Research

Cosmology - gravitational lensing - synergies between strong and weak lensing - Hubble tension - dark enegy - quasar Hubble diagram. Rubin Observatory

Publication list on Inspire HEP

Science questions

I use astrophysical observations to address fundamental questions about the Universe:

How can independent cosmological probes improve measurements of the Hubble constant and the expansion history of the Universe?
How can strong and weak gravitational lensing be combined to reduce systematic uncertainties in time-delay cosmography?
What can gravitational lensing reveal about the distribution of dark matter in galaxies and clusters?
How can quasars and gamma-ray bursts extend cosmological distance measurements to the high-redshift Universe?

Methodology

Strong gravitational lensing and time-delay cosmography
Galaxy-galaxy weak gravitational lensing
Population simulations and forward modeling
Statistical inference and cosmological parameter estimation
Machine learning for strong-lens discovery and classification
Scientific software and analysis-pipeline development
Quasars and gamma-ray bursts as high-redshift cosmological probes

Below are the main research topics I work on, spanning strong-lensing cosmography, Rubin Observatory science, and high-redshift cosmology.

Figure 1. Reduced shear profile of a lens at z = 0.5, illustrating the strong- and weak-lensing regimes. The light-blue shaded region on the right represents the weak-lensing regime, where galaxy shape measurements are generally available. The vertical red dashed line indicates the lower radial limit typically adopted for weak-lensing measurements in large-scale clustering analyses. The light-green shaded region on the left represents the strong-lensing regime. The orange dashed line marks the Einstein radius, while the brown dashed line indicates twice the Einstein radius, which generally defines the outer limit of strong-lensing observations. The light-lavender shaded region represents the intermediate-lensing regime. The vertical magenta dashed line indicates the radial scale to which weak-lensing measurements must extend to achieve approximately 1% precision on the mass-sheet degeneracy (MSD) constraint. The yellow shaded region highlights the additional radial coverage required from weak-lensing observations to reach this precision. M_halo and M_* in the light-blue shaded region denote the halo mass and stellar mass, respectively.

Research topics

Strong-lensing cosmography and the mass-sheet degeneracy

Strong gravitational lensing offers a way to measure cosmological distances that is independent of several commonly used cosmological probes. In particular, time delays between multiple images of strongly lensed quasars and supernovae can be used to infer the time-delay distance and constrain the Hubble constant (H₀). I am interested in developing this technique into a precise and robust probe of the expansion history of the Universe.

A central limitation is the mass-sheet degeneracy (MSD): different lens mass distributions can reproduce the same strong-lensing observables while implying different time-delay distances. I investigate how galaxy-galaxy weak lensing can provide complementary information about the mass distribution beyond the Einstein radius and help constrain this degeneracy (See Fig. 1 for an illustration of the strong and weak lensing regimes). One part of this work is to identify non-lens galaxies that match the properties and environments of strong-lens galaxies. Their stacked weak-lensing profiles can then be used to infer the population-level mass distribution and inform strong-lens mass models.

Figure 2. Forecast for the Hubble constant measurement precision achievable by combining strong and weak lensing measurements from the LSST survey.

This approach connects strong-lensing measurements on small physical scales with weak-lensing information at large scales. It offers a route to controlling an important systematic in time-delay cosmography and improving the reliability of cosmological constraints (the detailed methodology is described in Khadka et al. (2024) ). We hope to achieve percent level constrain on Hubble constant combining strong and weak lensing measurements from LSST survey (see Fig. 2 for the forecast).

Preparing strong-lensing science for the Rubin Observatory

The Vera C. Rubin Observatory's Legacy Survey of Space and Time (LSST) will repeatedly image a large fraction of the southern sky, creating an exceptional data set for discovering variable and multiply imaged sources. The expected discovery of large samples of lensed quasars and supernovae will open new opportunities for time-delay cosmography, studies of dark matter, and the characterization of lens-galaxy populations.

First SLSim simulation plot — **Figure 3.** Illustration of the capabilities of the strong lensing simulation pipeline **SLSim** . Left panel: Predicted distribution of strong lenses for galaxy-galaxy strong lenses, lensed quasars and lensed supernovae with certain cuts in lens and source magnitudes in the LSST survey. Right panel: a sample of galaxy-galaxy strong lenses produced by SLSIM and injected into CosmoDC2. Such realistic images can be used to train machine learning methods to find and model strong lenses, and to assess the selection biased in the discovery and inference.

