SED_Model

Synthetic photometry and stellar parameter inference.

SED_Model is a Python package for computing observer-ready synthetic spectral energy distributions (SEDs) and broadband magnitudes from stellar atmosphere grids, and for recovering stellar parameters from observed photometry via Bayesian MCMC inference. It is designed to work directly downstream of SED_Tools and as a standalone drop-in for workflows that use MESA-style atmosphere grids and filter transmission curves.

Version: 0.4.0

Documentation: https://nialljmiller.github.io/SED_Model/

---

Overview

SED_Model provides two paired models that share the same parameter language:

• Forward model — maps (Teff, logg, [M/H], R, d) to an interpolated SED, bolometric flux and magnitude, and synthetic magnitudes in any set of loaded filters.
• Inverse model — maps observed magnitudes with uncertainties to a posterior distribution over (Teff, logg, [M/H]), and optionally over Av and distance, via emcee MCMC sampling.

Both directions operate through a shared FitParams object. The same parameter specification flows from the forward model to the likelihood evaluator inside the sampler and back out again.

---

Features

• Bounded Hermite (default), Hermite and linear interpolation over 3D atmosphere grids (Teff, logg, [M/H]).
• Photon-counting filter convolution in Vega, AB, and ST photometric systems.
• Precomputed zero-points matching MESA colors/private/synthetic.f90 conventions.
• Pure-NumPy interstellar dust extinction with multiple supported laws.
• Bayesian stellar-parameter inference via emcee with fixed, bounded, and fully free parameter modes.
• Structured result containers with summaries, persistence to CSV and NPZ, and interpolation diagnostics.
• Fortran kernels for performance-critical paths, compiled via f2py/Meson and shipped as prebuilt wheels.
• Forward tests validated against MESA colors reference outputs; inverse tests verified by self-consistent synthetic recovery.

---

Installation

pip install sed-model

Prebuilt wheels (Fortran extension included, runtime bundled) are published for Linux x86_64/aarch64 and macOS arm64/x86_64. Every release also provides a pure-Python wheel, standalone Fortran libraries (no Python needed), and the source distribution — see the installation guide for choosing between them.

From source

git clone https://github.com/nialljmiller/SED_Model.git
cd SED_Model
python -m pip install -e .      # needs gfortran; meson-python builds the extension

The extension compiles:

fortran/public/colors_def.f90
fortran/private/{colors_utils,hermite_interp,linear_interp,hermite_interp_bounded,synthetic,bolometric}.f90
fortran/cc_api.f90

into sed_model.cc_api. make is a quick alternative that builds it in place.

Compatibility

Platform	pip install sed-model	Python	How it was verified
Linux x86_64 (glibc ≥ 2.28)	prebuilt wheel	3.10 – 3.14	CI on every push; built and tested in manylinux_2_28 (emulated)
Linux aarch64 (glibc ≥ 2.28)	prebuilt wheel	3.10 – 3.14	CI on every push; all five Pythons tested on Debian 11/12/13
macOS arm64 (macOS 14+)	prebuilt wheel	3.10 – 3.14	CI on every push; build and repair path tested locally
macOS x86_64 (macOS 15+)	prebuilt wheel	3.10 – 3.14	CI on every push; not tested locally
Linux musl (Alpine)	builds from source	3.10+	tested; needs gcc gfortran musl-dev
macOS older than the above	builds from source	3.10+	needs gfortran
Windows	not supported natively	–	use WSL

The Fortran runtime is bundled in the Linux wheels and linked statically into the macOS wheels, so nothing besides Python and NumPy ≥ 2.0 is needed to use a wheel. "Builds from source" means pip compiles the Fortran on your machine and requires gfortran and a C compiler; if either is missing the build stops with instructions.

Each release also provides, on the GitHub releases page, a pure-Python wheel (no compiled extension) and standalone Fortran libraries (no Python) for the Linux and macOS platforms above. See the installation guide.

Requirements

• Python >= 3.10
• numpy >= 2.0
• pandas
• emcee
• gfortran (building from source only)

matplotlib is only needed for plotting/demo workflows.

---

Quick Start

from sed_model import load_grid, load_filters, run_forward, run_inverse

grid = load_grid("/path/to/Kurucz2003all/")
filters = load_filters([
    "/path/to/filters/GAIA/G.dat",
    "/path/to/filters/2MASS/J.dat",
])

# Forward model
result = run_forward(
    teff=5777,
    logg=4.44,
    meta=0.0,
    R=6.957e10,
    d=3.086e19,
    grid=grid,
    filters=filters,
    mag_system="Vega",
)

print(result.magnitudes)

# Inverse model
posterior = run_inverse(
    obs_magnitudes=[5.03, 4.17],
    obs_uncertainties=[0.01, 0.02],
    filter_names=["G", "J"],
    R=6.957e10,
    d=3.086e19,
    grid=grid,
    filters=filters,
    mag_system="Vega",
)

posterior.print_summary()

---

Grid Loading

from sed_model import load_grid

grid = load_grid("/path/to/Kurucz2003all/")

load_grid() expects a MESA/SED_Tools-style atmosphere grid containing a flux_cube.bin, with optional validation against a co-located lookup_table.csv.

