| Older posts | July 2010 -- December 2010 | Newer posts |
|
This page represents only my own views, and not those of any university or other body.
Posted Saturday 18th December 2010 at 9.06pm How to learn French There we go... watch my hit count speed along now! Things I won't miss about Paris: Guy going "Ahhhhhhhhhh" RER B Jussieu canteen The fact that supermarkets don't sell pharmaceutical goods, so pharmacies can charge exorbitant prices Rubbish internet Rubbish heating Things I will miss about Paris: Vélib' Speaking French Friends Fantastic probability group Ping pong table I am sad to be leaving Paris, and there's no way I thought I would be saying that 6 months ago. Not that I didn't enjoy my first 6 months, just that there were enough things that annoyed me about Paris that I would rather have been somewhere else. But I guess it was the same with Bath - it takes a while to get used to any city, to explore its depths... I still find the French attitude odd in some respects, but again probably the English attitude is peculiar in different ways - I just don't notice because they're the ways I know best. For example the heating in my office and those around it has worked for about one week in the last two months, and in that time it has been freezing - I've been sat shivering in two jumpers. That would just not be acceptable in England - a fuss would be made and something would be done very quickly. In France, nothing happens. Two people have resorted to bringing hairdriers into their offices to warm themselves up, and one person even has a raclette-making machine turned on all day sans cheese! Finally, the language. I'm convinced French is not as good a language as English. Sorry if that offends you - I realise I'm not exactly fully qualified to make this point, being a limited French speaker at best. But it's less flexible and has more unneccessary rubbish. However, you'll see that I've put speaking French in the "Things I will miss" list. There are some things one can express better in French than in English - English isn't perfect, just great! - and I've really enjoyed trying to speak it over the last few months. So here's how to learn French: I recommend hanging around with Italians - they tend to speak pretty decent French (bar some pronunciation problems) but speak way, way slower than real French people! This makes conversations far less difficult and conversations are what you need - it's all about practice. To be fair to my office-mate, Clément, he has really tried hard to talk to me and has improved a lot at speaking slowly and taking time out between sentences, which has helped me a lot. Posted Monday 6th December 2010 at 10.06pm I wish I had an interview last week. I don't think I put on a very good performance. My presentation was ok - it's tough to give a very good talk in 25 minutes and I think I (along with the other 4 candidates whose presentations I saw) was solid if unspectacular. But the actual interview caught me a bit cold - I should have done more thinking about my answers in advance. One of my main problems was questions about the future - where do I see myself in 5 years, am I going to be a research star, do I want a research group of my own? The honest answer (and more or less the answer I gave, subject to a lot of waffling) is that I don't know. I still don't really have an accurate picture yet of where I fit in. I don't know how good a mathematician I am, or how good I can be. I know I'm not the most talented mathematician in the world but that I work hard and am very interested in what I do, which can make up for a shortfall in raw ability. Andrew Thomason, one of my directors of studies and supervisors in Cambridge, wrote in one of my first-year supervision reports "quiet but sometimes has good ideas" which I think is the most accurate and concise description of me I've seen. So far it's been enough to just chug away and work on maths to the best of my ability, and let other people decide how good I am. But at some point I have to take charge of my own career. Maybe it's time to be more confident in myself. I wish I hadn't written "maybe" though - there's the rub! Posted Monday 29th November 2010 at 4.17pm Binary search tree pictures I drew some new pictures of the binary search tree (aka quicksort) at the weekend, for a presentation I'm giving at Warwick on Friday.   Here's the description from my research page: "Drawing discrete trees in a useful way is tricky, as the number of possible leaves in each generation increases exponentially. Here are a couple of attempts at drawing a random binary search tree (random quicksort) with 20,000 insertions. Colour represents age (red means the node was added early on, blue late). Click for higher-res." Posted Tuesday 23rd November 2010 at 1.50pm Fighting off a rhino with Graham Gooch and a set of stumps A funny article about what the Ashes means, by Mark Steel. "... for the rest of the summer [in 2005] millions began the day by asking "Did you see the cricket?" It felt like being a scientist who'd spent his life defending a theory, but everyone else thought he was crazy, and now finally he'd proved right, as those who'd derided cricket as a pointless exercise started to follow every moment." Posted Sunday 14th November 2010 at 3.21pm The youth and beauty brigade pt 2 Will Hutton says what I