MyFilter

MyFilter is a web app for interactive visualization of astronomical filter transmission curves.

Demo

https://preview.lmytime.com/myfilter

JWST NIRCam customized version:

https://preview.lmytime.com/myfilter?cus=JWST

Features

• Interactive transmission curves for common astronomical filters
• Overlay multiple filters for comparison
• Zoom in on specific wavelength ranges
• Check emission or absorption lines at any redshift
• Narrow-band / broad-band emission-line decomposition (/emline)

Usage

Select filters from the sidebar. Hover the graph for details, and:

Gesture	Effect
Drag	Zoom
Shift + drag	Adjust line redshift
Alt/Option + drag	Pan the wavelength window
Double-click	Zoom out

Presets are available via the cus query parameter: ?cus=JWST, ?cus=CSST, ?cus=Roman, ?cus=NIJI. The Share button builds a link that restores the current filter set, redshift, and preset.

Architecture

The app runs entirely on Cloudflare: a static single-page app plus a small Worker, with filter data in R2. There is no application server.

Browser ──► Cloudflare Worker ──► R2 (filters/v1/…)
   │             │
   │             ├─ /getfilter              CSV of resampled curves
   │             ├─ /myfilter/getemlineparam emission-line parameters
   │             └─ /api/v1/…               instrument index + bundles
   └── static assets (SPA) served directly from the edge

Python is a build-time dependency only. It mirrors the SVO catalog and compiles it into per-instrument JSON bundles; it does not run in production.

Path	Purpose
src/	Vue 3 + TypeScript SPA
src/lib/	Numerics shared between the browser and the Worker
worker/	Cloudflare Worker: API routes and R2 access
pipeline/	Python: SVO mirror → instrument bundles → R2
data/custom_filters/	User-contributed curves, tracked in git
tests/fixtures/	Golden outputs frozen from the original Python

Why the numerics are pinned to fixtures

/getfilter and the emission-line endpoint were ported from SciPy/pyphot to TypeScript. To make that safe, the original Python was run over real curves — including the sparsest (4-point) and densest (15,730-point) in the catalog — and its outputs frozen as golden fixtures. The port must reproduce them exactly.

Two consequences are worth knowing before you "fix" something that looks wrong:

• The resampling grid step is (wmax + wmin) / precision — a sum, not a difference. precision is therefore not a point count. This is preserved deliberately; changing it would alter every CSV the app has ever served.
• CSV floats follow Python's repr, not JavaScript's String. They differ on integral values, on the scientific-notation threshold, and on exponent padding. src/lib/pyfloat.ts reproduces Python's rules and is validated against 11,689 real tokens.

Development

Requires Node 20+ and Python 3.9+.

npm install
npm run dev

Tests — the pure numerics run in Node, the Worker routes run in the real Workers runtime with a simulated R2 binding:

npm test

Filter data

The SVO mirror (~11,000 curves, 114 MB) is not tracked in git; it is reproducible. Custom curves in data/custom_filters/ are tracked — see PROVENANCE.md.

npm run sync:svo      # refresh the local SVO mirror (incremental)
npm run data:build    # compile curves into dist-data/filters/v1/
npm run data:publish  # upload changed objects to R2

Deployment

Code and data deploy separately.

• Code — Cloudflare Workers Builds deploys on every push to main. The build command is npm run ci:verify, so a type error or a failing test aborts the deploy rather than shipping it.
• Data — a manual GitHub Actions workflow (Publish filter data to R2) syncs SVO, validates, builds bundles, and uploads the changed ones.

See docs/deployment.md for one-time setup: creating the R2 bucket, the three GitHub secrets, and the Workers Builds dashboard settings.

To deploy by hand instead:

npm ci && npm run deploy

Data sources

• Most curves come from the SVO Filter Profile Service.
• JWST_* — JWST ETC v2.0.
• CTIO-Merian — filters designed for the Merian Survey.
• Subaru-MOIRCS — Subaru MOIRCS.
• CSST-MS, CSST-MCI25 — CSST instrument teams.

Contributing

Contributions are welcome — please open an issue or a pull request. To add a filter, see PROVENANCE.md.

License

MIT.

Acknowledgment

If you publish work that uses MyFilter, please cite it via the Zenodo DOI:

In bibtex:

@misc{MyFilter,
  author       = {Li, Mingyu},
  title        = {{MyFilter: A Web Application for Interactive Visualization of Astronomical Filter Transmission Curves}},
  month        = nov,
  year         = 2023,
  publisher    = {Zenodo},
  version      = {1.0.0},
  doi          = {10.5281/zenodo.10210201},
  url          = {https://doi.org/10.5281/zenodo.10210201}
}

Optionally, attach the GitHub repo link in a footnote:

\footnote{\url{https://github.com/lmytime/MyFilter}}