2020 11 19

Meeting November 19, 2020

Recap

• Ask Jo

  • about JupyterHub on his server
  • Spherical Harmonic Expansion
  • Numerical methods for going to a DF from a non-axisymmetric, analytic potential / density

• Upload Mathematica Derivation as PR (@CCAstro35)

  • Create branch
  • put derivation in branch
  • publish branch to github
  • create Pull request & request @nstarman review

Meeting Notes

@CCAstro35 showing Residual Grid in AGAMA

@nstarman showed the relative merits of different fit statistics. One statistic that needs further consideration is the RMS.

Sampling lattice discussion: the difficulties of how to lay down a lattice where the points are determined by the density of the potential.

We are now working on creating a coherent code framework to do the sampling.

Our residual function is wrong. We should be looking at the differential of the potential, not the potential itself.

@nstarman talked to Jeremy Webb about lattices. We are looking for an Adaptive Mesh Refinement code. This is a list of related links:

• http://flash.uchicago.edu/site/flashcode/
• https://stackoverflow.com/questions/32389538/adaptive-mesh-refinement-python
• https://stackoverflow.com/questions/59216521/easy-to-use-adaptive-mesh-refinement-for-characteristic-function-python
• https://github.com/TUD-RST/symbtools/blob/master/symbtools/meshtools.py
• https://github.com/adamdempsey90/NDTAMR
• https://scicomp.stackexchange.com/questions/923/what-simple-methods-are-there-for-adaptively-sampling-a-2d-function
• https://www.salome-platform.org/forum/forum_10/303518492/view
• https://github.com/topics/adaptive-mesh-refinement
• https://fenicsproject.org/olddocs/dolfin/1.3.0/python/demo/documented/subdomains-poisson/python/documentation.html
• https://cims.nyu.edu/cmcl/software.html
• https://ngsolve.org/docu/latest/whetting_the_appetite/adaptive.html

@nstarman talked to Jo about Lattices. Suggest Voronoi tesselations and Barnes-Hut tree. Also, as a good ab initio, just lay down a cylindrical mesh for the disc. Alternatively, lay down a fine mesh using the most-applicable symmetry and weight each lattice point by the potential / density / gradient thereof.

• https://en.wikipedia.org/wiki/Centroidal_Voronoi_tessellation is promising
• https://epubs.siam.org/doi/10.1137/S0036144599352836 on tessellations

So, lattice options, ranked in increasing complexity:

1. just lay down a mesh
2. weight each point in the mesh by the potential / density / gradient thereof
3. adapt the mesh

For Next Week

• Continue clean everything. (@nstarman)

  • Continue PRs for project configuration
  • Make some Plots of the the DFs in the notebook

• Poisson Noise for Hernquist Spheres

  • Sample Galpy (@nstarman)
  • Sample AGAMA (@CCAstro35)

• consider the RMS statistic