Thành viên:Phạm Thạch Thảo/Note: Computation tools

• HPC.

• Install R from source:

1. Install gcc. For Linux, that would be easy, usually already included in the OS basic. For Mac, one has to install Xcode (free from apple), which includes gcc. Installing gcc directly from source is kind of difficult because it requires a bootstrap process which asks for some C-compiler. Unfortunately Xcode doesn't have gfortran, which R loves. The compiler can be found in the GNU page, under the session gfortran.

2. Just after that one is able to use "./configure and make"! This installs R "on the spot".

3. To install the whole tree of R to the a specific place, one needs to use

./configure prefix={location, usually /usr/local/}
make install

4. For some locations, one certainly needs the administrator's control, which is... painful.

-> Better use the built-package of R in order to save you stupid hours, unless you want to look like an expert... (This graphic interface takes 5 minutes! And you still can use the terminal, pretending to be... an expert!)

5. Usually the default setting for text encoding in Mac seems not to be UTF-8, this makes the R initiation appears with a warning. As a result, X11 does not function properly and plotting is out of work. To reset the text encoding, we can feed the terminal with the command

defaults write org.R-project.R force.LANG en_US.UTF-8

• Open sources for computer algebra systems. According to wikipedia benchmark, among the leading softwares in performance capability is SAGE (only the diophantine equation solving lacking). However, with some more lacking of performance, Maxima is probably more friendly to users used to Mathematica.