the nice cover! The main plot is about a smuggler+mountain guide taking a hidden mage away from a magicians’ city. The Whitefire is the mountain range the group must cross to reach a safe haven where magic is banned. The first part of the book is quite enticing, taking place in the mountains with several stories of climbs and rescues. There is however a limit on the number of climbs you can describe in a book and the second part of The Whitefire Crossing is more tepid, in my opinion. This is the author’s first book and the way characters interact with one another somehow reflects upon this. The plot is indeed rather predictable and the very final twist not really unexpected. (The [unavoidable] love relation is clear to anyone but the main character from the very beginning of the book!) The cover is also going against mountaineering (obvious) practice that the most experienced climber stands at the back when going down…   The Whitefire Crossing still remains an enjoyable book (I had to rescue over and over from my son’s room as  he kept stealing it from me!) and I am looking forward the sequel, The Tainted City, as obviously are more enthusiastic reviewers, here and there. And there.

## Bridal Veil fall, Provo

Posted in Mountains, pictures, Travel with tags , , , on December 10, 2011 by xi'an

On Friday, Shane Reese (who so superbly organised the MCMSki III conference early this year and helped us so much for the Adap’ skiii workshop!) took me ice-climbing on one of the most iconic ice routes near Provo, Bridal Veil Fall. There, we met with a guide, Scott Adamson, who lead-climbed both pitches we experimented and belayed us as well.

This was a superb day of climbing where we did about six pitches each, including an attempt towards mixed climbing which was very interesting in its closer connection with rock climbing. Scott was immensely encouraging and it was only towards the end of the day that he told us about the new route he had opened on Moose’s Tooth, Denali, a story I had read in a climbing magazine at that time… (He was trying another route there last spring as well.)

## Provo, Utah

Posted in Mountains, pictures, Statistics, Travel, University life with tags , , , , , , , on December 9, 2011 by xi'an

Prior to my attending WSC 2011 in Phoenix next Sunday, I was invited to give a seminar at Brigham Young University in Provo, by the department of Statistics. (This is a private religious university run by the LDS Church, named after of the founders. Students and faculty have to adhere to an “Honor Code” that prohibits, among other things, tea. As an illicit substance. Fortunately, this does not apply to visitors and I can keep drinking tea all night.) The surroundings of Provo are superb,  especially in the current crisp dry weather, the forefront of the Wasatch mountains being the actual Eastern boundary of the town. I hope to get some ice-climbing done today, as Provo is a great spot for doing this!

The visit to the department was very pleasant with a very warm welcome by all and a lot of interesting discussions. I gave my seminar on ABC model choice, using the slides already presented in Madrid last month:

which is quite appropriate given that one of the papers (about the limitations of ABC model choice) was conceived in Utah, early this year (at the MCMC’Ski conference). There were an amazing lot of graduate students in the audience and I hope I managed to get the message out to them, despite the heavy math part at the end. (I personally got a better understand of [A4] and of a way to rewrite it slightly more clearly. I also spotted a typo on mad(y) that should have been corrected weeks ago once Natesh had mentioned it!) Natalie Blades gave a very kind (if rather embarrassing!) intro to my talk and concluded with a “two truths and a lie” game with the audience, asking which one of three facts

• I worked in a Camembert cheese factory
• I played the trumpet in a French Navy band
• I climbed Mont Blanc

was a lie. Producing a very interesting outcome!

## yet more questions about Monte Carlo Statistical Methods

Posted in Books, Statistics, University life with tags , , , , , , , , , , on December 8, 2011 by xi'an

As a coincidence, here is the third email I this week about typos in Monte Carlo Statistical Method, from Peng Yu this time. (Which suits me well in terms of posts as  I am currently travelling to Provo, Utah!)

I’m reading the section on importance sampling. But there are a few cases in your book MCSM2 that are not clear to me.

On page 96: “Theorem 3.12 suggests looking for distributions g for which |h|f/g is almost constant with finite variance.”

What is the precise meaning of “almost constant”? If |h|f/g is almost constant, how come its variance is not finite?

“Almost constant” is not a well-defined property, I am afraid. By this sentence on page 96 we meant using densities g that made |h|f/g as little varying as possible while being manageable. Hence the insistence on the finite variance. Of course, the closer |h|f/g is to a constant function the more likely the variance is to be finite.

“It is important to note that although the finite variance constraint is not necessary for the convergence of (3.8) and of (3.11), importance sampling performs quite poorly when (3.12) ….”

It is not obvious to me why when (3.12) importance sampling performs poorly. I might have overlooked some very simple facts. Would you please remind me why it is the case? From the previous discussion in the same section, it seems that h(x) is missing in (3.12). I think that (3.12) should be (please compare with the first equation in section 3.3.2)

$\int h^2(x) f^2(x) / g(x) \text{d}x = + \infty$

The preference for a finite variance of f/g and against (3.12) is that we would like the importance function g to work well for most integrable functions h. Hence a requirement that the importance weight f/g itself behaves well. It guarantees some robustness across the h‘s and also avoids checking for the finite variance (as in your displayed equation) for all functions h that are square-integrable against g, by virtue of the Cauchy-Schwarz inequality.