fridapy 0.1.0

Integrated Assessment Model, based on FRIDA, written in python using sostrades

fridapy

A coupled climate–economy–society system-dynamics model, based on the WorldTrans FRIDA v2.1 integrated assesment model and re-implemented in Python as a set of SoSTrades disciplines solved by an MDA (multidisciplinary analysis) coupling. This version replicates the original FRIDA model v2.1 results, using a simplified (tuned) ocean carbon model.

Installation

fridapy depends on sostrades-core, which is not published on PyPI, so it must be installed first from GitHub (this is not our package — see os-climate/sostrades-core):

pip install git+https://github.com/os-climate/sostrades-core.git@v5.3.4
pip install fridapy            # or, from a checkout:  pip install -e .

Equivalently, install the pinned runtime in one step:

pip install -r requirements.txt
pip install -e .

For the diagnostic / ensemble / plotting scripts under diagnostics/, install the extra:

pip install -e ".[diagnostics]"     # adds matplotlib, scipy

Quickstart

Run the full model for a single climate case:

import numpy as np
from sostrades_core.execution_engine.execution_engine import ExecutionEngine

DISCIPLINES = [
    'Demographics', 'Economy', 'Energy', 'Emissions', 'Climate', 'LandUse',
    'GovernmentRegulations', 'BehavioralChoices', 'BehavioralChange',
    'Resources', 'SeaLevel', 'SeaLevelRiseImpacts', 'PlanetaryBoundaries',
]
time = np.arange(1980, 2150 + 1e-9, 1.0)

study = 'fridapy'
ee = ExecutionEngine(study)
ee.select_root_process(repo='fridapy', mod_id='process')
ee.configure()

values = {f'{study}.{study}.{d}.time': time for d in DISCIPLINES}
values[f'{study}.{study}.Climate.climate_case'] = 50          # 1–100
ee.load_study_from_input_dict(values)
ee.dm.set_values_from_dict({
    f'{study}.{study}.max_mda_iter': 300,
    f'{study}.{study}.tolerance': 1e-3,
    f'{study}.{study}.inner_mda_name': 'MDAGaussSeidel',
})
ee.execute()

gdp = ee.dm.get_value(f'{study}.Real_GDP')

See diagnostics/usecase.py for a complete 9-panel example, and diagnostics/run_ensemble.py for ensemble runs (local or Modal).

Repository structure

Path	Contents
fridapy/disciplines/	the 13 SoSTrades disciplines
fridapy/process/	the MDA process builder
fridapy/data/	packaged runtime reference data (forcing series, climate cases)
diagnostics/	ensemble runners, calibration, validation, plotting
docs/	design notes for each major mechanism
data/	full development dataset (Stella FRIDA.isdb source, calibration inputs)

Data sources

The packaged fridapy/data/ files (frida_input_data.csv, climate_cases.json) are derived from the FRIDA v2.1 model. The FRIDA reference uncertainty envelopes used by some diagnostics come from Schoenberg, W. (2025), FRIDA v2.1 Endogenous Model Behavior (EMB) 100000 member ensemble, Zenodo, 10.5281/zenodo.15396799 (CC-BY-4.0); that ~270 MB archive is not bundled — see diagnostics/analyze_ensemble.py for how to obtain it.