Scientific Simulations

Scientific simulations in real notebooks.

Vicena gives the agent a persistent cloud JupyterLab for scientific Python. Ask for modeling, simulations, calculations, plots, fitting, or computational chemistry, and the agent builds the notebook with visible code and outputs.

JupyterLab

Python + SciPy

Persistent state

kinetics_model.ipynb
Vicena Scientific Simulations JupyterLab notebook interface

Agentic JupyterLab

Vicena creates notebooks, writes cells, runs them, reads output, fixes errors, and keeps the notebook visible so you can inspect every step.

Scientific Python stack

Use Python, NumPy, SciPy, pandas, matplotlib, RDKit, PySCF, ASE, and installable packages inside the same persistent environment.

Persistent simulation state

Notebook files, generated figures, local datasets, and installed packages stay with the Science Computer feature so later chats can continue from prior work.

Explicit execution boundary

Scientific Simulations runs in the Science Computer feature. Workspace Files are durable research records, so the agent should keep the distinction clear.

What it is for

A simulation workspace for doing science, not just talking about it.

Use it for work that needs execution: notebooks, code, calculations, simulation, modeling, plots, and reproducible analysis. The agent can reason in chat, then move into Jupyter when the work needs code.

Model a system

Create a notebook for ODEs, reaction kinetics, transport, growth curves, population dynamics, or parameterized scientific models.

Run calculations

Do dimensional analysis, numerical integration, uncertainty propagation, fitting, optimization, statistics, and sanity checks with visible code.

Simulate and compare

Sweep parameters, generate plots, compare scenarios, inspect residuals, and iterate until the result matches the scientific question.

Notebook evidence

The notebook stays visible while the science runs.

Notebook cells, plots, figures, and generated artifacts stay in the foreground so the work can be inspected, extended, and challenged instead of disappearing into a chat summary.

CFD wake visualization

CFD wake visualization

Flow simulations and vortex/wake fields can be generated as notebook artifacts, then inspected with the model code beside them.

Graphene band structure

Graphene band structure

Interactive 3D electronic-structure plots can be generated from notebook code and inspected alongside the assumptions.

Molecular orbital visualization

Molecular orbital visualization

Quantum-chemistry outputs such as HOMO/LUMO orbitals, energies, and generated figures stay with the notebook.

Agent workflow

From question to executed notebook.

The important part is not that Vicena can write code. It can keep the code, the result, the error trace, the figure, and the scientific explanation in one reviewable notebook.

1

Plan

Define the scientific question, assumptions, inputs, and expected outputs.

2

Write

Create notebook cells with code, formulas, markdown notes, and plotting logic.

3

Run

Execute cells, inspect errors and outputs, then update the notebook in place.

4

Review

Summarize results, limitations, next experiments, and artifacts to preserve.

Example requests

Start with a scientific job, not a blank notebook.

Each request opens Vicena with a concrete notebook task. The agent should create the notebook, execute cells, and report what it learned from the outputs.