Previous Up Next
Dutch / Nederlands

Dairy, July 2011



Sun Mon Tue Wed Thu Fri Sat
                      1   2
  3   4   5   6   7   8   9
 10  11  12  13  14  15  16
 17  18  19  20  21  22  23
 24  25  26  27  28  29  30
 31


Friday, July 1, 2011

The world of carton

When Annabel heard about the design contest The world of carton (De wereld van karton in Dutch), she quite quickly got the idea to transform a drinking package into a general purpose holder that can be used to hold a mobile device close to the wall when it is being charged. It is a simple but clever design, I think. She joined the contest, but might be just too late, because the closing date is today. On the right top a picture of her design in case it does not win a price. The picture on the right shows her design.


Saturday, July 2, 2011

Skûtjesilen

Today, we had our yearly Skûtjesilen trip. (Except for last year, when we went to Barcelona.) We left around 8 o'clock from the office and arrived in Elahuizen around 10 o'clock. We sailed with three skûtjes. I went with the Eelkje II again. We sailed from 11 till 1. I stayed on the front of the ship helping with the the jib (foresail). Then we had a packaged lunch at an island. We sailed through some canals and had to use the engine for some part. We also had an extra stop for some icecream, a reward for a conteste that some colleagues won. When we left around four, the wind was quite strong, and we did some real sailing. I stayed at the back of the ship with the captain and helped a little. Then around half past four, we noticed a skûtje that had capsized. (See the picture on the right.) It was quickly surrounded by some other skûtjes. First we thought that it was not one of our ships, but after a phone call it turned out that it was one of ours. We lowered the sailed and went back to collect our colleagues, who had already transfered to another skûtje. Not all of them had fallen in the water. Luckily, there were no personal injuries, except for some small cuts. But almost all of them had wet feets. Luckily, we had some dry sailing clothes on our ship. The water was not that cold, but staying in the wind with wet clothes is not good. Later all the personal belongings where collected from the ship. Some electronic devices did not survive the water and died. We had some BBQ. And a close colleague of mine, who deals with the helpdesk, was nominated for employee of the year. Then we heard that the skûtje had sunken shortly after it had been erected again, because water flowed into it. The lake is at most 3 meters (10 feet) deep, and they where going to try to salvage it with some crane ships later that evening. Around 8 we left for home. We arrived in Enschede around 10 o'clock. (The KML file of todays GPS tracks to be viewed in Google Earth or shown with Google Maps.)


Thursday, July 7, 2011

Two support beams

A few minutes after noon, we heard a loud rumble outside. Yesterday, a colleague of mine and I had watched the cranes beside De Grolsch Veste, the stadium of the soccer club F.C. Twente, slowly swinging due to some strong winds, and discussed the possibility of them crashing into the stadium. It immediately came to my mind up. I jumped to the window and saw the last second of some part of the roof collapsing. Everybody looked out of the window. I took two pictures. For our point of few we could only see part of the roof, as our view is obstructed by another building. I wrote some messages on twitter including one (in Dutch) mentioning that two support beams (twee steunbalken in Dutch, had collapsed. I meant to say: two roof beams. Shortly after that, I discovered that the whole roof on one side (except for the corner) that was being constructed, had collapsed. I wrote another tweet to correct my previous tweet. A little later, I discovered that the expression twee steunbalken had been copied in a very brief news item of Twentse Courant Tubantia, a local newspaper. I strongly suspect that this was taken from my tweet, and I wrote a comment to remark that this was wrong. Comments are always reviewed before they are shown, so, I had expected that the reviewer would have noticed my remark and correct the mistake, but instead the comment was placed below the news item. During the afternoon, I noticed that the expression twee steunbalken was copied by many other news websites. I even found something resembling this on Wikipedia and removed it. Later on the day, I also found the expression two support beams on many websites around the world. I am afraid that all those uses can be traced back to me using the expression twee steunbalken on twitter, as I was the first to use it.