said, and more, more articulately. Posted Monday 8th November 2010 at 9.31pm The youth and beauty brigade I feel like I have paid My debt to society... They said son, Come join the youth and beauty brigade Nothing will stand in our way. I think I wrote a while ago about how being a researcher made me able to see more of a case for tuition fees, as a way of getting more money into UK universities, than when I was an undergraduate. However from what I can see the new proposals, while increasing tuition fees by a large amount, don't actually provide universities with any extra cash. State funding will be cut by as much as - if not more than - the higher fees will raise. There are some improvements like the increased salary threshold for repayments, but they're just window dressing when students will be paying two or three times as much for the same service as they're getting now (and disincentivising early repayment is just cutting off everyone's noses to spite rich kids' faces). They're claiming this is a good deal for students: but would you rather be given £24,000 and have to pay back £9,000 slowly when you're earning £15,000 a year, or be given £24,000 and have to pay back £24,000 slowly when you're earning £21,000 a year? The Browne report was meant to find the fairest way of getting more money into university budgets1, not the fastest way to cut state funding for higher education. 1: Reference: Mandy interview. Ignore the tail-on-donkey stuff from UCU. Posted Tuesday 2nd November 2010 at 12.04pm Unscaled paths Yay! The main chunk of my PhD thesis (unscaled paths of BBM) has been accepted for publication by AIHP. This was written about a year ago but for various reasons (one of which was me worrying more about getting my PhD finished than submitting papers) has been delayed a while. It's by some distance the most original part of my thesis, and I think it contains quite a few non-trivial ideas. The other big part (scaled growth in an inhomogeneous breeding environment) was also hard work, but mainly from a technical point of view - most of the ideas can be found elsewhere, even if putting them together in the right order was a challenge. One (very) small disappointment is the following. At the end of the unscaled paper there is a proof of a theorem about a "very critical" case. Here "very critical" means a certain parameter is 1/3. The reason why 1/3 is critical is slightly complicated, involving various integrals of various functions etc. Completely separately from this, the proof makes use of the fact that 1/3 + 2 * 1/3 = 1. It's not clear at all that this has anything to do with why 1/3 is critical, and yet it turns out to be crucial in the proof. To me this is still rather mysterious and coincidental. So I wrote Our proof ... takes advantage of the convenient - and well-known - fact that 1/3 + 2 * 1/3 = 1. And of course the referee report asks that I take this line out. Spoilsport. Do papers have to be 100% serious at all times? Although thinking about it I guess I should actually try to explain in the article what I've said above rather than just making a rubbish joke. Posted Monday 1st November 2010 at 5.30pm What do we want? 
Posted Monday 18th October 2010 at 9.09pm Il faut agir maintenant So, the French continue to strike hard in protest against a very sensible new law requiring them to work till they're 62, having sat and watched as two stupid laws (banning the Burka and expelling the Roma) were written. And meanwhile, the British get worked up over an (admittedly stupidly constructed) attempt to get rid of some unneccessary child benefit payments, but just frown a bit as the rest of the benefit state collapses under the weight of ridiculous cuts. If this goes belly up, can we give Osborne an ASBO? Posted Saturday 9th October 2010 at 9.34pm Mmmmmmmmm curry I very much agree with this article - even though it does slightly fall into the London-is-the-centre-of-the-world school of thinking. Living in Paris it's not like I can't find food from a wide range of cultures reasonably easily - but it's still the exception rather than the rule, and you have to go out of your way to find it. I think a big part of the difference is that Britain doesn't look down on any other cultures' diets (quite possibly because the "traditional" British cuisine used to be so bad!), unlike for example France, which looks down on - well, certainly British food, and many other diets too. And the UK's giant high-street-killing supermarkets have something to do with it too (but that's another story). And you can't get a decent curry here for love nor money! On the plus side, today I wore shorts and a T-shirt all day, even though it's the 9th of October.  