BOINC!

I decided to try and do something useful with my computer, and set up BOINC so I could choose from a list of projects to use my computing cycles.

Ok, so I just pulled in the debian package with aptitude but after installation then it didn't run as a daemon as the document describes so I decided to run it from the command line. That's where my troubles began.

I understood from the documentation that you could run boinc from the command line and choose some projects to support and voila, everything would be up and running.

What happened to me was that I choose some projects but I didn't get any work to do, my client was querying the projects for work but none was given (sometimes I got a message saying that the project needed x amount of disk space but I had 0 Mb available when in fact I had more then 9 Gb available). I was getting a bit frustrated but decided to try and run the program as a daemon instead.

That on the other hand did work! Now I am happily doing my work and my computer is doing someone else's work for me ;-)

So if you try to run the boinc client from command line as a normal user then it wont work, you have to be logged in as user boinc or run the client as a daemon. At least that's how it works here in debian land.

Ohh yeh, I decided to go with Rosetta@home and climateprediction.net

High temperature superconductors

This says it all.

It would be excellent to get easy to produce hight temperature superconductors. High speed railways for everyone!!! :-)

The Enemies of Reason

Richard Dawkins of whom I am a great admirer hosted a two episode documentary about non-science things like astrology, alternative healing and the like called The Enemies of Reason. The shows are available on google video here: Part 1 and Part 2.

Highly recommended.

Some cool things

Here are a few cool things I have found during my wanderings over the net lately:

A draft to a paper to be published in Scientific American about the limit of quantum computing.

A blog about cool stuff found on the preprint site for papers on physics stuff.

A blog post about why linux doesn't cut it in the mainstream and an experiment the author is conducting to try to prove his hypothesis (won't be a conclusive experiment, but could give reason to further experiments in sociology :-) )

Solar evidence points to human causes of climate change, take that climate warming by human intervention sceptics! :-)

The most powerful laser ever built.

That all for now.

error bars in gnuplot

I use gnuplot to make all my graphs for my thesis.

Some days ago I decided that it was about time I put some errorbars on the experimental values in my graphs (this is something that is pretty important in physics, and should be in all subjects that handle measured data) so I calculated my standard deviation from some fits I made and used them as my errors.

Now to plot them. Hmmm .. how do I do that?

Well, if you have a datafile called "data" and in it you have 3 columns, the first column representing the x-axis, the second column the data (the y-axis) and the 3 column the +/- error values on the data values in the second column then you just give the command

plot "data" using 1:2:3 with yerrorbars

and you get the data values with the errorbars set correctly.

Except it didn't really plot well for me. The lines that made up the errorbars were dashed in stead of solid when I printed them out into a eps file and if I wanted to have different colors, if they were all red like the default is then there was no problem, but having one data set green and another blue and a function that was supposed to be compared to the data as red, then I had problems.

I wondered a lot about that, tried several other plotting utilities (gees those graphical utilities are bad ... not intuitive at all I think) and went no where. I asked the nice people on the news group comp.graphics.apps.gnuplot and there I was just told to either use the candlesticks mode or define the color by using the linetype option.

So I tried and tried with no luck.

Until finally I came across one sentence there that said in plain text to my question about why the error bars were dashed

That would be because your terminal setting has patterned linetypes (no "solid" option to the postscript driver)

and then it continued on to answer some more questions so this part did not register with me by some reason the first time I read this.

It seems that the default of gnuplot is to set the dash option with the postscript driver when you issue

set terminal postscript eps

So to get a solid line with your error bars in your eps files you have to issue

set terminal postscript eps solid

Took me quite a time to figure this little thing out but should have taken me a lot less time if I had read better what was written to me in that post :-)

Lesson of the day, read what is written for you :-)

GeoTIFF and NetCDF

A friend of mine is studying glaciology in New Zealand and he is working there with data taken with the GeoTIFF format but needs it in NetCDF to put it into the model they are using at the university so he wrote a little C program to convert between the two using the libgeotiff and NetCDF libraries.

I got permission for him to put the program out on the net for others who are interested to have a look at it and also to put a GPL license on it. You can find it here along with a small makefile to compile it (read the makefile before you try it, it might not work on your computer), or you can just compile it with your own compiler.

At the moment the software is limited to his specific needs but it is well commented so people should not have to much problem in adapting it to ones needs. Perhaps we will work on it more in the future and get it to be more general.

Wikipedia for experts

Scholarpedia is a new wikipedia style encyclopedia, or should it be called an encyclopedia when it is not meant to cover everything from the smallest to the biggest?

Scholarpedia is supposed to be a specialized wikipedia where experts in their fields write about the stuff they know about. So it is supposed to take subjects from wikipedia like saltatory conduction and make a pretty in depth article about it by an expert in the field. When the first article is created a curator is nominated for the article and every time someone wants to edit the article the edits go through the curator first and he decides if the edits are appropriate or not. It's not a completely free editing encyclopedia but almost (of course people could get dissatisfied with the curator, how the community of scholarpedia will react to that I have no idea).

Hope it will do good and get more physics articles in there :-)
At the moment it is more focused on computer engineering and biology related articles.

Climate change sceptics thwarted

The BBC have just put out a pretty nice listing of the top 10 topics that climate change sceptics have made in the past and the scientists response to them.

Read it here.

I never remember stuff like this, I read it and then I forgett it, and then I meet a sceptic and try (unsuccsessfully) to explain to him why he is most certainly wrong and just end up saying like a complete idiot "your wrong because your wrong" :-)

I should print this out and read it every time I go to bed. Thats a good brainwashing.

A little update

I am so close to finishing my masters now I think, that I have begun for serious to apply for Ph.D. programs.

So far I have applied for one at DTU, a really interesting project there that is about making quantum feedback controllers so that measurements that have to be so precise that the quantum noise that follows every experiment made is the lower bound of the precision in the experiment. Usually experiments have lots more uncertainty in them, the equipment that is used has some uncertainty but using feedback controllers than you can try to minimize this uncertainty.

The other position is in Stuttgart, in the group of Prof. Tilman Pfau. They are investigating a lot of things but they are searching for a Ph.D. student in their dipolar BEC of chromium atoms. Sounds fancy ayyhh?? :-)

So I am going down to Stuttgart in october and giving them a talk and they will show me around the lab and we will have a little discussion about the prognosis of me being hired in the position .... jaaaayyyy! :-)

So not much blogging going on, but I will update this when I know something about my future plans.

Bill Bryson

Just finished Bill Brysons book A Short History of Nearly Everything. I really liked it, gave a good overview of its subject, not to detailed so they layman looses interests but enough to keep the scientist interested in the history part of it. As for my own preferences I liked the part about life, I dont know much about biology so that part was really fun for me to read.

I recommend this book for everyone.