Archive for the Running Category

likelihood-free inference by ratio estimation

Posted in Books, Mountains, pictures, Running, Statistics, Travel, University life with tags , , , , , , , , , , , , , , , on September 9, 2019 by xi'an

“This approach for posterior estimation with generative models mirrors the approach of Gutmann and Hyvärinen (2012) for the estimation of unnormalised models. The main difference is that here we classify between two simulated data sets while Gutmann and Hyvärinen (2012) classified between the observed data and simulated reference data.”

A 2018 arXiv posting by Owen Thomas et al. (including my colleague at Warwick, Rito Dutta, CoI warning!) about estimating the likelihood (and the posterior) when it is intractable. Likelihood-free but not ABC, since the ratio likelihood to marginal is estimated in a non- or semi-parametric (and biased) way. Following Geyer’s 1994 fabulous estimate of an unknown normalising constant via logistic regression, the current paper which I read in preparation for my discussion in the ABC optimal design in Salzburg uses probabilistic classification and an exponential family representation of the ratio. Opposing data from the density and data from the marginal, assuming both can be readily produced. The logistic regression minimizing the asymptotic classification error is the logistic transform of the log-ratio. For a finite (double) sample, this minimization thus leads to an empirical version of the ratio. Or to a smooth version if the log-ratio is represented as a convex combination of summary statistics, turning the approximation into an exponential family,  which is a clever way to buckle the buckle towards ABC notions. And synthetic likelihood. Although with a difference in estimating the exponential family parameters β(θ) by minimizing the classification error, parameters that are indeed conditional on the parameter θ. Actually the paper introduces a further penalisation or regularisation term on those parameters β(θ), which could have been processed by Bayesian Lasso instead. This step is essentially dirving the selection of the summaries, except that it is for each value of the parameter θ, at the expense of a X-validation step. This is quite an original approach, as far as I can tell, but I wonder at the link with more standard density estimation methods, in particular in terms of the precision of the resulting estimate (and the speed of convergence with the sample size, if convergence there is).

Japan’s Kumano Kodo pilgrimage [book review]

Posted in Books, Mountains, Running, Travel with tags , , , , , , , , , , , , on September 8, 2019 by xi'an

When preparing our hiking trip to the Kumano Kodo pilgrimage route, I was extremely pleased to find a dedicated guidebook that covered precisely the region we wanted to explore and provided enough background material to make the walk sound feasible. However, once I found the Kumano Travel reservation website, run most efficiently by the Tanabe City Kumano Tourism Bureau, the information contained in this site made the guidebook less relevant. And when we arrived in Tanabe at the start of the trail, I found that the Bureau was also distributing free leaflets in English for each of the three main routes, which described day-by-day the stages of the hikes, as well as recommendations and tips. Making in the end or a posteriori the guidebook superfluous. (As the detailed description of the routes was not necessary, given how clearly they are identified. The leaflet managed to stand the five days on the trail despite rain, humidity, frequent consultations and a general lack of care, as shown above!)  Hence, while there is nothing wrong with the guidebook which also includes an extra day-hike along the Eastern coast of the Kii peninsula and another one from Koyasan to the bottom of the cablecar [again covered by leaflets at the local tourism bureau], I would not strongly recommend it. Interestingly (?), when I stated these mere facts as a review on Amazon, I was rejected as contravening their review guidelines without further precision… (I can only post comments on the French portal of Amazon as my associate gains mean that I never “buy” anything on the US portal!)

 

über Salzburg [jatp]

Posted in Mountains, pictures, Running, Travel with tags , , , , , , , , on September 5, 2019 by xi'an

off to SimStat2019, Salzburg

Posted in Mountains, Running, Statistics, University life with tags , , , , , , , , , , , , , on September 2, 2019 by xi'an

Today, I am off to Salzburg for the SimStat 2019 workshop, or more formally the 10th International Workshop on Simulation and Statistics, where I give a talk on ABC. The program of the workshop is quite diverse and rich and so I do not think I will have time to take advantage of the Hohe Tauern or the Berchtesgaden Alps to go climbing. Especially since I am also discussing papers in an ABC session.

temples on Mount Koya

Posted in Mountains, pictures, Running, Travel with tags , , , , , , , on September 1, 2019 by xi'an




trailers versus mountaineers?

