Custom Software Installation¶

On ScienceCloud, custom software can be installed using command line, see Update and upgrade. Below, we provide links and a step-by-step guide for installing R, MATLAB, and Gaussian.

Run R¶

Instructions for installing and running R are provided in the general R guide.

Run MATLAB¶

You can run MATLAB on ScienceCloud using containerized modules. Follow the steps below to create an instance, update your system, and load MATLAB.

Create an instance. Use one of the latest images that includes Singularity. Select an image with the *** prefix in its name, e.g., ***Singularity on Ubuntu 24.04 (2024-11-21).

Update and reboot the system

sudo apt update
sudo apt full-upgrade
sudo reboot

List available modules and package versions
```
module av
```
Load MATLAB module. Select the version you want to run and load it, e.g.:
```
module load matlab/2023b
```
On the first load, the MATLAB container will be downloaded (this can take 15–20 minutes, with no progress bar). On subsequent loads module load matlab, the module will start almost instantly.
(Optional) Enable GPU support. If you created an instance with a GPU, you may be able to use it with MATLAB by also loading the GPU module:
```
module load gpu
module load matlab/2023b
```
Note: loading multiple versions of MATLAB modules on a single instance may exhaust disk space due to their significant storage requirements.
Run MATLAB. Once the module is loaded, you can run MATLAB commands as usual. Test MATLAB with a simple command:
```
matlab -nodisplay -r 'disp(magic(5)*magic(5)); exit'
```

Hints and suggestions¶

GUI usage. To run MATLAB with a graphical interface, enable X forwarding when connecting via ssh:
```
ssh -XY ubuntu@<ip-address>
```
GPU usage. If your instance has a GPU and was created from an image with CUDA drivers, load the GPU module together with MATLAB for GPU-accelerated computations:
```
module load gpu matlab
```
Error handling. If you see the error Unrecognized function or variable 'Settings', open MATLAB and run settings. This initializes MATLAB settings properly.

Run Gaussian¶

You can run Gaussian on ScienceCloud using containerized modules. Follow the steps below to create an instance, update your system, and load Gaussian.

Create an instance. Use one of the latest images that includes Singularity. Select an image with the *** prefix in its name, e.g., ***Singularity on Ubuntu 24.04 (2024-11-21).

Update and reboot the system

sudo apt update
sudo apt full-upgrade
sudo reboot

List available modules and package versions
```
module av
```
Load Gaussian module. Select the version you want to run and load it, e.g.:
```
module load gaussian/16_c.01
```
On the first load, the Gaussian container will be downloaded (this can take a few minutes, with no progress bar). On subsequent loads, the module will start almost instantly.
Run Gaussian. Once the module is loaded, you can run Gaussian commands as usual. Test Gaussian with a simple command:
```
mkdir test && cd test
cat << EOF > water.com
%Chk=water
# RHF/6-31G(d)

water energy

0  1
O
H  1  1.0
H  1  1.0  2  120.0

EOF
g16 < water.com > water.log
```
Check the output in water.log for errors. In particular, the last line should begin with Normal termination of Gaussian.

Note: you may encounter the following error when running Gaussian inside a Singularity container:

source: open /opt/gaussian/g16/bsd/g16.profile: permission denied
FATAL:   "g16": executable file not found in $PATH

To resolve this issue, try disabling setuid in the Singularity configuration file and then run your Gaussian script again:

sudo singularity config global -s "allow setuid" "no"

Common issues with `docker compose`¶

Sometimes the entropy value on the base operating system can be too low. This setting can cause docker compose to hang for a few minutes. You can check this value by running: cat /proc/sys/kernel/random/entropy_avail

If the value is less than 1000, then this is potentially the issue. To help solve the problem, consider installing the haveged package using:

sudo apt update
sudo apt install haveged

Then reboot your VM in the ScienceCloud dashboard using the Soft Reboot Instance option.