Proof: obvious

Paper accepted

Tue, 21 Feb 2006 13:40:45 +0100

The paper I wrote about last time got accepted. So that's great. Unfortunately, the paper I had been working on for the past couple of weeks had to be postponed because the algorithm changed the day before the deadline. Luckily, it's now a stronger result, though the proof isn't as cool as it was. Win some, lose some I guess.

Updates

Mon, 05 Dec 2005 16:13:06 +0100

Since ipe was released containing one of my ipelets, I got a bug report. Otfried told me how to fix it, so I have updated my ipelets with the bug fixed. (The bug was that if you applied the ipelet and then moved the object and then applied the ipelet again, the ipelet would work on the original version of the object.) So if you follow these sorts of things, both of the ipelets have new versions.

The other thing that has been going on lately is that I have submitted a new paper. My co-authors, Mark and Boris , really did a great job in getting the paper into decent shape. It still amazes me how much a paper can change from when I think I'm done with it to the final product. So anyway, we submitted it to SoCG . It's the only paper I'm submitting there, so I really hope that it gets in.

New version of ipe

Fri, 02 Dec 2005 16:33:59 +0100

Otfried has released the new version of ipe . As I had hoped, it uses QT4, and it includes one of my ipelets. Go check it out.

Ipe in QT4?

Tue, 06 Sep 2005 14:02:29 +0200

I don't know how hard the port would be, but I think it would be nice to get ipe into qt4 as soon as possible, mainly so that people on Windows can start using ipelets not compiled by Otfried. They could in the past, of course, but the license fees for the qt development libraries on Windows before 4 are in the thousands of Euros range per developer. Starting with 4, the development libraries seem to be free.

Ipelet update

Tue, 06 Sep 2005 13:57:03 +0200

I fixed a fairly subtle bug in the ipelet that appeared when the input polygon spiraled heavily. As usual, get it on the ipelets page.

Ipelet update

Fri, 12 Aug 2005 13:43:49 +0200

Now with memory management.

New ipelet

Thu, 11 Aug 2005 15:02:35 +0200

Just added a new ipelet to my ipelets page. This one computes the points visible to a point inside a given polygon assuming that the edges of the polygon are walls that can't be seen through. The code for this ipelet is still a bit rough, so don't use it without saving your file first.

Submitted

Thu, 07 Jul 2005 14:19:35 +0200

I submitted the paper I was working on with Mark to SODA yesterday. It's a good feeling to be done, and we weren't even all that frantic at the last minute. The paper has some good results in it, so it might even get accepted, though I hear that acceptance at a big conference like SODA can be something of a crapshoot.

Work is fun

Tue, 24 May 2005 17:47:45 +0200

For almost the entire year until now, I had been writing a paper about an algorithm that Mark and I came up with. The algorithm was pretty complicated, so a few weeks ago I decided to step back and see if I could make it simpler. The result was good -- we shaved off a log n factor from the time and space complexity. However, the paper that I had been working on for months is now obsolete, never to be published. I could salvage a large part of it for the new paper that I am now working on, but it was still a bit discouraging.

Event based web app

Tue, 24 May 2005 17:24:12 +0200

I recently had an idea for something similar to evdb. It is a web application that lets you search based on a location, a name of something like a sports team or a band and it tells you what events are coming up that are related to these things.

The current implementation clearly needs a bit of work (there is a Blue Jays game listed as being in "Toronto, United States") and more data (there is only one page about things related to the Netherlands) but it has definite potential. What really needs to happen is for a service like this to gain critical mass. That way, promoters of events have a simple way to publicize the events and event-goers can be more confident that the events that they are interested in will be listed.

Of course, there is a slight computational geometry angle that could be explored here: I would really like a way to find events that happen within 100 km of my town. This would correspond to a range-searching query with a circle as the range. This can be accomplished using O(n^2) space and O(k) query time. If that is too much space, there is apparently a data structure that uses O(n log n) space and has O(sqrt(n log n) + k) query time. This is a pretty bad result, and one would guess that a more practical solution would be to use a rectangular query area where there are data structures that use O(n log^2 n) space and O(log^2 n + k) query time.

First ipelet

Mon, 21 Feb 2005 12:04:32 +0100

I wrote an extension for the image editor ipe that computes the smallest circle enclosing a set of points. Aside from the usual C++ headaches, it was pretty fun -- it is always nice to see visual proof that a program that you write works. I will probably write more when I see the need for them, so I have created an ipelets page that I will update when necessary.

