An interactive mechanistic simulator that runs entirely in the visitor's browser. No Python server, GPU, paid inference API, external scripts, fonts or analytics are used. Hugging Face serves static files.
All outputs are synthetic and uncalibrated. This is a mathematical model, not a trained ML predictor or experimental evidence. It compares static quenching, dynamic quenching and inner-filter attenuation for Fe/Cu mixtures.
The default population contains 256 synthetic CQD classes, seed 42. Python exports the exact per-class coefficients for pH 3–9 (integer steps), 365/405 nm, and the optional simplified iron-equilibrium scenario. Browser JavaScript solves the shared Fe/Cu mass balances for your continuous concentration, capacity and coefficient-scale settings. It does not interpolate a response table. Tests compare its outputs to the Python reference.
F0 is a CQD-only blank. Copper absorption is unmodeled and assumed zero. The optional iron scenario assumes fixed pH, ideal activities, 25 °C, mononuclear hydrolysis and amorphous Fe(OH)3 equilibrium; it excludes buffers, ligands, redox and solid scattering. Predicted solids suppress optical outputs. The pH grid does not establish chemical feasibility for real samples.
The fuller Python notebook additionally supports custom absorption spectra, parameter distributions, simulated instrument noise and experiment-condition ranking. The browser version focuses on deterministic hypothesis comparisons.
Use the in-app downloads for the current response CSV, assumptions/provenance JSON and plot SVGs. A portable Colab notebook is also included.
This folder is prepared with python prepare_huggingface.py from
https://github.com/jeorgexyz/cqd-fluorescent-sensing . Source, model and dataset
publication instructions are in the repository's docs/huggingface.md.
Licensed under the MIT License. See LICENSE in the source repository.