Sign in

Timo Hillmann

@timohillmann.bsky.social
504 followers 173 following 43 posts

PhD student @ Chalmers University of Technology

PostsRepliesMedia
Timo Hillmann @timohillmann.bsky.social · 14/02/2026
I heard decoding qLDPC codes? Let’s see what we can do :)
010
Reposted by Timo Hillmann
Lucas Berent @lucasberent.bsky.social · 13/10/2025
happy to share this preprint from last week scirate.com/arxiv/2510.0... - my first with Iceberg! big thanks to Larry and Nouedyn, who led this cool project ❄️ we have also written a short blogpost about it www.iceberg-quantum.com/blog/fast-su... comments welcome 🧊
scirate.com
Fast surgery for quantum LDPC codes
Quantum LDPC codes promise significant reductions in physical qubit overhead compared with topological codes. However, many existing constructions for performing logical operations come with distance-...
0191
Timo Hillmann @timohillmann.bsky.social · 03/09/2025
Thanks for this collaboration @lucasberent.bsky.social, Armanda Quintavalle, @jenseisert.bsky.social, Robert Wille, and @qec.codes (Joschka Roffe)
010
Timo Hillmann @timohillmann.bsky.social · 03/09/2025
LSD can get its soft information from a pre-processor like BP or the device's noise model directly, see also VibeLSD. arxiv.org/abs/2508.15743 It matches the performance of the best decoders out there, like OSD-O, but with a reduced runtime complexity.
arxiv.org
Colour Codes Reach Surface Code Performance using Vibe Decoding
Two-dimensional quantum colour codes hold significant promise for quantum error correction, offering advantages such as planar connectivity and low overhead logical gates. Despite their theoretical ap...
110
Timo Hillmann @timohillmann.bsky.social · 03/09/2025
All of this is made possible through a novel parallel matrix factorization strategy, which we call on the-fly elimination, to identify, validate, and solve local decoding regions on the decoding graph.
110
Timo Hillmann @timohillmann.bsky.social · 03/09/2025
Working on LSD was fun! - Our paper on Localized statistics decoding (LSD) for qLDPC codes is now in print. LSD is a highly parallelizable, reliability-guided inversion decoder that exploits the clustering of errors. Think of it as a parallel version of OSD. www.nature.com/articles/s41...
nature.com
Localized statistics decoding for quantum low-density parity-check codes - Nature Communications
Quantum low-density parity-check (QLDPC) codes offer lower overhead than topological quantum error-correcting codes, but decoding remains a key challenge for scalable fault-tolerant quantum computing....
1163
Reposted by Timo Hillmann
Arne Semsrott @arnesemsrott.bsky.social · 01/07/2025
Der Polizist behält seinen Job, denn: "Aus den Nachrichten ergebe sich noch keine eindeutig verfassungsfeindliche Gesinnung."
26378123
Timo Hillmann @timohillmann.bsky.social · 30/06/2025
I think I’m missing :)
100
Timo Hillmann @timohillmann.bsky.social · 03/06/2025
Since even for surface codes Relay-BP (Fig. 2) competes with matching-based decoders, it would be interesting to see whether Relay-BP outperforms known color code decoders, relevant, for example, to the performance of magic state cultivation.
010
Timo Hillmann @timohillmann.bsky.social · 03/06/2025
As Gallager already noted, BP is often more art than science, justified “only by the fact that it works.” But it does work in many cases, and I think it’s worth exploring how far we can push it with careful heuristics and structured approaches.
120
Timo Hillmann @timohillmann.bsky.social · 03/06/2025
Another thing I noticed: there has been only limited effort to combine different strategies to mitigate trapping sets. I had actually written this down as a future project idea but didn't get to it, so I’m glad to see that the IBM Quantum team has already shown that this direction is viable.
110
Timo Hillmann @timohillmann.bsky.social · 03/06/2025
While writing my thesis, I revisited a lot of the BP literature and dug into the basics. One thing that stood out is that heuristics for trapping sets are rarely tested under circuit-level noise, even though decoding in that setting is significantly different from code capacity.