Saturday, July 9, 2011

Horizontal and cross beams

It is becoming clear that the collapse of part of the roof of De Grolsch Veste is due to missing horizontal and cross beams. Possibly adding the additional weight of a large screen on the edge of the roof might also contributed to the collapse of the roof. Although it should be remarked that this screen already had been there for some weeks. It seems that seven very solid horizontal beams connecting the most outside points of the roof trusses where still missing. Also two horizontal beams on the top of the roof trusses where missing. From an important combination of four cross beams, three beams where still lacking. Of another fourteen cross beams only four where placed so far. The theory that the missing beams where essential for the collapse to happen is supported by the fact that the first roof truss that did not collapse, had horizontal and cross beams on one side attached. From some airial pictures it is clear that some strong forces worked on this truss, but nevertheless it looks unharmed. Another interesting picture shows the collapsed roof construction from the outside. Seven roof trusses where damaged. With four the most outside node collapsed to the right, The following two collapsed to the left and the last did not collapse on the outside node. Which is rather surprising. It could be explained by the fact that the horizontal beam of this truss is disconnected at the spot where it should have been connected with bolts. It would be interesting to know when this connection snapped, at the beginning or at the end of the collapse. What surprises me about the picture is that how little the trusses moved to the side. From the fact that they only had to move so little to the side, (I think) one could conclude something about the forces on the trusses, especially for the sideway force needed to make the trusses collapse. It is possible that only a slide horizontal movement (of the edge of the roof) was sufficient to make it collapse due to the instability of the roof construction. I think it will be interesting to know the construction planning and to find out why the horizontal and cross beams where not installed yet, while the construction was put under load. Who was responsable for the planning and/or was there any deviation from the planning.


Sunday, July 10, 2011

Oak tree

In the evening, Annabel and I, cut down our oak tree in the back garden. It grew from an acorn that Annabel put in the ground six or seven years ago. Now it was about 5 meter (15 feet) tall. The problem is that it is very close to the wall of the extention of our neighbours, and I am afraid that it will damage the foundation in due time. (Picture from June 6, 2010.) I also cut some of the roots going out from the main root. Our neigbour also helped a little to cut the trunk in some smaller pieces. Annabel found a very small oak tree on the lawn. I don't know how it got there, but I will cherish it.


Monday, July 11, 2011

Katla showing activity

It seems Katla is showing some activity lately with swarms of earthquakes and harmonic tremors. I found a good blog: Iceland Volcano and Earthquake blog.

Monte-Carlo Tree Search

This morning, I accidently found the PhD thesis Monte-Carlo Tree Search by Guillaume Chaslot in some bookshelve. I wondered how it got there, as I could not imagine that anybody in the office knew what it was about. Then I heard that a colleague of mine often received PhD thesis because his involvement with some kind of board. I asked him of I could get it and he replied: Sure! The subject of the thesis is related to computer programs playing Go. I am also interested into the subject, because it might be a technique applicable to Havannah as well. There seems already to be some work on this: At home, I started reading the thesis.


Wednesday, July 13, 2011

Computer learns language by playing games

Today, I came across Computer learns language by playing games, which links to the paper Learning to Win by Reading Manuals in a Monte-Carlo Framework. I submitted a story to Slashdot, which got accepted with some slight editting.

The Symple challenge

I received an email from Christian Freeling in which he states that the abstract game Symple will be the next major challange for the AI community. A decade ago, he predicted that the abstract game Havannah would pose great problems for the AI community. That indeed proved to be the case and it still holds, but, as he acknowledges, Monte-Carlo Tree Search and UCT (Upper-Confidence-Tree) have given the efforts a firm handle. Already good results have been shown for computer programs playing Go. Havannah received less attention so far, but maybe that may change when his 1000 Euro challenge is coming to a close next year. Symple indeed poses a new challenge because of its extreme branching factor in the middle phase of the game, when there are in the order of 1020 different moves to consider. The reason that there are so many moves, is because each move can consist of placing more than one stone depending on the number of groups you have. With Symple you can either decide to create a new group or to (optionally) grow each group on one of its direct neigbours. But maybe it is possible to use a Monte-Carlo technique to determine which is the best move for each group and from this information construct a best multi-stone move, or at least limit the search space by trying various combinations of good moves from each group.


Saturday, July 16, 2011

Summer crossing

Today, I visited bookshop De Slegte rondgekeken. Outside on a table I found Een zomer lang, the Dutch translation of Summer crossing by Truman Capote, ISBN:9789023420590. At 11:57 bought this book for € 3.49.


