Overview

The only laboratory is the simulation.

Cosmology is an observational science with a single, un-repeatable experiment already in progress. One cannot rerun the formation of galaxies under altered conditions; one can only observe the result and ask which initial conditions and which physics could have produced it. The simulation is the closest thing the field has to a laboratory.

Program RP-053 builds and runs such simulations. It models how matter, under gravity and across cosmic time, assembled itself into the filaments, clusters, and voids that surveys now map — and it asks where the standard account fits the data and where it strains.

Structure

From smoothness to cosmic web.

The early universe was very nearly uniform. The structure we observe today — a vast web of matter threaded with empty voids — grew from minute density fluctuations amplified over billions of years by gravity. Reproducing that growth faithfully requires tracking enormous numbers of bodies across an enormous span of time.

The program's N-body simulations do exactly this, and the engineering challenge is as real as the physics. Gravity couples everything to everything, but the matter is distributed sparsely and unevenly, and an algorithm that does not exploit that sparsity will exhaust any computer that exists.

Dark Sector

Modeling what cannot be seen.

The overwhelming majority of the universe's contents neither emit nor absorb light. Dark matter shapes the gravitational scaffolding on which visible structure hangs; dark energy governs the accelerating expansion of space itself. Neither is directly observed, and both must be inferred from their effects.

The program treats the dark sector as a set of hypotheses to be tested against simulation. By varying the assumed properties of dark matter and dark energy and comparing the resulting structure to observation, it works to narrow the range of models the universe will tolerate.

Gravitational Waves

The signature of colliding remnants.

When compact remnants — black holes, neutron stars — spiral together and merge, they radiate energy as ripples in spacetime itself. The program models these inspirals to predict the waveforms detectors should expect, sharpening the templates against which faint signals are recognized.

This is among the program's most directly testable work: a prediction that meets, or fails to meet, an observed event. The field has lately delivered such events at a quickening pace, and each one is a verdict on the modeling that preceded it.