Posted Monday 20th September 2010 at 10.45pm Make way for winter's eerie glow Not nice reading. (And it begs the question: why are job centres so rubbish?) Nice Reading. More nice Reading. My friend Emily played an album called "The ghost that carried us away" by a band called Seabear while I was staying at her house at the weekend. It's good. Seabear are from Iceland, and sound kind of like if Sam Beam fronted The Boy Least Likely To, with Architecture in Helsinki as their backing band. Posted Friday 3rd September 2010 at 11.25pm Gold Soundz? Even if I don't love them, I can see that Paranoid Android and Only Shallow are well-constructed songs that had a big effect on rock music. Loser, too, was really something new back then. I don't agree with the choice of Say It Ain't So, but only because I think My Name Is Jonas or Surf Wax America or the Sweater Song are even better. The State That I'm In is just a fantastic song, and as with REM it's no criticism to say Belle and Sebastian's best stuff was done right at the start of their career. I never really liked Smells Like Teen Spirit, but that's probably partly because I only heard it after grunge had already been ripped off by a million other inferior bands. And finally, I love - LOVE - Holland, 1945 and Common People. Two of the best songs ever. But Gold Soundz? No, really. Gold Soundz? It's a nice song. Cute melody, half-mysterious, half-pretty lyrics. It sounds off the cuff, effortless. I appreciate that I was only 10 at the time, and that Pavement don't really mean anything to me like they do to people five or ten years older than me. Maybe I just don't get it. But is it really the best track of the 90s? Listen to what Jarvis has to say - acerbic, relevant, with arrogance and wit and bombast and one hell of a tune. Listen to Jeff Mangum on Holland, 1945 (on the whole of Aeroplane in fact) - the surreal lyrics, those churning guitars and blaring horns and the singing saw, and the drums - the drums! - and the feeling, the soul in that music. Heck, even listen to Paranoid Android - so precise, so carefully thought out, so brilliantly put together. I could have accepted Paranoid Android. But Gold Soundz? Posted Wednesday 1st September 2010 at 10.39pm Some thoughts and a recipe On Pakistani cricket: only Pakistan could have Shahid Afridi playing the straight man. But I think the calls for life bans for anyone involved in the match-fixing stuff are a step too far. The idea behind that, presumably, is that we should make an example of them and then no-one will commit the crime again. But that's the cricket equivalent of advocating the death penalty. We should instead realise that these kids have been misled by people with a lot more money and experience than them - they were greedy, which is of course a bad thing, but personally I'm not sure they know any better. As far as I know all they've done (if proven guilty) is deliberately bowl no-balls occasionally in test matches, which very rarely affects anything. It's wrong, but in the scheme of match-fixing it's like nicking three fizzy cola bottles and a giant strawb from Woolies pick'n'mix. On the Labour leadership race: I know nigh-on nothing. I would more naturally side with the redder Ed, but I read an article by Ed and an interview with David, and I was more impressed with David's. My dad's school's new building opened yesterday. You can see him, and it, here (buzzwords abound): On the BBC (I think this works worldwide) On ITV (UK only). I have made a couple of gardener's pies (like shepherd's pies but veggie) recently, one with chick peas and the other with kidney beans. Here are the instructions I sent to a friend: I boiled potatoes and carrots in one pan, and meanwhile fried onion and leek with garlic, then added some rosemary, sage, and a stock cube, then the carrots when they were ready. Cook for a while, then add a little bit of tomato and a tin of kidney beans. Once it's got to the right consistency add spinach and stir till the spinach has wilted. Put in a dish, mash the potatoes with butter and spread on top. Grate some cheese on top of that (preferably cheddar but seeing as we're in France that can be difficult). Cook in the oven until the cheese is slightly browned (30 mins?). Before...  And after.  Posted Friday 20th August 2010 at 9.26pm A* for effort First of all, congratulations to Stas Smirnov and all the other winners at the ICM in Hyderabad yesterday. Smirnov gave one of the courses at the summer school I was at in Brazil a few weeks ago. I've seen him talk a couple of times before, and whether because of my (at the time) almost complete ignorance of the subject or otherwise, I barely understood a word. In Brazil he was brilliant, explaining things clearly despite the difficulty level of the material, and he also seemed a lot happier, regularly joking with the audience and attempting to run and hide whenever there was a power cut. On a rather lower level, congratulations also to the many people picking up A-level results this week. This year has seen the launch of the A* grade, which seems to me a bit of a half-hearted solution to a difficult problem. It has also led to a re-realisation that kids at private schools get much better grades on average than kids at comprehensive schools: this article by Mary Dejevsky is a good example of someone jumping on that particular horse. But I think she largely misses the point. Firstly, the A* results may make this issue more transparent to her, but the universities are already well aware of it and generally attempt to take it into account. Secondly, it is entirely reasonable that the private schools in the UK should gain better results on average than the state schools. They have bigger budgets and their intake is overwhelmingly upper class. I am willing to conjecture, too, that the second of those factors is by far the more important of the two. Money is clearly extremely important, as it pays for better facilities and (on average) cleverer