Note
Go to the end to download the full example code.
Modified blackbody: T_dust × β grid¶
Casey 2012 modified blackbody dust SEDs across the canonical fitter’s
two knobs — dust temperature T_dust and emissivity index β.
Each curve in the top panel is a fixed β = 1.8 MBB swept in T;
the bottom panel fixes T = 30 K and sweeps β. The peak shifts by
~40 μm per 10 K of warming; the sub-mm slope steepens by one
power-law index per Δβ = 1.
Useful when interpreting FIR fits as the (T, β) degeneracy
projected onto a single sub-mm photometric point.

import os
os.environ["TF_CPP_MIN_LOG_LEVEL"] = "2" # suppress XLA/PjRt C++ INFO+WARNING logs
import warnings
import jax
import jax.numpy as jnp
import matplotlib as mpl
import matplotlib.pyplot as plt
import numpy as np
import tengri
from tengri.plot import setup_style
setup_style()
warnings.filterwarnings("ignore", message=".*BakedInBackend.*")
C_AA_PER_S = 2.998e18
ssp = tengri.load_ssp()
def _build(t_dust=None, beta=None):
dust = {
"law": "power_law",
"type": "two_component",
"all_params": tengri.Fixed(tengri.DEFAULT),
"tau_diff": 0.5,
"tau_bc": 1.0,
}
# ``dust_T``/``dust_beta_ir`` are overridden per grid point below (#2296:
# a params-dict key the spec declared Fixed is refused), so both must be
# free; bounds pad the T_grid (20-60 K) / beta_grid (1.0-2.5) sweeps.
dust_emission = {
"type": "modified_blackbody",
"all_params": tengri.Fixed(tengri.DEFAULT),
"T": tengri.Uniform(10.0, 80.0),
"beta_ir": tengri.Uniform(0.5, 3.0),
}
model = tengri.SEDModel.build(
ssp,
sfh={"type": "const", "all_params": tengri.Fixed(tengri.DEFAULT), "log_total_mass": 11.13},
dust_attenuation=dust,
dust_emission=dust_emission,
redshift=tengri.Fixed(0.05),
)
p = dict(model.spec.sample(jax.random.PRNGKey(0)))
if t_dust is not None:
p["dust_T"] = jnp.float64(t_dust)
if beta is not None:
p["dust_beta_ir"] = jnp.float64(beta)
return model, p
# Sanity-check what knobs the MBB exposes
_, p_check = _build()
mbb_keys = sorted(
k
for k in p_check
if k.startswith("dust_")
and ("temp" in k.lower() or "beta" in k.lower() or "t_dust" in k.lower())
)
# Will print to log so the reader can re-discover the actual MBB knob names
# if the API ever drifts; not part of the figure.
T_grid = np.linspace(20.0, 60.0, 7)
beta_grid = np.linspace(1.0, 2.5, 6)
fig, (ax_T, ax_b) = plt.subplots(
2, 1, figsize=(7.2, 6.4), sharex=True, gridspec_kw={"hspace": 0.06}
)
norm_T = mpl.colors.Normalize(vmin=T_grid.min(), vmax=T_grid.max())
for T in T_grid:
model, p = _build(t_dust=float(T), beta=1.8)
out = model.predict(p)
w = np.asarray(model.wavelengths)
ax_T.loglog(
w, C_AA_PER_S / w * np.asarray(out.rest_sed()), color=plt.cm.viridis(norm_T(T)), lw=1.4
)
norm_b = mpl.colors.Normalize(vmin=beta_grid.min(), vmax=beta_grid.max())
for beta in beta_grid:
model, p = _build(t_dust=30.0, beta=float(beta))
out = model.predict(p)
w = np.asarray(model.wavelengths)
ax_b.loglog(
w, C_AA_PER_S / w * np.asarray(out.rest_sed()), color=plt.cm.viridis(norm_b(beta)), lw=1.4
)
for ax in (ax_T, ax_b):
# Frame the FIR dust peak (~4.5e44, at 40-120 um across the T grid) with
# headroom. Start at 40 um: redward of that the modified blackbody
# dominates cleanly; blueward the stellar mid-IR tail (~1e42) pokes
# through the cool-dust Wien side and muddies the panel.
ax.set(xlim=(4e5, 1e7), ylim=(1e40, 1e45), ylabel=r"$\nu L_\nu$ [erg s$^{-1}$]")
ax_b.set_xlabel(r"Rest-frame wavelength $\lambda$ [$\mathrm{\AA}$]")
cb_T = fig.colorbar(plt.cm.ScalarMappable(norm=norm_T, cmap="viridis"), ax=ax_T, pad=0.01)
cb_T.set_label(r"$T_{\rm dust}$ [K] (β = 1.8 fixed)")
cb_b = fig.colorbar(plt.cm.ScalarMappable(norm=norm_b, cmap="viridis"), ax=ax_b, pad=0.01)
cb_b.set_label(r"emissivity index $\beta$ ($T_{\rm dust}$ = 30 K fixed)")
plt.savefig("plot_mbb_temperature_beta_grid.png", dpi=150, bbox_inches="tight")
Total running time of the script: (0 minutes 3.514 seconds)