120
Timo Hillmann @timohillmann.bsky.social · 03/06/2025
Great work by the IBM Quantum team on Relay-BP, a BP-only heuristic decoder for general qLDPC codes. From a first look, the performance seems strong and further supports qLDPC architectures from a decoding-focused perspective. arxiv.org/abs/2506.01779
arxiv.org
Improved belief propagation is sufficient for real-time decoding of quantum memory
We introduce a new heuristic decoder, Relay-BP, targeting real-time quantum circuit decoding for large-scale quantum computers. Relay-BP achieves high accuracy across circuit-noise decoding problems: ...
1151
Reposted by Timo Hillmann
Seok-Hyung Lee @seokhyung-lee.bsky.social · 16/05/2025
I recently created a simple Python wrapper for the Tesseract decoder for personal use, but I’ve decided to make it public in case others find it useful. Feel free to let me know if you encounter any issues! github.com/seokhyung-le...
github.com
GitHub - seokhyung-lee/tesseract-python: Python Wrapper for Tesseract Decoder
Python Wrapper for Tesseract Decoder. Contribute to seokhyung-lee/tesseract-python development by creating an account on GitHub.
061
Reposted by Timo Hillmann
Henrik Wilming @arrr.de · 14/05/2025
There is now an official announcement: The TIB has archived the current status of the arXiv: blog.tib.eu/2025/05/14/p...
blog.tib.eu
Protecting Science: TIB builds Dark Archive for arXiv - TIB-Blog
Research and science are international; it is not for nothing that we speak of international specialist communities. Although a service such as arXiv is operated by an institution based in the USA, na...
1156
Timo Hillmann @timohillmann.bsky.social · 07/05/2025
Did the text change significantly? If I recall correctly the previous ones then they read very similar. Only the word „exceptional“ didn’t appear every time if my mind is not lying to me
020
Reposted by Timo Hillmann
Laura Lopez @ohdearz.bsky.social · 03/05/2025
New AIP report discusses impact of funding cuts in physics & astro depts: “No one can be productive in a climate like this—it is the end of the US scientific and technical dominance in the world—we will not recover from this in a decade...”. ⚛️🔭 physics.aps.org/articles/v18...
Excerpt from APS article on the new AIP report on impact of funding cuts in physics & astro departments: "One cannot work like this - all energy goes in thinking how to survive and not in physics - it is like being in the war," another department chair wrote. "No one can be productive in a climate like this - it is the end of the US scientific and technical dominance in the world - we will not recover from this in a decade," they added.
04623
Reposted by Timo Hillmann
Max Czollek @maxczollek.bsky.social · 02/05/2025
Ok, heute ARD-Brennpunkt zur AfD, auf einen der Gäste kommt ihr nie. Richtig, Tino Chrupalla. Vor 80 Jahren hätte der zumindest auf der Anklagebank gesessen. Woher diese unbändige Lust, weiter mit Nazis zu reden? Ist es Masochismus? Erinnerungskultur tm? Oder vermisst ihr einfach eure Omas und Opas?
1636759
Reposted by Timo Hillmann
qctip2025.bsky.social @qctip2025.bsky.social · 23/04/2025
Next up in session A is Lucas Berent telling us about LSD, a popular topic in Berlin.
1133
Reposted by Timo Hillmann
Mark M. Wilde @markwilde.bsky.social · 12/04/2025
This past week, I received a second stop work order on one of my grants. This now means that we can no longer spend on *two* of my major grants. Several of my group members will now graduate earlier than expected, I am looking into being a consultant for the summer, and students will do internships
35412
Reposted by Timo Hillmann
Ingwar Perowanowitsch @ingwarpero.bsky.social · 10/04/2025
Dass ausgerechnet die CSU-Politikerin Doro Bär, die noch vor wenigen Jahren den menschengemachten Klimawandel leugnete, jetzt neue Forschungsministerin werden soll - ein Job, bei dem es ganz besonders darauf ankommt, der Kraft der Wissenschaft zu vertrauen, ist m.E. hochproblematisch.
