phytorch-lib 0.1.4

PhyTorch is a PyTorch-based modeling toolkit for fitting common plant physiological models of photosynthesis, stomatal conductance, leaf hydraulics, and optical properties.

PhyTorch

A Comprehensive Physiological Plant Modeling Toolkit

PhyTorch is a PyTorch-based package for fitting and analyzing plant physiological models. It provides a unified, simple API for model fitting with efficient GPU-accelerated optimization and automatic differentiation.

Features

• Unified API: Simple fit(model, data) interface for all models
• Photosynthesis Models: FvCB (Farquhar-von Caemmerer-Berry) model with flexible temperature and light response functions
• Stomatal Conductance: Multiple empirical and semi-empirical models (Medlyn, Ball-Woodrow-Berry, Buckley-Mott-Farquhar)
• Leaf Optical Properties: PROSPECT model for leaf spectral reflectance and transmittance
• Leaf Hydraulics: Pressure-volume curves (SJB2018), sigmoidal vulnerability curves
• Canopy Architecture: Leaf angle distribution models with automatic classification into canonical types
• Generic Models: Arrhenius, peaked Arrhenius, Weibull, beta distribution, Gaussian, sigmoidal, and hyperbolic functions
• GPU Acceleration: Leverage PyTorch for fast parameter optimization on CPU or GPU
• Automatic Differentiation: Efficient gradient-based optimization for model fitting
• Custom Plotting: Model-specific diagnostic plots with publication-ready styling

Installation

From PyPI (recommended)

pip install phytorch-lib

From source

git clone https://github.com/ktrizzo/phytorch.git
cd phytorch
pip install -e .

Quick Start

PhyTorch uses a unified fit(model, data) API for all models:

Pressure-Volume Curves

from phytorch import fit
from phytorch.models.hydraulics import SJB2018
import pandas as pd

# Load pressure-volume data
df = pd.read_csv('pv_data.csv')
data = {
    'w': df['RWC'].values,      # Relative water content
    'psi': df['Psi_MPa'].values # Water potential (MPa)
}

# Fit model
model = SJB2018()
result = fit(model, data)

# View results
print(f"πₒ = {result.parameters['pi_o']:.3f} MPa")
print(f"w_tlp = {result.parameters['w_tlp']:.3f}")
print(f"R² = {result.r_squared:.4f}")

# Plot with custom diagnostics
result.plot(save='pv_curve.png')

Leaf Angle Distribution

from phytorch import fit
from phytorch.models.canopy import LeafAngleDistribution, bin_leaf_angles
import pandas as pd

# Load and bin leaf angle measurements
df = pd.read_csv('leaf_angles.csv')
data = bin_leaf_angles(df['zenith'].values, angle_type='zenith')

# Fit model
model = LeafAngleDistribution()
result = fit(model, data)

# Classify canopy architecture
classification = model.classify(result.parameters)
print(f"Canopy type: {classification['type']}")
print(f"R² = {result.r_squared:.4f}")

# Plot with classification
result.plot(save='leaf_angles.png')

Generic Models

from phytorch import fit
from phytorch.models.generic import Arrhenius, Weibull, Beta

# Temperature response with Arrhenius
model = Arrhenius()
data = {'x': temperatures, 'y': rates}
result = fit(model, data)

# Weibull distribution
model = Weibull()
result = fit(model, data)

# Beta distribution
model = Beta()
result = fit(model, data)

Package Structure

phytorch/
├── fit.py               # Unified fit() function
├── models/              # Model library
│   ├── generic/         # Generic mathematical models
│   ├── hydraulics/      # Leaf hydraulics models
│   └── canopy/          # Canopy architecture models
├── photosynthesis/      # FvCB photosynthesis models (legacy API)
├── stomatalconductance/ # Stomatal conductance models (legacy API)
├── leafoptics/          # PROSPECT leaf optical properties (legacy API)
├── leafhydraulics/      # Leaf hydraulics (legacy API)
├── core/                # Core optimization and fitting infrastructure
└── data/                # Example datasets

Documentation

Full documentation is available at https://phytorch.org

Citation

If you use PhyTorch in your research, please cite:

Lei, T., Rizzo, K. T., & Bailey, B. N. (2025). PhoTorch: A robust and generalized
biochemical photosynthesis model fitting package. (In preparation)

Credits

PhyTorch is an extension and reorganization of PhoTorch, developed by:

• Tong Lei
• Kyle T. Rizzo
• Brian N. Bailey

License

PhyTorch is licensed under the GNU General Public License v3.0. See LICENSE for details.

Contributing

Contributions are welcome! Please feel free to submit a Pull Request.

Contact

For questions or issues, please open an issue on GitHub.