27th September 2024, the European Quantum Industry Consortium (QuIC) hosted its first ever webinar, under the Q WAVE banner, with the topic the Quantum industry is most concerned with as we all work towards rolling out our quantum enabled tools.
“The journey towards fault-tolerant quantum computing in Europe” much anticipated webinar began with guest speakers on board who were able to explain what efforts they were making in advancing quantum error correction.
ALICE & BOB + RIVERLANE
Attendees were in great company with Earl Campbell, VP of Quantum Science at Riverlane and Jérémie Guillaud, Chief of Theory at Alice & Bob at the helm. The webinar was moderated by David Morcuende, the QuIC Network Manager.
Earl Campbell tailored his session on QEC software advancements that would facilitate the optimization of quantum systems as a whole. For sure the race is underway in Europe and globally for robust solutions to fault-tolerant computing and the exciting developments over the past two years. Campbell specified that the current error rate of 0.1- 1% was not robust enough and his goal was to reach optimum levels of correction by developing logical qubits that have the ability to run off thousands of qubits and therefore advanced algorithms. Useability of algorithms is the key to error correction measurement as they require trillions of quantum operations. Current figures see error rates of up to one in 1,000. Campbell confirmed the quantum industry needs a minimum error rate ratio of 1 in a million in order to calculate quantum at a reliable value. BUT to deliver real utility, the average user requires error rates of one in a trillion.
GOOGLE 7 X 7 SURFACE CODE PATCH OUTPUT
Along with a study from ETH Zurich that succeeded in showing improvement factors based on simulations during tests of repeated quantum error correction in a three-surface distance, it was further determined that the industry will need to improve the physical error rate by a factor of 2. Campbell also highlighted a study from Google that uses an alternative technique to suppress quantum errors by scaling a surface code logical qubit.
The Google team saw an improvement in each logical qubit as physical qubit use increases, significantly closing the error correction threshold.
By mid-September 2024 we learnt that the Google quantum team had reached 7 x 7 surface code patch, an experiment that shows how logical qubit has finally improved to a factor of 2 in the error rate. In fact, any increase to the surface code patch has led to a clear suppression of a factor of 2 in the logical qubit error rate. The Google experiment indicates that logical errors can indeed be suppressed.
INDUSTRY SUMMARY OF ACHIEVEMENTS
ETH Zurich 2022- Logical qubit suffers higher error rates than physical qubit
Google 2023- Logical qubit improves slightly as more physical qubits are utilized
Harvard 2024- Neutral atoms leap to 48. Error detected for logical qubits
Google September 2024- logical qubit is recorded as achieving an improvement of over 2 x more each time physical qubits are used.
eFTQC
early Fault Tolerant Quantum Computing
Jérémie Guillaud, Chief of Theory at Alice & Bob, examined the CAT architecture methodology to reduce the overheads required for fault tolerant quantum computing so that the technology could become highly scalable and thus more accessible.
Guillaud confirmed the number of qubits required to generate high value logical qubits required is a few hundred to a few thousand.
CAT QUBIT
AWS- easy to stabilise
Polarizes noise
GKP QUBIT
Nord Quantique
Considered more difficult to stabilize
Minimizes noise
THE AI BUBBLE, LLMs VERSUS SLMs
In keeping with the LatestSale.com goal to develop sustainable technology solutions under its organic technology™ banner, hence the use of an SLM strategy and opposed to an LLM build, the webinar assessed the need to work for a purpose and value, whilst ensuring positive societal impact was achieved. As the tech industry as a whole looks at the parameters surrounding Artificial Intelligence and asks itself the questions that matter the most such as, “Will AI serve meaningful problems?” Guillaud asserts, “To complete a task faster will not stop an AI bubble if no legitimate client problem is solved.”
The industry is equipped with the quantum tools to reach error correction, but it was clear during this webinar that questions still remain as to how genuine client problems can be identified and then validated. AI for the purpose of mere execution is not sufficient if measurement metrics are not in place.