ml_ai_augmentation
GPU-first ML workflows for symbolic regression, PDE surrogates, pattern recognition, and derivation explanations.
How to use it
ml_ai_augmentation is exposed by the Phys MCP MCP server. Add the server to your MCP client (Claude Desktop, Cursor, Windsurf and others), and the ml_ai_augmentation tool becomes available to the model automatically. See the full listing for setup details and every tool this server provides.
Install Phys MCP
pip install -r requirements.txtOther tools in Phys MCP (16)
Report available acceleration hardware and the active ACCELMODE/ACCELDEVICE.
Access external scientific APIs such as arXiv, CERN Open Data, NASA datasets, and NIST references.
Computer Algebra System operations for evaluating expressions, differentiation, integration, solving equations and ODEs, and propagating uncertainty.
Retrieve CODATA and astrophysical constants including c, h, G, Msun, pc, ly, and more.
Unified data toolkit for HDF5/FITS/ROOT I/O plus GPU-first FFT, filtering, spectrogram, and wavelet analysis via the action parameter.
Distributed job submission, session sharing, lab notebooks, and artifact versioning.
DAG-driven orchestration with validation, execution, publishing, and collaboration hooks.
Publish research artifacts to Overleaf, GitHub, Zenodo, Jupyter, and immersive formats.
Full-featured calculator with CAS, graphing, statistics, matrices, and programmable utilities.
Translate natural language physics requests into structured MCP tool calls.
Generate 2D/3D plots, vector fields, phase portraits, contours, volume plots, animations, and interactive visualizations.
Quantum computing utilities for operators, solvers, and Bloch/probability visualizations (scaffold).
Summarize MCP sessions into Markdown reports with linked artifacts.
Build partition functions and derived thermodynamic quantities from energy levels.
Compute Christoffel symbols, curvature tensors, and geodesics (scaffold).
Convert between units via the Pint registry with SI, imperial, and specialized physics unit coverage.