Skip to content
This repository has been archived by the owner on Aug 6, 2024. It is now read-only.

2020 10 02

Jump to bottom
Nathaniel Starkman edited this page Oct 25, 2020 · 3 revisions

Meeting Oct 2, 2020

This is our first meeting.

Goals

  • Covariance Matrix (or full non-linear "covariance") of potential expansion reconstructions from both analytic and N-Body sims

  • how many terms to go to?

  • the errors in the terms

  • observational errors

  • non-axisymmetric features: bars, spirals, perturbations

Perturb simulations. Run the expansion. see how they differ. Use the best-fit expansion for "adjoining" potentials

Run on a bar, on a disk. How are they when separated, when together? Can we "integrate" out pieces of a potential. Can we understand the "cross"-terms in a a relaxed potential between components. And how they contribute to the reconstruction error.

How do the errors evolve between similar sets of parameters. Can the likelihood around one set, be applied to a neighboring set of points?

Go forwards -- obs error, points bootstrapping, time evolve, components strength — to errors / likelihood.
Go backwards from likelihood to errors on MLE-fit. This feels like ABC or maybe also machine learning.

Stretch Goals

  • SCF for N-Body snapshots in Galpy

  • CylSP (Vasiliev 2015) in Galpy

    • Analytic
    • Snapshot

Projects

  • Add data from a sim [#6]
  • Data Loader API [#9]

Next Week

Read:

Start making data loaders.

Meeting Notes:

Clone this wiki locally