Using Galaxy Evolution as Source of Physics-Based Ground Truth for Generative Models
Authors:
Yun Qi Li,
Tuan Do,
Evan Jones,
Bernie Boscoe,
Kevin Alfaro,
Zooey Nguyen
Abstract:
Generative models producing images have enormous potential to advance discoveries across scientific fields and require metrics capable of quantifying the high dimensional output. We propose that astrophysics data, such as galaxy images, can test generative models with additional physics-motivated ground truths in addition to human judgment. For example, galaxies in the Universe form and change ove…
▽ More
Generative models producing images have enormous potential to advance discoveries across scientific fields and require metrics capable of quantifying the high dimensional output. We propose that astrophysics data, such as galaxy images, can test generative models with additional physics-motivated ground truths in addition to human judgment. For example, galaxies in the Universe form and change over billions of years, following physical laws and relationships that are both easy to characterize and difficult to encode in generative models. We build a conditional denoising diffusion probabilistic model (DDPM) and a conditional variational autoencoder (CVAE) and test their ability to generate realistic galaxies conditioned on their redshifts (galaxy ages). This is one of the first studies to probe these generative models using physically motivated metrics. We find that both models produce comparable realistic galaxies based on human evaluation, but our physics-based metrics are better able to discern the strengths and weaknesses of the generative models. Overall, the DDPM model performs better than the CVAE on the majority of the physics-based metrics. Ultimately, if we can show that generative models can learn the physics of galaxy evolution, they have the potential to unlock new astrophysical discoveries.
△ Less
Submitted 9 July, 2024;
originally announced July 2024.
A systematic TMRT observational study of Galactic $^{12}$C/$^{13}$C ratios from Formaldehyde
Authors:
Y. T. Yan,
J. S. Zhang,
C. Henkel,
T. Mufakharov,
L. W. Jia,
X. D. Tang,
Y. J. Wu,
J. Li,
Z. A. Zeng,
Y. X. Wang,
Y. Q. Li,
J. Huang,
J. M. Jian
Abstract:
We present observations of the C-band $1_{10}-1_{11}$ (4.8 GHz) and Ku-band $2_{11}-2_{12}$ (14.5 GHz) K-doublet lines of H$_2$CO and the C-band $1_{10}-1_{11}$ (4.6 GHz) line of H$_2$$^{13}$CO toward a large sample of Galactic molecular clouds, through the Shanghai Tianma 65-m radio telescope (TMRT). Our sample with 112 sources includes strong H$_2$CO sources from the TMRT molecular line survey a…
▽ More
We present observations of the C-band $1_{10}-1_{11}$ (4.8 GHz) and Ku-band $2_{11}-2_{12}$ (14.5 GHz) K-doublet lines of H$_2$CO and the C-band $1_{10}-1_{11}$ (4.6 GHz) line of H$_2$$^{13}$CO toward a large sample of Galactic molecular clouds, through the Shanghai Tianma 65-m radio telescope (TMRT). Our sample with 112 sources includes strong H$_2$CO sources from the TMRT molecular line survey at C-band and other known H$_2$CO sources. All three lines are detected toward 38 objects (43 radial velocity components) yielding a detection rate of 34\%. Complementary observations of their continuum emission at both C- and Ku-bands were performed. Combining spectral line parameters and continuum data, we calculate the column densities, the optical depths and the isotope ratio H$_2$$^{12}$CO/H$_2$$^{13}$CO for each source. To evaluate photon trapping caused by sometimes significant opacities in the main isotopologue's rotational mm-wave lines connecting our measured K-doublets, and to obtain $^{12}$C/$^{13}$C abundance ratios, we used the RADEX non-LTE model accounting for radiative transfer effects. This implied the use of the new collision rates from \citet{Wiesenfeld2013}. Also implementing distance values from trigonometric parallax measurements for our sources, we obtain a linear fit of $^{12}$C/$^{13}$C = (5.08$\pm$1.10)D$_{GC}$ + (11.86$\pm$6.60), with a correlation coefficient of 0.58. D$_{GC}$ refers to Galactocentric distances. Our $^{12}$C/$^{13}$C ratios agree very well with the ones deduced from CN and C$^{18}$O but are lower than those previously reported on the basis of H$_2$CO, tending to suggest that the bulk of the H$_2$CO in our sources was formed on dust grain mantles and not in the gas phase.
△ Less
Submitted 8 April, 2019;
originally announced April 2019.