Second SLSim simulation plot — **Figure 3.** Illustration of the capabilities of the strong lensing simulation pipeline **SLSim** . Left panel: Predicted distribution of strong lenses for galaxy-galaxy strong lenses, lensed quasars and lensed supernovae with certain cuts in lens and source magnitudes in the LSST survey. Right panel: a sample of galaxy-galaxy strong lenses produced by SLSIM and injected into CosmoDC2. Such realistic images can be used to train machine learning methods to find and model strong lenses, and to assess the selection biased in the discovery and inference.

I contribute to strong-lens finding challenge within the LSST Dark Energy Science Collaboration (LSST-DESC) and the LSST Strong Lensing Science Collaboration (LSST-SLSC). This work prepares for the development of methods to find lenses from actual LSST survey data. My work includes developing simulated lens populations and images, supporting the development and validation of lens-finding methods, and building analysis pipelines for Rubin-era studies (see Fig. 3). I am also involved in Rubin Observatory observing operations, which gives me direct experience with the observational processes that produce the survey data.

I led the development of SLSim , an open-source Python package for simulating realistic strong-lens populations, including galaxy-galaxy lenses, lensed quasars, lensed supernovae, cluster lenses, and lensed kilonovae. SLSim supports the generation of realistic populations and images for the development, training, and validation of strong-lens discovery and characterization methods. These simulations help prepare the community to identify and study the strong-lensing systems that Rubin will discover.

The development and scientific applications of SLSim are described in our publication: Khadka et al. (2026) .

High-redshift cosmology with quasars and gamma-ray bursts

My Ph.D. research at Kansas State University focused on using quasars and gamma-ray bursts (GRBs) as cosmological probes. I studied the nonlinear relation between quasar X-ray and ultraviolet luminosities and its application to constructing a quasar Hubble diagram (see left panel of Fig. 4). Because quasars can be observed at substantially higher redshifts than Type Ia supernovae, this relation offers a possible way to extend cosmological distance measurements to earlier epochs of cosmic history.

qhubble diagram — **Figure 4.** **Left panel:** Hubble diagram of 1,598 quasars constructed using the L_X–L_UV relation. The black solid line represents the best-fit flat ΛCDM model with Ω_m = 0.60 from the 2019 QSO data. Red points show the mean distance moduli and their uncertainties in narrow redshift bins for the quasar data. These averages are not used in the statistical analysis and are shown only for visualization purposes. The black dashed line represents a flat ΛCDM model with Ω_m = 0.30. **Right panel:** Hubble diagram of 78 Mg II QSOs in the best-fit flat ΛCDM model. The magenta solid line shows the prediction of the best-fit flat ΛCDM model with Ω_m0 = 0.27 from the Mg II QSO-78 data set. Black and red data points represent the observed distance moduli and corresponding uncertainties for the Mg II QSO samples, respectively, in the best-fit QSO-78 flat ΛCDM model. The blue dotted line represents the standard flat ΛCDM model with Ω_m0 = 0.30.

second qhubble diagram — **Figure 4.** **Left panel:** Hubble diagram of 1,598 quasars constructed using the L_X–L_UV relation. The black solid line represents the best-fit flat ΛCDM model with Ω_m = 0.60 from the 2019 QSO data. Red points show the mean distance moduli and their uncertainties in narrow redshift bins for the quasar data. These averages are not used in the statistical analysis and are shown only for visualization purposes. The black dashed line represents a flat ΛCDM model with Ω_m = 0.30. **Right panel:** Hubble diagram of 78 Mg II QSOs in the best-fit flat ΛCDM model. The magenta solid line shows the prediction of the best-fit flat ΛCDM model with Ω_m0 = 0.27 from the Mg II QSO-78 data set. Black and red data points represent the observed distance moduli and corresponding uncertainties for the Mg II QSO samples, respectively, in the best-fit QSO-78 flat ΛCDM model. The blue dotted line represents the standard flat ΛCDM model with Ω_m0 = 0.30.

I also investigated cosmological constraints from reverberation-mapped quasars (see right panel of Fig. 4) and GRBs in combination with baryon acoustic oscillation (BAO) measurements and direct measurements of the Hubble parameter, H(z). This work explored the potential of these astrophysical probes to extend distance measurements to redshifts approaching z ∼ 8 and to test cosmological models over a broader redshift range.

Publications

For a complete and up-to-date list of publications, see my Inspire HEP profile.