Researcher profile

Mi Dai

Mi Dai contributes to research discovery and scholarly infrastructure.

ResearcherAffiliation not importedOpen to collaborate

astro-ph.CO astro-ph.GA astro-ph.HE astro-ph.IM astro-ph.EP astro-ph.SR Machine Learning physics.soc-ph

Trust snapshot

Quick read

Trust 17 - UnverifiedVerification L1Unclaimed author

4works

0followers

8topics

4close collaborators

Actions

Decide how to stay connected

Follow researcher0

Claiming links this public author record to a researcher profile and unlocks direct collaboration workflows.

Direct collaboration

Open a focused conversation when the fit is right

Claim this author entity first to unlock direct invitations.

Inspect adjacent work, topics, institutions and collaborators without jumping out to a separate graph page.

Building this graph slice

BZPEER is loading the nearby papers, people, topics and institutions for this page.

preprint2026arXiv

Feeling Blue: Constructing a Robust SALT3 UV Template and Constraining its Redshift Dependency

Upcoming cosmological surveys will obtain numerous rest-frame ultraviolet (UV) observations of Type Ia supernovae (SNe Ia), yet there is concern about how standardizable SNe Ia are in the UV. In this work, we train a robust optical--UV SED model for SNe Ia (SALT3-UV) with the open-source model-training software $\texttt{SALTshaker}$. We incorporate a spectroscopic UV data sample from HST, including 67 UV spectra from 18 nearby SNe Ia. Unlike previous training spectra, the HST spectra have sufficiently precise calibration that they do not require additional warping to match coincident photometric data. Additionally, while including this new SN Ia sample necessitates incorporating auxiliary photometric data from ZTF and ATLAS that have insufficient calibration for cosmological analyses, the improvements in the calibration of these data is anticipated in the near future. Compared to the previous SALT3-K21 model, the SALT3-UV model shows a significant improvement in the UV down to $2000\mathring{\text{A}}$, with over a threefold improvement in model uncertainty and a more physically accurate continuum and line features. We further evaluate potential redshift evolution in the UV template by separating the UV training sample into low- and high-$z$ subsamples. Our results reveal a non-negligible $\gtrsim 0.05$ mag difference between low- and high-$z$ SALT3-UV models in the $g-$band at $z\gtrsim0.5$ and the $u-$band at $z\gtrsim0.2$. We demonstrate that, if confirmed, such evolution could lead to a few-percent bias in the measurement of $w$ if high-$z$ rest-frame UV data are included in future cosmological surveys such as LSST and $\textit{Roman}$.

preprint2026arXiv

Hyrax: An Extensible Framework for Rapid ML Experimentation and Unsupervised Discovery in the Era of Rubin, Roman, and Euclid

The NSF-DOE Vera C. Rubin Observatory, Roman Space Telescope, Euclid, and other next-generation surveys will deliver imaging, spectroscopic, and time-domain data at scales that increasingly shift the bottleneck in astronomical machine learning (ML) projects from model design to infrastructure. We present Hyrax, an open-source, modular, GPU-enabled Python framework that supports the full ML lifecycle in astronomy: from data acquisition and training to inference and experiment comparison, with capabilities including multimodal dataset support, integrated vector databases for similarity search, and interactive two- and three-dimensional latent-space exploration for unsupervised discovery. We demonstrate Hyrax's versatility through five representative applications on real survey data: (i) unsupervised representation learning on $\sim 4\times10^5$ Rubin Legacy Survey of Space and Time (LSST) Data Preview 1 (DP1) galaxies, surfacing new merger and low-surface-brightness candidates missing from reference Euclid and Dark Energy Survey catalogs, while also isolating imaging artifacts -- all without labeled training data; (ii) hybrid density-based clustering for identifying cluster-scale gravitational lens candidates in DP1 data; (iii) multimodal early-time transient classification in the Zwicky Transient Facility leveraging light curves, spectra, images, and metadata; (iv) supervised false-positive filtering in shift-and-stack searches for distant solar system objects in the Dark Energy Camera Ecliptic Exploration Project survey; and (v) supervised detection of semi-resolved dwarf galaxies in Hyper Suprime-Cam and LSST-like imaging using synthetic source injection. Together, these results demonstrate that Hyrax provides astronomy-specific ML infrastructure that enables systematic discovery and rapid methodological iteration across next-generation astronomical surveys.

preprint2025arXiv

Picture an Astronomer: Best Practices for Retaining Talent in Astrophysics

Women are consistently underrepresented in astrophysics yet are simultaneously subject to disproportionate attrition at every career stage. This disparity between demonstrated efficacy in job performance and ultimate career outcome was the primary motivation for the Picture an Astronomer series, which included both targeted public outreach to increase representation of women in astrophysics and high-level, solution-oriented discussions among professional astronomers. In March 2025, more than 200 astronomers came together in a hybrid-format symposium focused on the state of the field for female scientists, combining scientific exchange with discussions of policies and practices to strengthen retention of talent in the field. This white paper is the result of those discussions, offering a wide range of recommendations developed in the context of gendered attrition in astrophysics but which ultimately support a healthier climate for all scientists alike.

preprint2021arXiv

The Foundation Supernova Survey: Photospheric Velocity Correlations in Type Ia Supernovae

The ejecta velocities of type-Ia supernovae (SNe Ia), as measured by the Si II $λ6355$ line, have been shown to correlate with other supernova properties, including color and standardized luminosity. We investigate these results using the Foundation Supernova Survey, with a spectroscopic data release presented here, and photometry analyzed with the SALT2 light-curve fitter. We find that the Foundation data do not show significant evidence for an offset in color between SNe Ia with high and normal photospheric velocities, with $Δc = 0.005 \pm 0.014$. Our SALT2 analysis does show evidence for redder high-velocity SN Ia in other samples, including objects from the Carnegie Supernova Project, with a combined sample yielding $Δc = 0.017 \pm 0.007$. When split on velocity, the Foundation SN Ia also do not show a significant difference in Hubble diagram residual, $ΔHR = 0.015 \pm 0.049$ mag. Intriguingly, we find that SN Ia ejecta velocity information may be gleaned from photometry, particularly in redder optical bands. For high-redshift SN Ia, these rest-frame red wavelengths will be observed by the Nancy Grace Roman Space Telescope. Our results also confirm previous work that SN Ia host-galaxy stellar mass is strongly correlated with ejecta velocity: high-velocity SN Ia are found nearly exclusively in high-stellar-mass hosts. However, host-galaxy properties alone do not explain velocity-dependent differences in supernova colors and luminosities across samples. Measuring and understanding the connection between intrinsic explosion properties and supernova environments, across cosmic time, will be important for precision cosmology with SNe Ia.

Institution

Affiliation not imported yet

This author record came from a source that does not expose affiliation metadata. Once the author claims the profile or we enrich the record from another provider, this section will link to the concrete institution.

Topic footprint