Nice and hot

If you like your food hot, you will have pondered about what chili to use for which occasion. Here you see a small selection as available in my local grocery store:


The lone green one (I've eaten all others yesterday) is a Jalapeño. Mild (2000 Scoville) and with agreeable bell pepper taste, so you can use it like another veggie to get some volume (as potatoe replacement). The slim red ones are already in an entirely different league: they're Bird's Eyes (or Rawits), and the winners of this informal contest. They're hot, though not extremely so (100000 Scoville), and have a nice hot taste which will go along with a lot of different food. The yellow ones, of course, are orange Habaneros with 250000 Scoville (*hust*) and the distinct Habanero taste not always desired or liked. Personally, I love it...but the Rawit is a universal hot Chili you can enjoy even on a steak! 😉


Folding@Home is, as so many times before, broken on Linux systems with a reasonably up-to-date version of libc and either segfaults or stops working with bizzare error messages such as 'Couldn't CosmHTTPOpen' or 'Cannot get ID from server'. Since many solutions posted in the interwebs are wrong or deprecated, here's a viable workaround which I know to work on Debian, Ubuntu, Mandriva, and Arch (and Fedora, I've heard).

Install nscd, then check your /etc/nscd.conf with your $EDITOR. Find the following line:

enable-cache            hosts           no

Change "no" to "yes", if necessary, then save and exit the editor. Finally, issue a

service nscd restart

and fold away.


A friend asked me if my enthusiasm to use my bike to go to work still persists after my move to a new place.

It does.

As a matter of fact, I'm truly privileged in this respect. My new location allows me to take partly the same route as previously, namely, along the entire west-east axis of the Tiergarten with the Interconti and the Sony Center as the respective entry points. What's more, I've realized that my new route touches some of the most scenic places of Berlin:

Prager Platz
Witttenbergplatz und das KaDeWe
Großer Stern

My daily ride to the lab could be sold to tourists as a sight-seeing tour. 😊

Keep pushin'

Mandriva 2010 Spring has been finalized (after several serious delays) at the 9th of July. The 'Mandriva Update Applet' apparently missed this event, since it did not yet offer me an upgrade, though several mirrors have been populated days ago. I finally lost my patience, and told the senile applet that a new version is out:

mdkapplet-upgrade-helper --new_distro_version=2010.1

Since then, the applet is dutyfully doing its job:

upgrade to 2010 spring

PS: Since probably nobody will be able to make the link, I'll explain: the title is a reference to a favored song of mine. 8)

Particle analysis (Berlin, 37°C...)

Inhomogeneous, granular material is encountered in many scientific disciplines. For instance, think about living tissue and bacteria cultures in biology, or polycrystals and composite materials in material science. As an example, look at the following micrograph, which was acquired by a scanning electron microscope and shows a top-view of upright standing GaN nanowires on a Si substrate. The horizontal size of this micrograph is 5 µm, i.e., about 10 times smaller than the diameter of an average human hair. The nanowires are really quite small: much more tiny than bacteria, they are similar in size to common virii.

Quite some bed of nails, huh? As you can easily see, all of these nanowires are individual entities, quite different in size and shape from their neighbors. It is an interesting, but very cumbersome task to study the properties of individual nanowires. Fortunately, one is often interested in the properties of a nanowire ensemble, thus averaging over the response of the individuals consituting the entire population.

To correlate this average to the geometrical reality of the given ensemble, one desires to obtain an impression on the actual distribution of sizes in this ensemble. As a physicist, I'd like to simplify that even further, and get a histogram of the size (i.e., area) of the individual entities represented by a disk of the same area (the eqivalent disk).

There are two open-source programs available which are capable of performing this task in a straightforward way: Gwyddion and ImageJ.

