Uhlmann's theorem states that the fidelity of two states rho and sigma can be expressed in terms of their purifications (where the fidelity simplifies). In arxiv.org/abs/2502.01749 we (with @giulia_qp and Renato Renner) generalize Uhlmann's theorem to Rényi relative entropies.
Follow up work on the effective dimension (with M. Datres, GP. Leonardi, @AFigalli):
(i) stronger generalization bound
(ii) lower bound that can be computed efficiently (for Markovian models)
(iii) new simulation results
--> check out arxiv.org/abs/2401.09184
We finally understand how to (optimally) cut arbitrary two-qubit gates. Two interesting findings:
(1) cutting multiple gates jointly is always cheaper than cutting them individually
(2) classical communication does not help
---> check out arxiv.org/abs/2312.11638
There is something mesmerising about seeing one's paper(s) completely plagiarised, theorem by theorem, without any reference, and posted on arXiv. Compare
arxiv.org/abs/2010.02152
with our
arxiv.org/abs/1604.03023
In recognition of his contributions to mathematical physics, Jürg Fröhlich of @ETH_en is being inducted into the National Academy of Sciences. Watch the #NAS159 livestream: ow.ly/uuXh50IVNP2
New arXiv paper out! Purely classical machine learning :) in collaboration with @quantum_sutter@AFigalli and Stefan Woerner. We define and motivate a measure of capacity for ML models: the local effective dimension. We show how this quantity bounds generalization error 1/3
Happy to finally get this paper published and looking forward to seeing more results that bring error mitigation and correction together. @quantum_sutterjournals.aps.org/prl/abstrac…
I'm at #iccmp2021 today to receive the 2019 AHP prize awarded jointly with my co-authors Ralph Silva and @postquantum for arxiv.org/ct?url=https:/…. I wish they could be there. Nice to see @quantum_sutter (AHP prize 2018) and J. Bauch (AHP prize 2017) in person again!
When we first derived error mitigation we tried to make it clear it was not error correction. Since then I have wanted to bring them together and in this work we start to close the gap. Still a long way to go but I really like this direction of research arxiv.org/abs/2103.04915
Error mitigation techniques may allow a smooth transition into the fault-tolerant computing era. Together with Christophe and Stefan from @IBMResearch and @ETH_physics we improve the sampling overhead for the quasiprobability method arxiv.org/abs/2101.09290
Happy to share a new quantum algorithm to solve convex dynamic programming problems with a quadratic speedup. Joint work with Giacomo-@sutter_tobias-Stefan from @IBMResearch/@EPFL_en (see arxiv.org/abs/2011.11654)