Every so often I get bored writing about statistical analyses, software and torturing data and spend time in alternative creative endeavors: taking and processing pictures, writing short stories or exploring new research topics. The former is, mostly, covered in 500px, I keep the stories private and I’m just starting to play with bioacoustics.
While I’ve been away from this blog came the Google Reader debacle; Google announced that Reader will be canned mid-year, probably because they want to move everyone towards Google+. Let’s be straightforward, it is not the end of the world but a (relatively) minor annoyance. The main consequence is that this decision led me to reevaluate my relationship with Google services and the result is that I’m replacing most services, particularly those where what I consider private information is stored.
My work email (Luis.Apiolaza@canterbury.ac.nz) stayed the same while I moved my Google calendar back to my work’s exchange server. I setup my personal email address in one of my domains, served by Zoho. There are no ads in this account. I opted for this to avoid worrying about maintaining email servers, spam filtering, etc. I’ll see how it works, but if it doesn’t, will swap it for another service: my email address will stay the same.
I barely use Google Docs, so it won’t be a big deal to move from them. I deleted my Google+ account, no big loss. I’m keeping my Gmail account for a little while while I transition registration to various services. Nevertheless, the most difficult services to replace are Search, Maps and Scholar, which I’m now using without being logged-in in Google. I’m testing Duck Duck Go for search (kind of OK), while I’m sticking to Maps and, particularly, to Scholar. Funnily enough I have access to Web of Science and Scopus—two well-known academic search services—though the university and I will often prefer to look in Scholar, which is easier to use, more responsive and much better coverage of the literature; particularly of conferences and reports.
Google didn’t remove a small service. It did remove my confidence on their whole ecosystem.
I’ve been moving part of my work to university servers, where I’m just one more peasant user with little privileges. In exchange, I can access the jobs from anywhere and I can access multiple processors if needed. Given that I have a sieve-like memory, where configuration details quickly disappear through many small holes, I’m documenting the little steps needed to move my work environment there.
The server provides a default R installation but none of the additional packages I often install are available (most people accessing the servers don’t use R). I could contact the administrator to get them installed, but I’ve opted for installing them under my user space. For that I followed the instructions presented here, which in summary require adding the name of the default folder (
/hpc/home/luis/rpackages) for the local library of packages to my .bashrc file:
R_LIBS="/hpc/home/luis/rpackages" export R_LIBS
I also have a temporary folder (called
rpackages)in the account where I move the source of the packages to be installed (using SFTP). Once the R session is started it is possible to check that the local folder is first in the library path, confirming that
R_LIBS has been made available to R.
Then I can install the packages I moved to the server with SFTP from the temporary folder to the local library using
.libPaths() #  "/hpc/home/luis/rpackages" #  "/hpc/home/projects/packages/local.linux.ppc/pkg/R/2.15.1/lib64/R/library" # install.packages("~/temporary/plyr_1.8.tar.gz", lib="/hpc/home/luis/rpackages", repos=NULL) # * installing *source* package 'plyr' ... # ** package 'plyr' successfully unpacked and MD5 sums checked # ** libs # gcc -std=gnu99 -I/usr/local/pkg/R/2.15.1/lib64/R/include -DNDEBUG -fPIC -O2 -c loop-apply.c -o loop-apply.o # gcc -std=gnu99 -I/usr/local/pkg/R/2.15.1/lib64/R/include -DNDEBUG -fPIC -O2 -c split-numeric.c -o split-numeric.o # gcc -std=gnu99 -shared -Wl,--as-needed -o plyr.so loop-apply.o split-numeric.o # installing to /hpc/home/luis/rpackages/plyr/libs # ** R # ** data # ** moving datasets to lazyload DB # ** inst # ** preparing package for lazy loading # ** help # *** installing help indices # ** building package indices # ** testing if installed package can be loaded # # * DONE (plyr)
Now we can load the package as normal:
require(plyr) # Loading required package: plyr
Nothing complicated or groundbreaking, just writing down the small details before I forget them.
This is one of those times of the year: struggling to keep the head above the water, roughly one month before the last lecture of the semester. On top trying to squeeze trips, meetings and presentations in between while dealing with man flu.
- Creating a list of introductory-level Bayesian blogs. Some suggestions in the comments: Probably Overthinking It, Maximum Entropy & Deus Dia Pente.
- Materials for Statistical Modelling in Stata. HT: Christophe Lalanne.
- That damned R-squared! understanding through simulation by Arthur Charpentier.
- Strange headlines: ‘Livestock Improvement Corporation offers semen credit in wake of ‘hairy calves’ genetic defect’. HT: Mem Sommerville.
- Attention Deficit Disorder sufferers have moved, at least for a while, from talking about Julia to talking about Stan.
- I bought a copy of Wes McKinney’s Python for Data Analysis. Worth the price.
This picture reminded me of The Oatmeal’s How to tell if your cat is plotting to kill you.