Why Climacs is a bad idea

Tue, 25 Jan 2005 16:01:10 +0100

Brian Mastenbrook blogged about five lisp projects to watch and included climacs as number 2. It has a great name, but other than that, I don't see the value of climacs. First of all, it is not a new project, just a new name on an old one (called Hemlock). Secondly, getting rid of elisp is a horrible idea. You lose the huge amount of code that has been written in elisp; meaning that climacs will never replace emacs. Also, elisp is a language that is specialized to writing editors, while CL is not (though of course it could be turned into one, as Paul Graham always points out).

The solution I favor over climacs is rewriting the core of emacs in CL. I haven't done it, and the climacs people seem to be most of the way there, so I shouldn't talk. But it will happen eventually, and emacs will be much more fun to hack.

Emacs hacking status

Fri, 17 Dec 2004 15:41:33 +0100

I get about a question per month asking about the status of my emacs anti-aliasing work. Unfortunately, progress has pretty much stalled because I have hit a frustrating snag. I have taken my arch repository offline because everything in it segfaults without even displaying a window. I do plan to work on the problem a bit during my upcoming holiday, and I will try to keep updating this site with the status.

The CBC really gets it

Fri, 17 Dec 2004 14:39:41 +0100

Two days ago, I wrote about the CBC broadcasting in Ogg Vorbis. Today, I found the CBC's RSS page linked from the front page. What's more, if you scroll to the bottom, you'll notice that they recommend RSS readers for more platforms than Windows. It looks like they are taking my tax dollars and using them to hire people with a clue. Shocking!

Making a domain-specific language

Thu, 16 Dec 2004 11:31:29 +0100

Bill Clementson has an extremely impressive graph about the productivity improvements that arise from a domain-specific language. His conclusion is that one should use Common Lisp to create a domain-specific language and then use that language to make the application. I have never really done this, but it really makes sense as a way to program. I have a feeling that one really needs to be a wizard to accomplish this though.

Just to get a feeling of how to do this, I am going to do 2 things: first, I am going to reread On Lisp all the way through. Then I am going to try to write a language for writing compilers, with the aim of writing a compiler for elisp. This would be a start for completing my rather grandiose idea of rewriting the core of emacs in Common Lisp.

New title

Tue, 14 Dec 2004 16:45:48 +0100

I struggled with what the new name of this site should be, given that it couldn't just be <my name>'s blog any more. Yesterday, I thought "The outdoor type" would be a good idea, because it's kind of ironic as well as being the name of a good song. But then I went out on the web and found another blog with that name.

So now I have come up with a new name. "Proof: obvious" seems like a good title because it is one of my bad habits to use that construction in early drafts of papers and math assignments before I realize that the proof is not obvious and what I am trying to prove is in fact wrong. So having that as the title of this site might remind me about this bad habit and discourage me from doing it. Also, I'm not going to go look for another blog with the same title so that I don't have to change it again.

CBC in Ogg Vorbis

Mon, 13 Dec 2004 11:11:55 +0100

The BBC got a lot of press about a year ago for moving to Ogg Vorbis, but Canada's version, the CBC, has already done it . The Beeb still seems to be using RealPlayer. Real seems to be better, more open, and Linux-friendly company than they were in years past, but I still have trouble trusting them. So way to go, CBC. I'll be listening in.

The title

Sun, 12 Dec 2004 13:47:34 +0100

Well I thought the title of my blog would be unique (given that it is my name) but I guess it isn't. There is another Chris Gray out there. He seems to work for Microsoft (ugh) on Windows Server (ugh again). So I am going to have to come up with a new name for my blog since the other Chris was blogging first and I'd rather not be confused with him. No offense, Chris, if you're reading this -- I just think confusion would be a bad thing.

Playlist generation

Mon, 06 Dec 2004 18:45:22 +0100

A couple of weeks ago, I went to a workshop on intelligent algorithms. One of the talks was on generating playlists for little mp3 players. The limiting factor in these players is the amount of storage, and finding a set of mp3s that fill them is equivalent to the well-known knapsack problem. So I went home and whipped up a little evolutionary algorithm (hopefully that is the right term -- I don't look at much AI stuff) to fill up my mp3 player. It is working very well so far. It is not quite 100 lines of python, and fills up my mp3 player to within half a megabyte of being full after very little searching. That is an error of roughly a quarter percent. Pretty impressive.

You won't get arrested, but..

Mon, 06 Dec 2004 17:54:22 +0100

In a recent post Bruce Schneier makes the point that you won't get arrested at airport security if you are carrying something "forbidden" and you play dumb. That's true, but it misses the point that they will still take your stuff away from you and not give you a way to get it back. I had this happen to me in Montreal. They took away my Leatherman Micra. When I got back, I went to the security desk and asked for it back. All I got was quizzical looks. So only take stuff that's forbidden if you don't care about losing it.