In Gwyddion, you load the image, define its physical dimensions, set a threshold either by height or watershed (e.g., /Data Process/Grains/Mark by Threshold), and then plot or export the data you are interested in by selecting /Data Process/Grains/Distributions and checking "Equivalent Disc Radius" in the Area menu. The result using a height threshold of 25%:


This plot was obtained in a terminal by the following set of commands:

ipython -pylab

To do the same using ImageJ, you first load the image and draw a line across it. To set the scale, you enter the length of the line in /Analyze/Set Scale. To set a threshold, you might use the fully automated way (/Process/Binary/Make Binary) or do it manually (/Image/Adjust/Threshold). Then, use the '/Analyze/Analyze Particles' dialog. The advantage over Gwyddion is that you can set limits on both the minimum and maximum particle size, which I've used here to exclude diameters less than 10 nm. Apart from the smaller sizes, the result is quite similar:


For plotting the results, qtiplot may be used as a graphically oriented alternative. I've analyzed another (smaller) scanning electron micrograph with both programs and obtained the following histograms:

qtiplot distributions

You see that I'm working despite the suffocating heat. What else could I do? *Teilnamslosdalieg* ist too boring on a long term 😉

Berlin, 33°C

viel zu heiß

I planned to post an entry on encryption, but that will have to wait till the next weekend iceage. *teilnamslosdalieg*

Open source math

Years ago, when the rigid licensing conditions of Mathworks, OriginLab and Wolfram and the exorbitant costs of their software began to be more than an occasional irritation, I set my hopes on the open source world. Surely there would be alternatives, I reasoned, and I would take these and be free of any annoying complications arising from proprietary software and their licensing models.

I've indeed used SciLab for a while, but eventually everyday duties and the necessity to be compatible with the rest of the world got the better of me and I returned to the commercial offers mentioned above. 😞

Now, the licensing policies of the above producers of scientific software have not, generally speaking, evolved in any desirable way—to put it mildly. It's not only the prohibitive prices, but also the involved licensing models which require considerable time to comprehend for those responsible for the purchase of this software, i.e., me.

I hence recently decided to have a closer look onto the present status of open-source scientific software. To quickly get up-to-date, I've followed discussions concerning this subject across the interwebs. The usenet proved to be particularly resourceful, and I've learned a lot from discussions there. First, I've seen that I'm not alone with my growing resentments, and second, I singled out two projects which are frequently mentioned and mostly highly commended. I'm mightily impressed by both of these projects already after the first, cursory glance.

The first of these projects is PyLab, which is perceived by many users as a viable alternative to MatLab, but is stated by its main developer to be a vision rather than a completed computational environment. The situation, however, is not nearly as bad as described in this self-critical analysis of Keir Mierle. Pylab is based on Python and the modules NumPy, SciPy, Matplotlib, and iPython, the latter being an adapted python console for interactive work. All of these modules as well as the required dependencies are readily available in modern Linux distributions. After installation, just start the console with 'ipython -pylab', and issue the following commands taken straight from the plotting tutorial:

x,y = ogrid[-1.:1.:.01, -1.:1.:.01]
z = 3*y*(3*x**2-y**2)/4 + .5*cos(6*pi * sqrt(x**2 +y**2) + arctan2(x,y))
imshow(z, origin='lower', extent=[-1,1,-1,1])   
contour(z, origin='lower', extent=[-1,1,-1,1])
plot(x[:], z[50, :])

Three plots in one in five lines of code. The figure, by the way, can be saved in png format by a simple


Sympy and its isympy terminal adds symbolic abilities to pylab and is hoped to grow up to a fully-featured computer algebra system (CAS). Well, sympy is quite useful already at the present stage:

Without doubt, isympy is the most capable console calculator available. 😉 Oh, and before I forget (and since I like to clear consoles 😏):

import os

Finally, Spyder is a Qt4-based IDE for iPython and friends. As beautiful as practical!

The other project I became interested in is Sage, which is also based on python but represents an attempt to ... well, in my opinion, to become the ultimate math software on earth. 😄 Seriously, Sage is a unified interface for many major mathematical packages such as pylab, maxima, octave, scilab, R, singular, axiom, tachyon, jmol and many more I even don't know by name. Sage is executed in a client-server architecture with the client based on Ajax, thus working well in all modern web browsers.

