Accurate Unsupervised Photon Counting from Transition Edge Sensor Signals

NeuroLibre Reprod. Preprints 46 (2026) · DOI: 10.55458/neurolibre.00046

License: CC BY 4.0.

Abstract

We compare methods for signal classification applied to voltage traces from transition-edge sensors (TES) which are photon-number resolving detectors fundamental for accessing quantum advantages in information processing, communication and metrology. We quantify the impact of numerical analysis on the distinction of such signals. Furthermore, we explore dimensionality reduction techniques to create interpretable and precise photon-number embeddings. We demonstrate that the preservation of local data structures of some nonlinear methods is an accurate way to achieve unsupervised classification of TES traces. We do so by considering a confidence metric that quantifies the overlap of the photon-number clusters inside a latent space. Furthermore, we demonstrate that for our dataset previous methods such as the signal's area and principal component analysis can resolve up to 16 photons with confidence above 90% while nonlinear techniques can resolve up to 21 with the same confidence threshold. Also, we showcase implementations of neural networks to leverage information within local structures, aiming to increase confidence in assigning photon numbers. Finally, we demonstrate the advantage of some nonlinear methods to detect and remove outlier signals.

Figures

6 panels with data across 6 figures. Each panel page shows the plot, its columns and its files; each data.csv begins with a header naming the paper, the panel, the source, the license and the provenance route.

Fig. 1

Illustrative figure, no extractable data. Shown in the paper PDF.

Fig. 2

• panel (1): Voltage (a.u.) against time (a.u.) for 2,048 overlaid raw transition-edge-sensor traces of 100 samples each, every tenth trace of the training and of the test half of the demonstration dataset at 28.5 dB attenuation, drawn at 5% opacity. data.csv

Fig. 3

• panel (1): Kernel density estimate of the dataset in the one-dimensional latent space, density against the latent coordinate $s_1$. The printed figure fills the area under the curve and writes the cluster photon numbers above the peaks; neither is drawn here. data.csv

Fig. 4

• panel (1): Photon-number distribution of the demonstration dataset: probability against photon number, one bar for each photon number from 0 to 9. data.csv

Fig. 5

• panel (1): Confidence of the photon-number clusters against photon number on the Synthetic Uniform dataset for eleven methods: Max, Area, PCA 1D and 2D, Isomap 1D, t-SNE 1D and 2D, UMAP 1D and 2D, Param. t-SNE 1D and Param. UMAP 1D. Dotted lines are 1D latent spaces, solid lines 2D. The print names each curve at its end; here a legend does. data.csv

Fig. 6

• panel (1): Confidence of the photon-number clusters against photon number on the Synthetic Geometric dataset for eleven methods: Max, Area, PCA 1D and 2D, Isomap 1D, t-SNE 1D and 2D, UMAP 1D and 2D, Param t-SNE 1D and Param UMAP 1D. Dotted lines are 1D latent spaces, solid lines 2D. The print names each curve at its end; here a legend does. data.csv

Fig. 7

• panel (1): Confidence of the photon-number clusters against photon number on the Synthetic Large dataset for Area, PCA 1D, UMAP 1D and Param UMAP 1D, all dotted (1D latent spaces). As in the print, the PCA 1D curve falls below the axis between photon numbers 11 and 15. The print names each curve at its end; here a legend does. data.csv

Cite

Nicolas Dalbec-Constant, Guillaume Thekkadath, Duncan England, Benjamin Sussman, Thomas Gerrits, Nicolás Quesada. Accurate Unsupervised Photon Counting from Transition Edge Sensor Signals. NeuroLibre Reprod. Preprints 46 (2026). https://doi.org/10.55458/neurolibre.00046

When you use hosted data, cite the original paper and give the panel's URL so a reader can find the exact values you used.

All papers · About · Statistics · Submit data · API