The returned AtmosphereGrid exposes:

• teff_grid, logg_grid, meta_grid
• teff_bounds, logg_bounds, meta_bounds
• wavelengths
• in_bounds(teff, logg, meta)
• clamp(teff, logg, meta)
• nearest_grid_distance(teff, logg, meta)

---

Filter Loading

from sed_model import load_filters, load_filters_from_instrument_dir

filters = load_filters([
    "/path/to/filters/Generic/Johnson/B.dat",
    "/path/to/filters/Generic/Johnson/V.dat",
])

filters = load_filters_from_instrument_dir(
    "/path/to/filters/Generic/Johnson"
)

Each Filter stores:

• name
• wavelengths
• transmission
• Vega, AB, and ST zero-points
• zero_point(system)

Supported systems are:

Vega
AB
ST

---

Forward Model

from sed_model import run_forward

result = run_forward(
    teff=5778.0,
    logg=4.44,
    meta=0.0,
    R=6.957e10,
    d=3.086e19,
    grid=grid,
    filters=filters,
    mag_system="Vega",
    interp_method="hermite_bounded",  # or "hermite", "linear"
)

ForwardResult contains:

• wavelengths
• surface_flux
• observed_flux
• magnitudes
• band_fluxes
• bol_flux
• bol_mag
• nearest_grid_distance
• clamped
• teff, logg, meta, R, d, a_v
• mag_system
• extinction_applied

---

Inverse Model

from sed_model import run_inverse

result = run_inverse(
    obs_magnitudes=[5.03, 4.17],
    obs_uncertainties=[0.01, 0.02],
    filter_names=["G", "J"],
    R=6.957e10,
    d=3.086e19,
    grid=grid,
    filters=filters,
    n_walkers=32,
    n_steps=1000,
    n_burn=300,
    seed=42,
)

InverseResult contains:

• samples
• log_prob
• filter_names
• obs_magnitudes
• obs_uncertainties
• R, d, mag_system
• n_walkers, n_steps, n_burn, n_thin
• acceptance_fraction
• autocorr_time

It also provides:

result.summary()
result.print_summary()
result.save("posterior.npz")

---

Parameter Modes

FitParams and ParamSpec define the shared parameter language used by the forward and inverse models.

from sed_model import fit_params_from_grid, fixed, free, bounded

params = fit_params_from_grid(
    grid,
    a_v=(0.0, 2.0),   # free Av
    d_cm=3.086e19,    # fixed distance
)

Available modes:

fixed(value)
free(lo, hi)
bounded(lo, hi)

Example:

from sed_model import FitParams, fixed, free

params = FitParams(
    teff=free(5000.0, 6500.0),
    logg=fixed(4.44),
    meta=free(-1.0, 0.5),
    a_v=fixed(0.0),
    d=fixed(3.086e19),
)

---

Extinction

from sed_model import make_extinction_model, run_forward

ext = make_extinction_model(
    enabled=True,
    law="fitzpatrick99",
    a_v=0.3,
    r_v=3.1,
)

result = run_forward(
    teff=5778,
    logg=4.44,
    meta=0.0,
    R=6.957e10,
    d=3.086e19,
    grid=grid,
    filters=filters,
    extinction=ext,
)

Supported extinction-law keys include:

ccm89
odonnell94
fitzpatrick99
fm07
calzetti00
gordon23

---

Output and I/O

from sed_model import save_sed, save_magnitudes

save_sed(result, "forward_sed.csv")
save_magnitudes(result, "forward_magnitudes.csv")

Inverse results can be saved directly:

posterior.save("posterior_samples.npz")

---

Relationship to SED_Tools

SED_Model sits directly downstream of SED_Tools.

SED_Tools handles acquisition, standardization, filter downloads, grid construction, and SED completion/generation. SED_Model uses the resulting atmosphere grids and filter files for forward synthetic photometry and Bayesian stellar-parameter inference.

A typical workflow is:

SED_Tools  ->  build/download SED grids and filters
SED_Model  ->  generate synthetic magnitudes and infer stellar parameters

---

Documentation and Demos

Full documentation, including walkthroughs of every demo script in demos/, lives at:

https://nialljmiller.github.io/SED_Model/

The site is built with MkDocs from the docs/ directory and embeds the demo sources directly, so the documented examples are always identical to the code in this repository.

---

Testing

pytest tests/test_forward.py
pytest tests/test_inverse.py

The forward tests verify grid loading, bounds checking, interpolation, filter loading, zero-points, bolometric quantities, and synthetic magnitudes.

The inverse tests verify self-consistent posterior recovery and InverseResult persistence.

---

Repository Structure

sed_model/
  __init__.py
  grid.py
  filters.py
  forward.py
  inverse.py
  params.py
  io.py
  sed_extinction.py

fortran/
  public/colors_def.f90
  private/{colors_utils,hermite_interp,linear_interp,
           hermite_interp_bounded,synthetic,bolometric}.f90
  cc_api.f90

demos/
tests/
Makefile
meson.build
pyproject.toml
meson.build
meson_options.txt
tools/

---

License

SED_Model uses a mixed-license structure. The Python source code, documentation, examples, tests, and packaging files are distributed under the MIT License. The Fortran numerical kernels in fortran/ are distributed under LGPL-3.0-or-later. See LICENSE and source file headers for details.

---