Visualisation on duhpc

1. Start Jupyter on a compute node (in your master login shell)

module load viz
srun --partition=short --pty --time=04:00:00 duhpc-jupyter

It prints a block like:

Node : cn01 Port : 9139 Job : 6139
ssh -N -L 9139:cn01:9139 <you>@10.58.5.101
http://localhost:9139/lab?token=….

…followed by Jupyter’s own log, ending with a line containing your token:

http://cn01.hpc:9139/lab?token=0043fc6005dbea4fdf…

Leave this window running.

2. Open the tunnel (a NEW terminal ON YOUR PC)

Open a fresh terminal on your own computer — Windows Terminal, PowerShell, or MobaXterm’s local terminal (a shell on your laptop, not an SSH session to the cluster). Run the ssh -N -L ... line from Step 1, with your username:

ssh -N -L 9139:cn01:9139 you@10.58.5.101

Enter your cluster password. It will then sit silent — correct; leave it open.

3. Open JupyterLab in your browser

Paste into your browser’s address bar, using localhost and the token from Step 1:

http://localhost:9139/lab?token=0043fc6005dbea4fdf…

JupyterLab loads. You will see your /scratch/home/<you> files on the left.

4. Visualise (in a notebook cell in the browser)

Click Notebook → Python 3 (ipykernel). In a cell, type and press Shift+Enter:

import cclib, py3Dmol
from cclib.parser.utils import PeriodicTable
pt = PeriodicTable()

data = cclib.io.ccread('/scratch/home/YOURNAME/gaussian/myjob.log')
coords = data.atomcoords[-1]                       # last/optimised geometry
xyz = f"{data.natom}\n\n" + "\n".join(
    f"{pt.element[z]} {x:.4f} {y:.4f} {zc:.4f}"
    for z, (x, y, zc) in zip(data.atomnos, coords))

v = py3Dmol.view()
v.addModel(xyz, 'xyz')
v.setStyle({'stick': {}, 'sphere': {'scale': 0.3}})
v.zoomTo(); v.show()

The cclib library reads your job output and extracts the final geometry. A rotatable 3D molecule appears in the cell (drag to rotate, scroll to zoom).

5. Finish

Close the browser tab → press Ctrl-C in the Jupyter window (Step 1) → close the tunnel terminal (Step 2). This releases your compute allocation.