Home Office

The spread of SARS-CoV-2 has made it advisable for many people to work from home. I and my colleagues are doing that now for four weeks, and it's working very well. For me, home office isn't new: I use this possibility since a decade whenever I have a task at hand requiring particular concentration and focus. Writing papers or proposals is such a task, or developing and implementing a quantitative model to understand experimental data (that's what lucky physicists do for a living). In fact, I've been asked in January by colleagues to help with the development of such a model, which I thought to be challenging, but didn't expect to be as difficult as it actually turned out to be. For most of the time, I was rather cluelessly poking around in a forest of equations and not getting anywhere.

During the last days, I made an effort to refocus on this issue, and not only for a few hours, but a couple of days: you go to bed with the problem and wake up with it, and there's nothing to distract you from it. This kind of total concentration is simply not possible in the daily office routine, but I can do it at home, basically returning to my time as a student where every living moment was devoted to problem solving. What greatly helps with reaching this trance-like state is having no kids, an understanding wife, and softly purring cats that love to sleep in the chairs on my left and right. The breakthrough occurred after two days, all of a sudden, like a flash. I still had to solve technical problems, but the direction was clear. These are the moments that every scientist cherishes and holds most dear: the intense joy to have solved the problem, to have broken the code. 😌

I realize that not everybody has the same favorable boundary conditions as I do, or even the luxury to compare. And I understand that the situation is very different with young kids instead of cats. 😉 But still, I'm really tired reading commentaries in the newspapers moaning about the “solitary confinement”, and how unbearable it is. Most of them stem from rather young people with a smartphone glued to their right hand, and the strong belief to have the god-given right to party. Even worse are the characters with a political agenda, bitterly complaining about violations of our constitutional rights and predicting the end of democracy. What unites these two apparently very different groups is their failure to understand even the most simple arithmetic. And yes, there's no need for calculus to understand the simple concept of the exponential spread of a virus.

More realistic models are based on systems of differential equations similar to the ones describing the zombie apocalypse. The infection rate of the human population depends on the infection probability when a human and a zombie meet. Similarly, in epidemiology, the spread of an infectious disease is characterized by R0, the basic reproduction number. This number determines how fast the infection spreads (i.e., the slope of the exponential), and if it decreases with time (which is highly desirable), the curve “flattens”. The curve remains, however, an exponential as long as R0 > 1.

The greatest shortcoming of the human race is our inability to understand the exponential function. A. A. Bartlett


In a previous post, I've remarked:

“If the distributor commands over virtually unlimited resources, and compression speed is thus not an issue, brotli and zstd are clearly superior to all other choices. That's how we would like to have our updates: small and fast to decompress.”

And not even a year later, we get this announcement. My own measurements indicated a factor of 8 increase in decompression speed, but the Arch team even sees a factor of 14. Great! ☺

There are also a few settings in /etc/makepkg.conf that may greatly accelerate the installation of packages from the AUR. All details can be found in the Arch Wiki, but here are the modifications I'm using in the order of appearance:

# building optimized binaries
CFLAGS="-march=native -O2 -pipe -fstack-protector-strong -fno-plt"
# use all cores for compiling
# compile in RAM disk
# use package cache of pacman
# enable multicore compression for the algorithms supporting it
COMPRESSGZ=(pigz -c -f -n)
COMPRESSBZ2=(lbzip2 -c -f)
COMPRESSXZ=(xz -c -z - --threads=0)
COMPRESSZST=(zstd -c -z -q - --threads=0)
# use lz4 as default package format (the command line lz4 does not yet support multi-threading, but it's still faster than anything else)

I didn't perform any systematic measurements, but some AUR packages seem to install in seconds, when it took minutes in the default configuration. YMMV, but it's worth to give it a try.