Kaili Cao 曹开力

About · Publications

Publications

A complete list: first-author papers (including those I wrote alone), then coauthored papers by topic, each list newest first. Entries give my position among the authors, links to the code where it is public, and, for each coauthored paper, what I contributed. The same list is on NASA ADS (by ORCID).

Papers per year (chart)
0 5 10 2021: 1 first or sole author 2021: 1 coauthored 2 2021 2022 2023 2024: 4 coauthored 4 2024 2025: 3 first or sole author 2025: 5 coauthored 8 2025 2026: 6 first or sole author 2026: 4 coauthored 10 2026
  • first or sole author
  • coauthor
Papers per year
YearFirst or sole authorCoauthor
202111
202200
202300
202404
202535
202664
My papers by year, from the list below: 24 as of October 2026, with preprints counted in 2026. Made with AI assistance (Claude).

24 papers as of October 2026: 10 first-author (3 of them sole-author) and 14 coauthored; 20 are published in journals and 4 are preprints.

Highlights

Five papers in a sentence each. The full list, with my contribution to every paper, follows.

Four image panels: simulated sky images (top) and injected stars (bottom), coadded by PyIMCOM (left) and reconstructed by Effortless (right).

First two rows and columns of Figure 1 of Cao 2026a.

A faster way to make sharp images from Roman

Effortless turns the Roman Space Telescope's undersampled exposures into well-sampled images with a controlled point spread function, 50 to 60 times faster than IMCOM and with less memory.

Six image panels: simulated Roman sky images (top row) and injected stars (bottom row), each coadded by PyIMCOM's three linear algebra kernels.

Top half of Figure 5 of Cao et al. 2025, CC BY 4.0.

Making IMCOM practical

I led the rewrite of IMCOM into PyIMCOM, which cut its computing cost by about an order of magnitude; it produced the IMCOM images of the OpenUniverse2024 simulations.

Two panels of constraints on sigma 8 and Omega m, in Fourier space and real space: Fisher-forecast ellipses compared with MCMC samples.

Figure 6 of Cao et al. 2026, CC BY 4.0.

What Roman can tell us about the Universe

Forecasts of how well Roman's High Latitude Imaging Survey can measure cosmology by combining weak lensing and galaxy clustering; outside knowledge of the shape of the matter power spectrum improves the results by factors of 1.2 to 3.5.

Two panels of the predicted change in surface [C/N] against surface gravity along the red giant branch, for several metallicities and masses, marking the end of the first dredge-up and the center of the red giant branch bump.

Figure 2 of Cao and Pinsonneault 2026, CC BY 4.0.

Testing stellar models with thousands of red giants

Comparing stellar models with red giants from the APOKASC-3 catalog shows that the observed red giant branch bump sits about 0.15 dex higher in surface gravity than the models predict.

Four panels of asteroseismic mass and age against the frequency of maximum power for three combinations of models and weighting schemes: with the APOKASC-3 correction (top) the three disagree at low frequency, and with the raw correction (bottom) they agree.

Middle two rows of Figure 2 of Cao and Pinsonneault 2025, with its axis labels, CC BY 4.0.

More reliable masses for luminous red giants

A correction to the asteroseismic masses of luminous red giants that makes them much less sensitive to how a star's mean density is inferred from its oscillation frequencies.

First-author papers

Coauthored papers

Images

Cosmology

Stars

Dissertation