teachers. But probably if you compare the results of a high-performing state school in a leafy suburb somewhere with the results of a private school in a similar area the difference will be far, far less striking than the stats spouted repeatedly in the newspapers. There are good comps and there are bad comps. My rather narrow and biased view is that things are improving slowly as far as state schools are concerned, at least compared to 10 or 15 years ago. But please don't just judge our schools on how many A*s they get. EDIT: I meant they are improving compared to 10 or 15 years ago, but slowly - not that the improvements had slowed! Posted Tuesday 17th August 2010 at 11.13pm Another experiment with Processing I wanted to simulate some property of the binary search again today to try to decide whether it's worth following up a possible idea for a proof. I couldn't find my CCATSL code so I rewrote the thing in Processing. It turns out the the idea could be worth a shot, and I couldn't resist drawing some pictures to go with it. They're nowhere near as useful as my old CCATSL ones for actually visualising the process, but they do look nicer. There are no axes unfortunately - as I say, if you actually want to know anything about the process, the old pictures (also on my research page) are much better. Here the number of leaves in the fringe - the top level of the tree - is drawn in orange/red, with those on the second layer down in blue/purple, and the third level in green.   Posted Friday 13th August 2010 at 9.28pm "Intelligent folders" in Thunderbird 3.1 I updated Thunderbird today to version 3.1 - and what used to be called "intelligent folders" (I think) had disappeared. Turns out there's a quick fix though - above your mailboxes, where it says "All folders", click the right arrow and it changes to "unified folders", which seems to be exactly the same as intelligent folders used to be. Poor form from Mozilla on that. To be honest I don't think Thunderbird is great overall, but it's better than any of the alternatives I've seen for Windows now Eudora no longer exists. I'm still waiting for someone to make a nice, simple email client that just works nicely. Posted Monday 9th August 2010 at 11.28pm Dynamical percolation I was experimenting with a language called "Processing" today, which was recommended to me by Shaun Maguire and makes doing stuff visually nice and easy. It has some downsides but is certainly worth a spin if your programming skillz are as bad as mine. Anyway, here are the results:  Click here to see the program in action (and get the source code if you want it) - it requires Java. Also click here for a smaller, faster-moving version. We have Jeff Steif and collaborators to thank for the model, which is called "dynamical percolation" - although the program is a discrete-time approximation (the smaller version is a worse approximation but is a bit more lively than the bigger one). I just saw a course involving this model in Brazil and liked it, so it made a good example to play with as I got started with the language. I'll write something about it on my research page at some point, but for those who know what the following means, it's just critical bond percolation on Z^2 where each bond randomly updates its status at independent exponential times. The clusters belonging to the origin and four other evenly-spaced points (one in each corner) are shown in different colours. The important thing to note from a mathematical point of view is that "large" clusters (of each colour) appear and disappear very suddenly. Posted Friday 6th August 2010 at 4.17pm Black cab I feel like going home, but at the same time I don't. Posted Saturday 17th July 2010 at 1.36pm Who is this Martin Gale, and what's he done to his spine? It's surprised me this week how many of the people I've spoken to have never attended a proper course on martingales and Ito's formula. Hugo, the guy doing the TA sessions for the SLE course, asked who knew Ito's formula - and maybe slightly over half the audience put their hands up. I almost didn't because I thought he must be talking about something else - surely everyone knew Ito's formula? For me, doing probability without having martingales in your back pocket is like hiking in bare feet - you can do it, but it'd be so much easier and more enjoyable (and you'd probably get much further) if you had some boots. I've given a talk several times called "The unscaled paths of branching Brownian motion" in which, after stating the main theorem, I say "So, how are we going to prove this? We want to know about N, and I'm a probabilist, so when I want to know about something, I look for martingales associated with that something." People sometimes laugh, but I'm serious! Someone once said that you could get a chair in pure maths at Cambridge just by saying "Cauchy-Schwarz!" at the right times. Perhaps it's not quite so succinct, but my tactic for doing probability pretty much just involves finding martingales in the right places. I remember talking to James Norris, while I was a Part III student, about doing a PhD. It was the first term so I was going to the Advanced Probability course at the time (lectured by Gregory Miermont), and James asked if there was any part of the course that I particularly liked. I was kind of young and a bit nervous, so I think I just