SageMath on Hopper

Installing Sage with Conda Environments

To install Sage in your /home directory, the fastest way is to use Conda environments as in the steps below:

Install the latest miniconda:

wget https://repo.anaconda.com/miniconda/Miniconda3-latest-Linux-x86_64.sh -O miniconda3.sh
bash miniconda3.sh -u -b -p $PWD/miniconda3

2. Activate the installed miniconda to use conda and install mamba:

source miniconda/bin/activate

conda install mamba -c conda-forge

With the installed mamba, create a conda environment for sage
```
mamba create -n sage sage -c conda-forge
```
After the sage environment is created, activate it and add the kernel so you can use it in Jupyter Lab from Open OnDemand
```
conda activate sage

conda install ipykernel
python -m ipykernel install --user --name=SageMath
```

From the Command Line Interface

To run Sage from the command line:

Log on to Hopper from the command line

 ssh userid@hopper.orc.gmu.edu

2. Connect to a compute node with the salloc command

 salloc -p normal  -q normal --nodes=1 --cpus-per-task=12 --mem=15GB

Activate the Sage environment with the lines:

 source miniconda3/bin/activate
 conda activate sage

Run it with

```

sage

┌────────────────────────────────────────────────────────────────────┐ │ SageMath version 10.2, Release Date: 2023-12-03 │ │ Using Python 3.11.6. Type "help()" for help. │ └────────────────────────────────────────────────────────────────────┘

sage: 1+2 3

sage: exit

```

Once you exit sage, you can deactivate the environments, first the sage environment:

conda deactivate

then the main environment:

conda deactivate

And exit the compute node with:

exit

You should also be able to run sage scripts with SLURM in batch mode:

Example script, test.sage:

var('x')
assume(x > 0)
print (bool(sqrt(x^2) == x))

Example SLURM script, run.slurm:

#!/bin/bash
#SBATCH --job-name=sage_job
#SBATCH --qos=normal
#SBATCH --partition=normal

#SBATCH --output=sage_job-%j.out
#SBATCH --error=sage_job-%j.err

#SBATCH --nodes=1
#SBATCH --cpus-per-task=12

#SBATCH --mem=10G
#SBATCH --time=0-00:30:00

#Load needed modules/activate environment
module load gnu10 openmpi
source miniconda/bin/activate
conda activate sage


#Execute
sage test.sage

Submit it with:

sbatch run.slurm

and it will run in batch mode on the requested resources in the SLURM script. The outputs for this test script will be in the .out file and any error messages will be in the .err file.