Features

• Automatic construction of CALPHAD models from MLIP-based thermodynamic data
• Integration with ATAT and PyCalphad for Gibbs energy and phase diagram calculations
• Efficient parameter fitting using the Jansson derivative method with gradient-based optimization
• Support for physically-grounded adjustments using experimental phase equilibria
• Compatibility with ESPEI and PyCalphad toolchain for advanced thermodynamic modeling

Relationship to PhaseForge

PhaseForgePlus builds upon the foundation of PhaseForge, extending its capabilities for advanced CALPHAD model optimization. While PhaseForge enables users to automatically generate initial thermodynamic database (TDB) files from MLIPs, PhaseForgePlus takes the workflow a step further by providing robust tools for parameter fitting, integration with experimental data, and gradient-based optimization.

You can use PhaseForge to create an initial .tdb file, which serves as the starting point for further optimization in PhaseForgePlus.

By combining PhaseForge and PhaseForgePlus, you can seamlessly transition from automated TDB generation to advanced model optimization in a single, reproducible workflow.

Installation

You can install PhaseForgePlus via pip (once available on PyPI):

pip install phaseforgeplus

Or clone the repository directly:

pip install git+https://github.com/dogusariturk/PhaseForgePlus.git

Quick Start

Here's a minimal example of optimizing a CALPHAD model using PhaseForgePlus:

from phaseforgeplus import PhaseForgePlus

pfp = PhaseForgePlus(
    db="./data/pt-w.tdb",  # Path to your thermodynamic database
    zpf_path="./data",  # Path to Zero Phase Fraction (ZPF) data
    points=[1801, 1601, 1401, 1201, 1001, 802, 602, 402, 202],  # Points for optimization
    pressure=101325,  # Pressure in Pa
    temperature=298.15,  # Temperature in K
)

pfp.optimize()