mumbled something about liking it all. But actually there was one lecture that I really loved - it was when Gregory proved the recurrence and transience properties of Brownian motion, just by looking at the right martingales. I think that was the moment when I decided that maybe I should think about doing probability. I didn't like probability in my first year of undergraduate, and only did the Probability and Measure course in the third year because Julia Wolf (my Analysis II supervisor) promised me it was all analysis and no probability. But once I knew measure theory, probability became analysis and it turned out to be the bit of analysis I was best at, except perhaps Fourier analysis. The Cauchy-Riemann equations in complex analysis and the linearity in linear analysis introduced too much magic for me - they destroyed my intuition. Probability, on the other hand, made sense - and martingales suddenly made it exciting too. Posted Thursday 15th July 2010 at 1.56pm "I don't want to see the quality of universities cut, we don't want to narrow the opportunities for young people to go to university, so therefore the only possible way forward is by having a bigger graduate contribution." The words of Vince Cable. He's summed up in one sentence how easy things used to be for the Lib Dems, and how difficult they are now. On the one hand they ran around saying things like "we'll abolish tuition fees!" and never had to back up their claims. On the other hand, we should remember that actually things could have been different if the Lib Dems were in power by themselves - perhaps they would have held their nerve and not renewed the UK's nuclear deterrent, something that the Tories are committed to and which costs a lot of money. I was against tuition fees as a student - most people were - but having seen things from the other side a bit more, Cable's quote is pretty accurate. A graduate tax (one of the options Cable proposes) would be an interesting way of doing things in principle (one can perhaps imagine headstrong youngsters deciding not to go to university because they are convinced they are going to make millions and don't want to be taxed more), but would probably end up being fairly similar to the current system in reality, albeit with people paying more than they do now. Posted Wednesday 14th July 2010 at 9.15pm Burkas and Brazil Ignore the first two paragraphs and Andy Hamilton's article on banning burkas in the Independent is a good one. The more pleasant news is that I'm in Brazil, at the Clay Maths Institute summer school. There's an obscenely packed schedule (talks from 9am till 8pm) but lunchtimes give a bit of time to explore the nearby beaches.  Posted Tuesday 6th July 2010 at 1.04pm Some thoughts on giving talks I'm sat on a train and I was just thinking about giving talks. There is, of course, lots of advice available on how to give a talk, and I'm by no means an expert (although I have improved since my Pt III essay talk, where at one point I was waving at a blank blackboard in an attempt to explain something I had just rubbed out). But here are a couple of points - neither of which is new or profound - that I find it helpful to concentrate on. 1. Know what you are trying to do with your talk. Normally I'm talking to probabilists (fairly pure mathematicians). I generally hope that by the end of my talk, they will a) know what I've proved, b) think they understand the proof, and c) perhaps have learnt something. The "think they understand the proof" is the difficult part. I remember Tom Körner telling a story about an engineer friend of his who said, after watching the BBC Panorama documentary, that he thought he understood Wiles' proof of Fermat's Last Theorem - except for the details. Körner's point was that you don't understand a proof until you can do it fully yourself: the devil is in the details. And he's right, but when I'm giving a talk there isn't usually enough time or interest for the details. I actually want my audience to feel like the engineer - and perhaps, since they're mathematicians and many of them are cleverer than me, they can actually fill in the details themselves anyway. 2. Know how long your talk will take. If your talk is only supposed to be, say, 45 minutes long, then make sure you know that you can fit it into 45 minutes. It's no use thinking it'll be alright on the night - 45 minutes is 45 minutes is 45 minutes. I think of this a bit like understanding magic tricks. Often the reason people can't figure out how magic tricks are done is because they have some portion of their brain that wants to believe it's magic, and so suspends its usual logical rules (the Derren Brown episode where he "predicted" the winners of several horse races in succession is a good example of something that's obvious as long as you make the basic assumption that he doesn't actually have magical powers). Mathematicians give overlong talks because they want to believe they can fit in everything they want to say. Fit your talk into the time available, don't fit the time available around what you want to say. This one is, of course, pretty much the oldest piece of advice in the book, but it can't be overstated. It's not simplifying by much to say that if you overrun then almost everyone in the audience will be extremely unimpressed with both you and your talk!
|