Permalinks improvement

Mon, 06 Dec 2004 16:20:42 +0100

Made the permalinks point to an archive page for the day rather than the main blog page for obvious scalability reasons. I might have manually adjust the rest of the entries to do the same.

Approximation algorithm

Mon, 06 Dec 2004 12:26:42 +0100

See my previous work before reading this post. I have found a cute little approximation algorithm for the edge-restricted version of the pessimistic shortest path on uncertain terrains problem. (See? I told you to study before reading this.)

The algorithm is as follows: Make a graph where every edge corresponds to an edge of the given uncertain terrain. For each edge, give the edge a weight that is the longest the edge could possibly have in the uncertain terrain. Then find the shortest path on this graph. That's it. It is not too hard to see that this path is at most twice the length of the optimal path measured on the uncertain terrain, but that is a proof that is left as an exercise for the reader.

NS Lost and Found Website... Found

Mon, 06 Dec 2004 11:16:04 +0100

The last two posts have been about my panicking over the bag that I lost on the train. The Dutch railway service is known as NS and their website is pretty bad from an English-speaker's perspective. I know enough Dutch (and can infer enough from the context) to find out when the next train from Eindhoven to Amsterdam is for example, but finding the Lost and Found page eluded me for a long time. It wasn't until I remembered that the central Lost and Found is in Utrecht that I managed to find it via google. So anyway, here is a link that I hope the googlebot will find for the next hapless English-speaker in my unfortunate situation. NS Lost and Found

Disaster averted

Mon, 06 Dec 2004 10:01:02 +0100

The bag has beend found! It is on its way to Utrecht now, and apparently I can pick it up later this week. This makes me indescribably happy and reaffirms my faith in humanity. Humanity is good mmm-kay?

Disaster struck

Sun, 05 Dec 2004 17:34:53 +0100

On Wednesday night, I was going to a dinner to celebrate the fact that Otfried is moving to Korea soon (there must be a better way of saying that) and I left my bag on the train. Just in case anyone has seen a bag on a train lately, here are the relevant details: it is the MEC bag pictured below. I was on the intercity train between Eindhoven and 's Hertogensbosch that left Eindhoven at 17:27 Wednesday evening. The bag has some very important things for me in it, so it really sucks that I was so clumsy. If you have any info about it, please let me know.

Travel schedule

Sun, 05 Dec 2004 17:15:38 +0100

This is as much for me as for anyone else. I will be leaving the Netherlands on December 20 (at some time) and going through Heathrow to Calgary. I will probably spend the Christmas to New Year time with my parents in Calgary. After that time, I would like to go out to Vancouver and spend a few days there. Then I will be back in Calgary to begin my fabulous journey back to the Netherlands.

I will leave Calgary in the morning of January 9, then spend a 12 hour layover in Ottawa. After that, I will be on my way to London Heathrow, arriving on January 10 in the morning London time. Then I will be staying in the London area for about a day before leaving from London Stansted to return to Eindhoven. I have a ticket back from London to Calgary on July 19 but I don't have to use it.

Moving away from sexps for blog posts

Sun, 05 Dec 2004 16:58:34 +0100

The mechanics for posting to my blog aren't very interesting, but they sure are a pain in the ass for me. I need to make an emacs mode that helps me with it. The problem is that I can't think of a really good way to have the power I want -- making links is the hard problem. If I can't have a convenient way to make them in the mode, then there is a real problem with the mode.

The scope of this site

Sat, 04 Dec 2004 19:19:25 +0100

I have been thinking about what I should do with this site. I am a graduate student, and I don't really want to write about what I am working on because I don't think other people would think it's interesting. And frankly, my research moves so slowly that the blogging would probably get depressing. There are other sites run by academics that go into the politics and gossip of being an academic, but I'm really not into that and I don't think I would add anything.

Another thing I could write about is my life. This might be interesting to my family. Might. I am thinking about starting a moblog (a blog with pictures and snippets of text that say where I was and what I was doing when I took the picture) -- I have a cameraphone that takes really crappy pictures, so that could be fun.

I enjoy reading and learning about many things that other bloggers are already talking about -- for example technology, politics, a little bit of marketing, and some other things. I am afraid that my contribution to any of these spaces would simply be to link to things other people have said and whether or not I agree with them. That seems like something that wouldn't add a lot to the world.

So what am I going to do? I think I am going to just do a combination of all the above. Eventually, I might get really excited about something, but I would still post about my life.