Posted in Kids, Mountains, Running, Travel with tags , , , , , , , , , , , on August 31, 2019 by xi'an

A slight altercation in a swimming corridor during lunch put me back into this Le Monde paper I read yesterday about (real?!) mountaineers being annoyed at trailers, especially those currently running the Ultra-Trail du Mont Blanc (UTMB). A lady stopped me from going further for not crawling as this was a “crawl only” lane and started a lengthy tirade that I cut short by moving to another lane. I find such debates pretty absurd and rather hypocritical. When the fundamental goal is mostly to reduce the number of people on the trails, in the mountains, or in the pool by creating categories with those in and those out. This seems an unavoidable human trend that happened several times in mountaineering, from the early days when going above a certain limited was prohibited to those when climbing solo, rope-free, mixed style, without a registered guide or certificate, &tc. base-jumping, was or became taboo. It is annoying to see crowds in the mountains, whether on the Everest final sketch or on the UTMB track, for sure, but by nature these are singular events and the next peak is almost surely free. It is also annoying to find other climbers on one’s chosen route as they will certainly cause delays, but this is the nature of the game and the next route may well be free. I thus find pretty annoying that some claim their rights to enjoy mountains are higher or purer than others, whom they accuse of elitism and ill-placed competition, when themselves are far from free of the same defect.

 

hiking the Kumano Kodo Nakahechi imperial route

Posted in Mountains, pictures, Running, Travel with tags , , , , , , , , , , , , , , , , , on August 19, 2019 by xi'an

The Kumano Kodo is a network of paths of pilgrimage towards places seen as sacred by either buddhists or shintoists (or syncretists!) from the 700’s. (The Kumano Hongu Taisha shrine celebrated its 2050th anniversary last year!) Meaning for non-believers a well-established system of ancient hiking paths in the mountainous forests of the Kii peninsula, south of Osaka, Kyoto and Nara. Apart from the potential dangers of heavy rain, like massive mudslides, there is no particular difficulty in the hikes (which were done wearing braided straw shoes, the waraji sandals) and a dense network of guest-houses and bus routes makes it possible to adapt the length of the day trips to one’s speed. While this is the place of Japan with the maximum of rain, the hot temperatures actually make it more than bearable when it is a shower and not a typhoon!day one

14km, from Takijiri-oji to Chikatsuyu-Oui, 7:30 hours, 1100 positive gain, nice trail all the way, maximum temperature 32°, met two dozen hikers and two trail runners, got tired by the end, confused one minshuku guest house for another, the hosts of the first one drove us to the second and later brought me back my sandals I have left in their car, no English speaking at the second place and no Onsen but very nice bento box and a pleasant conversation with a couple of Ausso-Danish young women spending a month in Japan and four on the trail. Entire path under cover, making high heat so much more bearable! Not too much in terms of views despite following ridges and visiting tops. Plenty of huge butterflies.

day two

6km, from Doguyaba-bashi (reached by bus) back to Tsugizakura, 2:30-3:00 hours, maximum temperature 30⁰, mixed trail and road, rain spells and showers, met three other couples of hikers, arrived too early at the guest house, waited in a one-room thatched traditional tea house by the side of the road with free hōji-cha tea offered by a very nice old man (not English speaking, alas, which limited the exchanges), reached the guest house at the same time as a Canadian-Spanish couple embarked on a massive six month travel. The dinner was imperial with a whooping ten dishes, all delicate and beautiful. Plus getting our names in Japanese Kanji characters.

day three

9km, from Doguyaba-bashi (the host gave us a ride) to Hosshinmon-oji, 4 hours, 600 positive gain, maximum temperature 29⁰, heavier rain spells, first part really nice then switching to forest track due to closures, finishing detour on tar road, took the infrequent bus to Yunomine-Onsen due to the closure. A few more couples and lone hikers passed us. Stayed at a larger guest-house minshuku with its own natural onsen plus normally access to the river with swimming waterholes. Far from Hongu and Yunomine, alas. But also welcome place to wait two days for the typhoon Krosa to pass by Western Japan.


day four

14km, from Nachisan to Koguchi, 930 positive gain and 1260 negative gain!, alas cancelled as advised by local tourist office due to uncertainties on the state of the trail, took the bus instead and enjoyed a swim after a very hot day in the now quiet Kumanogawachonichi, next to a most congenial guesthouse. Very quiet place, superlative food!


day five

13km, from Koguchi to Ukegawa,  690 positive gain, 5:30 hours, maximal temperature 29⁰, sunny, very enjoyable part of the trail with many ridge walks, no road share, hardly any trace of the typhoon, and early arrival in Hongu. Met with a guest from the Yunomine-Onsen guest house hiking the other way. And very few others.