Sunday, July 17, 2011

MCTS and Symple

I finished reading the thesis Monte-Carlo Tree Search by Guillaume Chaslot, ISBN:9789085590996. I did not read every letter, because the thesis does contain a lot of repetition, such as the final chapter, which repeats all the conclusions that where already given at the end of each previous chapter. I also did not understand every part of it in great detail. I have to say that it is well written and that it has given me a clear understanding of what Monte-Carlo Tree Search (MCTS) is and how it is applied in programs playing Go. What I have learned is that although MCTS is a great technique it still requires a lot of domain language including expert knowledge from the game in order to create a stong game playing program. I am afraid that the results that have been achieved with Go, cannot be directly applied to a program playing Havannah and that Symple will be a great challenge.

After I wrote my previous entry about Symple, there has been some discussion with Christian Freeling and Benedikt Rosenau. I realized that my idea would work good in the later phases of the game, but that it would likely fail in the first half of the game. And it should be noted that it is this half which is the most important, because Symple lacks any 'dramatic' aspects, such as capturing. Also there is not much room to fix 'errors' made earlier on in the game. This makes it a very strategic game.


Monday, July 18, 2011

Havannah and Fuego

Last year, the CMPUT 655 - winter 2010 course from the Department of Computer Science of University of Alberta had a Havannah playing program as an assignment. A sample playing program based on Fuego was given. Fuego is a framework for Monte-Carlo Tree Search playing progams. The course was given by Martin Müller, who is also involved in the development of the Fuego framework.


Saturday, July 23, 2011

Birthday party Meindert

Today, Andy and I went to the birthday party of my old friend Meindert. There we met with his family, some of whom I had not seen in more than 20 years, and the family of his girl friend and some other friends. Andy did behave well. I played some connection games against Ton. In the past we often played Go. He now is really into connection games, such as Hex, Havannah, and Atoll. He is selling boards for those games. He is also one of the best, if not the best, Havannah player in the world. I lost from him with three stones ahead. I also asked him about the Havannah playing programs, and he thinks there is a good chance that Christian will lose his bet.

Because of a road being blocked we took a very different route home. I decided to take the highway instead of following an obscure detour like we did on our way going there. KML file of our trip which can be viewed in Google Earth or directly in Google Maps.


Tuesday, July 26, 2011

Extremely Loud & Incredibly Close

Yesterday, I finished reading Extremely Loud & Incredibly Close by Johanthan Safran Foer, ISBN:9780141025186, in which I started reading on April 10. On Thursday, December 23, 2010, I bought a second-hand copy of the book from bookshop De Slegte for € 4,50. On of the reasons was that I was intrigued by typographic experiments, such as the use of photographs, pages with a single line, and alternative spacing. The book even contains a short movie on the last 14 pages of the book. Yesterday and today, I tried to decypher the phone spell digits given on pages 269-271. The grandfather of the main character, who cannot speak, is trying to get a message across by typing the digits matching the letter after he has arrived from abroad on the airport. Last week I already started working on a program to decode the digits using a word list. I downloaded the SCOWL from Kevin's Word List Page and used some parts of it. Yesterday, I scanned the three pages containing the numbers. The I used FreeOCR version 3.0 to get some raw OCR output, which was not of a very good quality. But what can you expect from a free package. I spend some hours correcting the output using the MySample editor, which can display both images and text. Matching-up 'lines' from the scan with the text was relatively easy using Ctrl-Tab for switching between windows inside MySample and using the arrow keys to move around the image. Next I wrote a small program to transformed the corrected OCR output to a file containing the 'sentences' assuming that the question and exclamation marks are used for termination of sentences, or at least occure on word boundaries. When studying the sorted output of the program, I found one or two errors in the corrected OCR output. But it also seems that some errors where introduced when type-setting the book. There seems to be one '3' digit missing between pages 269 and 270. Somewhere else, an '3' seems to be accidently changed into a '7'. I also found How we work: Jonathan Safran Foer, author, which contains some remarks about the digit sequences stating that only the first sentences have some meaning, and that the rest is just random sequences being copied. You can have a look yourself by studying the results yourself.


Saturday, July 30, 2011

Books

Today, I went into the city with Andy. Of course, I also visited some bookshops. First we went to bookshop Kruimeltje, where I bought the PhD thesis: PARELLA: Measurement of Latent Traits by Proximity Items by Herbert Hoijtink for one Euro. I also walked around the second floor with Andy, which is kind of interesting, because in a way it is like a gallery along the walls with even a small 'bridge'. They have a nice collection of rare books by Harry Mulisch.

We also went to bookshop De Slegte. Andy looked in the childrens department and later on the second floor he looked through their collection of second hand comics. At 12:33, I bought the following two books:


This months interesting links


Home | June 2011 | August 2011