I just see that I don't have the time to describe my first impressions of Sage in any detail anymore, since it's close to 4 pm and thus time for:


Very briefly: Sage left me speechless. Try for yourself!

It's Vuvuzela time! Yeeeehaw! crazy bugger

Beyond Tabasco

I just love Tabasco, really, but its distinct taste coupled with its low heat of only 5000 Scoville makes it often a nonideal choice. Hot Mamas is a German manufacturer of Chili sauces. Despite the fact that their online representation as well as the design of their actual products (see below) leaves a rather dubious impression, the sauce I tried (No 14) is truly excellent. Very tasty, not as dominant as Tabasco (much less vinegar, more garlic, distinct Habenero note), and, most importantly, a 100 000 Scovilles (in reality, perhaps 30 000, which is still respectable for a sauce free of extracts). Besides, it's available in some of my local grocery stores. 😊

No 14

Beyond e-mail

We are currently experiencing a paradigm shift. Just a few years ago, e-mail was used to communicate, and file transfer was done via ftp. Nowadays, the majority of users employs e-mail for file transfer and communicates via skype. Diverting so extremely from the intended use of programs and protocols, however, may have adverse consequences for all of us, not only for the least-experienced of users (the LEOs). To illustrate this statement, read the following (only slightly distorted) anecdote:

The boss is calling. E-mail isn't working, he states gravely, and the half whining, half accusatory tone of his voice leaves no doubt as to who he believes to be responsible for this absolutely (!) unbearable (!) situation. Before hanging up, he takes the oppurtunity and emphasizes his enthusiasm and his devotion in general, and his bitter disappointment and utter frustration with the IT "situation" in particular. 😄

Instead of checking the mail servers (which you know to run since months without a single hitch), you immediately leave for the bosses office. And what you find there is a Thunderbird struggling to synchronize the local inbox after a hard reset has corrupted the index. Since said inbox is now at 60 GB, that takes some time ...

How can an e-mail inbox grow to such grotesque dimensions? Well, I tell you how. The secretary of a very important contact of the boss is receiving an interactive pdf form (212 kB). Printing this form, so that $VIP can mark the checkboxes with his MontBlanc fountain pen, our sweet secretrary next scans the document with the highest available resolution, and then uses Adobe Distiller in its highest quality settings to export it as a pdf (41 MB) which she then attachs to the mail to the boss. Since she finds this procedure to work flawlessly, Mrs. Sweet uses the same work flow for all documents, including those written by herself in MS Word (no, she did not yet discover the pdf icon there). A seminar announcement consisting of two lines of text thus takes 2 MB. You get the same seminar announcement, by the way, also from Mrs. Sweet's colleague Ms. Sour, who was mightily impressed by Mrs. Sweet's efficient work flow.

Now, of course one can try and educate the LEOs, explaining that e-mail isn't designed for exchanging files above a certain size. What's more effective, however, is to simply limit the LEOs' mailbox size to 5 MB ... 😉

The remaining question, of course, is what to offer the LEOs as alternative. We nerds can put large files on our own server (in my case the one serving the very page you're reading). Other nerds then access these files via scp and sftp. Other people may also exchange files via Ubuntu One or Dropbox. Hell, one could even use Opera Unite for that purpose. All of that, however, is way too complicated for Mrs. Sweet and her boss.

Honest answer? Well, there's no solution. Two LEOs just shouldn't exchange files. 😄

What I personally use for exchanging files with LEOs? A minimum webserver provided by the python scripts droopy and woof (and I'm not alone there). All the actual 'work' is up to me then (including forwarding ports on firewalls and the likes 😉 ).

#droopy -m "<p>This is the home of <i>Cobra</i>, the friendly snake.<br><br>
Please upload your files.</p>" -p ~/temp/upload/mandriva.jpg -d ~/temp/upload

What Mrs. Sweet sees is after I tell her the URL:

happy snake

#woof -p 8000 -Z ~/temp/download/

Here, the download dialog of the browser opens, directly offering the zipped directory defined in the shell script.