Dr. Jonathan Gordon was my first Ph.D. student, who worked tirelessly trying to understand if quasi-n-body simulations known as COLA could be applicable in the context of Rubin Observatory (Cosmic Shear). So what is COLA and why it is important?

The standard method for evaluating nonlinear gravitational collapse involves dark matter-only n-body simulations. Although these simulations provide excellent precision for weak lensing studies, they are computationally expensive. It is simply not feasible to create n-body-based nonlinear matter power spectrum emulators in all appealing models beyond LCDM. The COmoving Lagrangian Acceleration (COLA) is an exciting alternative to full simulations. COLA combines higher-order Lagrangian Perturbation Theory (LPT) on large scales with n-body code on small scales. One COLA run takes a around 3 hours on a 128-core node.

COLA simulations can reproduce the nonlinear response function down to k around 1h/Mpc when compared with the Euclid Emulator v2.0 within few percent accuracy. Dr. Gordon performed a COLA-based LSST Cosmic Shear analysis on wCDM models in collaboration with Profs. Kazuya Koyama and Hans Winther. Dr. Gordon created a novel way to combine high-resolution n-body in the LCDM plane (every extended cosmology contains the LCDM subset of parameters) with COLA sims in the extended parameter directions to create an emulator precise enough for cosmic shear LSST-Y1 with scale-cut theta_{min} = (11.0', 34.8') for cosmic shear components, respectively. His work was the most comprehensive end-to-end LSST-Y1-related study analyzing the precision of the COLA method.

Dr. Jonathan Gordon transioned to industry jobs but he continue to help our group to finish a second analysis expanding his previous work to test dynamical dark energy with Chevallier, Polarski, and Linder (CPL) parameterization for the dark energy equation of state.