Ecosyste.ms: Awesome

An open API service indexing awesome lists of open source software.

Awesome Lists | Featured Topics | Projects

https://github.com/hombit/hipscat-import-sfd


https://github.com/hombit/hipscat-import-sfd

Last synced: 7 days ago
JSON representation

Awesome Lists containing this project

README

        

# SFD map for HiPSCat and LSDB

This repository contains the SFD map tooling for HiPSCat and LSDB.

### Data exploration

`find_largest_gradient.ipynb` looks for the largest gradient between two points of the SFD map.
We use this to estimate the largest healpix order we can use to have interpolation errors to be smaller than the SFD map errors of 16%.
These 16% corresponds to the order 13, but we use the order 14 to be on the safe side.

### Get SFD map of the fixed healpix order

`interpolate_sfd.ipynb` interpolates SFD map into the healpix map of the order 14, using center of each tile as the interpolation point.
The result is saved into intermediate parquet files.

### Get multiorder SFD map

It is an alternative way to get the SFD map, chose one of the two: this one or previous one.

`build_multiorder_sfd_map.ipynb` builds a multiorder SFD map from order 17 map using the threshold of 0.01 (it is 16% scaled down from norder 13 to 17).
It uses Python package `mom_builder` to do the job.
The package is a Rust library with Python bindings, built with `maturin`.
Create virtual environment, install `maturin` and build the package with `maturin develop --release` from `mom_builder` folder.
The result is saved into intermediate parquet files.

### Import the map into HiPSCat

`import_sfd_as_catalog.ipynb` uses `hipscat-import` to create a HiPSCat catalog.

### Use the map in LSDB

`join_stars_and_sfd.ipynb` loads the SFD map and the test LSDB point source catalog and joins them together.
It implements the merging algorithm through the cross-matching interface of LSDB.

## Legacy

### Multiorder HiPSCat in Pure Python

`legacy/mom` folder contains a pure Python implementation of multiorder healpix map for SFD data.
It works extremely slow, because it merges the highest order map from bottom up with a lot of single tile work.
Python overhead makes it super-slow, new Rust implementation is in `mom_builder` folder.