1892600672
Timo Hillmann @timohillmann.bsky.social · 07/04/2025
If now for subsystem codes we are measuring non-commuting gauge operators and combining them to get stabilizers, there is a way to look at the problem in the same way as for the ordering of operations in ordinary syndrome extraction circuits. Probably with even fewer constraints though. 2/2
010
Timo Hillmann @timohillmann.bsky.social · 07/04/2025
I wonder if with hindsight we are just coming back to realize that the order of operations in the syndrome extraction circuit matters for the effective fault distance of ordinary codes. There we all agree that the order of operations matters. 1/2
110
Timo Hillmann @timohillmann.bsky.social · 07/04/2025
Section 3 of the Extractor paper is a very good overview I find as a non-expert. arxiv.org/abs/2503.10390
arxiv.org
Extractors: QLDPC Architectures for Efficient Pauli-Based Computation
In pursuit of large-scale fault-tolerant quantum computation, quantum low-density parity-check (LPDC) codes have been established as promising candidates for low-overhead memory when compared to conventional approaches based on surface codes. Performing fault-tolerant logical computation on QLDPC memory, however, has been a long standing challenge in theory and in practice. In this work, we propose a new primitive, which we call an $\textit{extractor system}$, that can augment any QLDPC memory into a computational block well-suited for Pauli-based computation. In particular, any logical Pauli operator supported on the memory can be fault-tolerantly measured in one logical cycle, consisting of $O(d)$ physical syndrome measurement cycles, without rearranging qubit connectivity. We further propose a fixed-connectivity, LDPC architecture built by connecting many extractor-augmented computational (EAC) blocks with bridge systems. When combined with any user-defined source of high fidelity $|T\rangle$ states, our architecture can implement universal quantum circuits via parallel logical measurements, such that all single-block Clifford gates are compiled away. The size of an extractor on an $n$ qubit code is $\tilde{O}(n)$, where the precise overhead has immense room for practical optimizations.
082
Timo Hillmann @timohillmann.bsky.social · 14/03/2025
Typing away on my PhD thesis - no matter where I go next, the views from my office will be hard to beat, unfortunately.
0140
Timo Hillmann @timohillmann.bsky.social · 14/03/2025
@arrr.de is working on something related.
110
Timo Hillmann @timohillmann.bsky.social · 13/03/2025
Read the paper here: journals.aps.org/pra/abstract...
journals.aps.org
Quantum error correction with dissipatively stabilized squeezed-cat qubits
Noise-biased qubits are a promising route toward significantly reducing the hardware overhead associated with quantum error correction. The squeezed-cat code, a nonlocal encoding in phase space based ...
000
Timo Hillmann @timohillmann.bsky.social · 13/03/2025
When we wrote the theory proposal for the dissipatively stabilized squeezed cat qubit, we got into an argument with the copy editors of the journal. I had a strong opinion against calling it „cat state qubit“ (as no one does) which got so far that we almost had to add a disclaimer, see the mail
1110
Timo Hillmann @timohillmann.bsky.social · 27/02/2025
cubeIP
000
Reposted by Timo Hillmann
Henrik Wilming @arrr.de · 15/02/2025
I think it would be a good time to reinstantiate the mirror network for the # arXiv. It’s an essential piece of the global research community in math, physics, computer-science and adjacent fields. Under the current circumstances it should not only be hosted in the US. #quantum #math #physics #cs
66319
Reposted by Timo Hillmann
Diego Ruiz @diego-ruiz.bsky.social · 14/02/2025
First realization of a dissipatively stabilized squeezed cat qubit (a slight variation that we called a moon cat 🌛 actually), it was super interesting to work on this with experimentalists! arxiv.org/abs/2502.07892 My two key takeaways ⬇️⬇️
arxiv.org
Enhancing dissipative cat qubit protection by squeezing
Dissipative cat-qubits are a promising architecture for quantum processors due to their built-in quantum error correction. By leveraging two-photon stabilization, they achieve an exponentially suppres...
