BBQLab @CLEO23

Saying goodbye to CLEO in San Jose from the air!

We recently attended the very exciting CLEO 2023 conference in San Jose, CA. The conference included a fascinating range of talks from academia and industry on topics as diverse as deep neural networks, multi-mode fibres, and critical coupling. Mehul presented an invited talk on “harnessing complexity for manipulating spatiotemporal entanglement” at the Symposium on Enabling Highly Multimode Nonlinear and Quantum Photonics, organised by Logan Wright and Marco Piccardo. He also presented a talk on our work on our work on noise and loss-robust quantum steering in the Quantum Network Protocols session. Besides all the great science, it was also very nice to catch up with old friends and colleagues from around the world!

PRL: Sorting overlapping Quantum States

Evolution of three input modes in red, green and blue that are sorted into respective outcomes with their overlap sorted into the ambiguous outcome that turns white.

We are excited to announce that our latest work ‘Simultaneously Sorting Overlapping Quantum States of Light‘ has been published in Physical Review Letters. In this collaboration with the QOCI Group, we demonstrate simultaneous and efficient sorting of non-orthogonal transverse-spatial states of light in up to seven dimensions. This has been made possible by employing a multi-plane light converter (MPLC) to program high-dimensional POVMs that correspond to unambiguous discrimination of the quantum states. The MPLC employs an additional auxiliary outcome that sorts the overlap of all the modes into an ambiguous outcome.

An implication of this method is that we can sort overlapping images encoded with coherent sources. We demonstrate this by sorting three smiley faces with an accuracy of 97.6%, implying accurate image classification with light!

arXiv: Multi-Plane Neural Networks

Characterising multiple complex media with machine learning

In our new preprint titled Referenceless characterisation of complex media using physics-informed neural networks, we show how multi-plane neural networks (MPNN) can be used to recover the complex transmission matrix of a commercial multi-mode fibre in a noise-robust manner, without using a reference field! We also show how the MPNN technique can be used to characterise a series of independent complex media, as shown in the figure above. This work will have many applications ranging from classical optical networks, biomedical imaging, to quantum information processing. As just one example, the MPNN technique forms a central part of our previous work on programming high-dimensional quantum gates inside a multi-mode fibre using inverse-design.

Congrats Dr. Herrera Valencia!

One of the many post-defense celebrations!

A huge congratulations to Dr. Natalia Herrera Valencia for successfully defending her PhD thesis! Natalia joined us in 2019 following a Europhotonics Erasmus Mundus Masters. Her thesis work has focused on the development of a new platform for high-dimensional entanglement and its unprecedented transport through a complex scattering medium. Natalia will continue as a postdoctoral research associate in BBQLab and we are very pleased that she is staying with us for some more time!

Poster advertising Natalia’s PhD seminar, designed by the talented Vatshal Srivastav