1123
Timo Hillmann @timohillmann.bsky.social · 13/02/2025
@alice-and-bob.bsky.social
000
Timo Hillmann @timohillmann.bsky.social · 13/02/2025
I’m super excited to see the experimental realization of a variant of the dissipatively stabilized squeezed cat that we proposed! Congrats to Alice & Bob and everyone else involved Especially great to see that it comes with similar performance gains as promised by the theory arxiv.org/abs/2502.07892
arxiv.org
Enhancing dissipative cat qubit protection by squeezing
Dissipative cat-qubits are a promising architecture for quantum processors due to their built-in quantum error correction. By leveraging two-photon stabilization, they achieve an exponentially suppres...
1234
Timo Hillmann @timohillmann.bsky.social · 11/02/2025
@adamwills1.bsky.social updates is on his recent work with collaborators on constant overhead magic state distillation and quantum codes with addressable and transversal non-Clifford gates #Coogee25
0100
Timo Hillmann @timohillmann.bsky.social · 11/02/2025
Armanda Quintavalle giving a great introduction to Quantum Tanner codes and talking about on going work of finding small~ish examples with good properties. #Coogee25
1150
Timo Hillmann @timohillmann.bsky.social · 10/02/2025
@mattmcewen.bsky.social kicking of #Coogee25.
0161
Timo Hillmann @timohillmann.bsky.social · 25/01/2025
But a uniform model is not the reality. A coupler between to chips has much higher attenuation than a single beam splitter (see methods).
100
Timo Hillmann @timohillmann.bsky.social · 25/01/2025
But that’s only true if we express attenuation as a percentage. As the dB conversion contains a logarithm, on a dB scale, losses add up. So in a uniform model you could still say that every component needs losses reduced by a factor of a 100, but starting from maybe .5 dB of total loss.
100
Timo Hillmann @timohillmann.bsky.social · 25/01/2025
These requirements really depend upon the specific choice of the state factory made, but there is one where for each path ~1% losses are tolerable. This is Fig. 4 in the paper. Thats all accumulated loss of all elements along that path, which multiplies as you correctly said.
100
Timo Hillmann @timohillmann.bsky.social · 25/01/2025
No, the architecture and simulations are structured based on loss paths, e.g. all losses from state factory to cluster state generation to final measurement. Currently there are for distinct ones, P1, P2, P3 and P4, with P1-3 having the most stringent requirements.
130
Timo Hillmann @timohillmann.bsky.social · 25/01/2025
Typing this last evening I made a typo. The attenuation is -10 log10(transmission) or equivalently -10 log10(1 - losses)
000
Timo Hillmann @timohillmann.bsky.social · 24/01/2025
Thresholds will improve, I'm confident about that.
010
Timo Hillmann @timohillmann.bsky.social · 24/01/2025
I'm not a big fan of the dB variant. I think the takeaway is that we need to get to ~99% transmission. My understanding is that many optical components are in general not super far away from this, but then you have input-output losses and other things that get you once you plug things together...
120
Timo Hillmann @timohillmann.bsky.social · 24/01/2025
Converting attenuation % to dB works as A = -10 log10(1 - transmission), this gives ~10 dB attenuation for the percentages quoted. With current threshold estimates, we need losses down to ~1% (transmission ~99%). this comes down to less then ~ one-tenth of attenuation in dB.
200
Timo Hillmann @timohillmann.bsky.social · 24/01/2025
Let me try (my background is not optics). In optics, people typically report losses/attenuation in dB. The paper also shows the losses in % for the paths given, with the worst being ~95% (5% transmission). For example, a single photon would be lost with a 95% probability.
100
Timo Hillmann @timohillmann.bsky.social · 23/01/2025
Back in the days here team.inria.fr/qasar/seminar/
team.inria.fr
Seminar – Quantum Architectures, Small And Reliable
030
Timo Hillmann @timohillmann.bsky.social · 23/01/2025
For example, lower thresholds requirement lower GKP squeezing which reduce the number of breeding layers, which reduce the number of input non-Gaussian states, reducing the requirement on multiplexing. This reduces the optical circuit depth and thus reducing losses along the optical paths.
040
Timo Hillmann @timohillmann.bsky.social · 23/01/2025
What we see is that with improvements in state generation and qec thresholds / logical error rates, numbers can go down very quickly.
120