Simon Benjamin on Architectures for Quantum Computing
Вставка
- Опубліковано 24 тра 2018
- Simon Benjamin - www.materials.ox.ac.uk/peoplep... - is a Professor of Quantum Technologies at Oxford. He is also the Principle Investigator for Oxford’s project on Quantum Optimisation and Machine Learning - quopal.com/
This episode is more technical than last week’s episode with John Preskill - blog.ycombinator.com/john-pre.... We start by covering some fundamentals then go into different approaches for constructing and scaling a quantum computer.
Read the transcript here - blog.ycombinator.com/simon-be...
The YC podcast is hosted by Craig Cannon - / craigcannon - Наука та технологія
Hi everyone. I'll keep my eye out here in case there are any questions -- feel free to ask anything!
Hello, straight into the question: When does an atom or any matter becomes a qubit? When you got it into an ion trap and isolate it, you get no information from it, but is there some kind of a limit to this information? What other ways to make a qubit? Is any isolated bit a qubit?
Hi. Well it's a good question. Anything could in principle be a qubit (even the famous Schrodinger's cat is one, in a way!) It's about isolating a single entity (e.g. an atom) from the rest of the world. If you think about a regular atom that's part of your body then it is constantly interacting with all the atoms around it. This interaction is quite like the other atoms constantly checking it "where are you? what are you doing?". In our ion trap devices the ions (an ion is just a charged atom) are isolated completely -- they don't touch any other matter. Then the make good qubits, because when we put them into a superposition of zero/one there are no other atoms interacting and (effectively) asking them to decide what state they are in.
As a follow up, what are the most common sources of decoherence? As I understands it, it's some part of "outside" interacting with the ion/qubit. Is it a matter of shielding from electromagnetic waves of all sorts, less than perfect vacuum, something more exotic like neutrinos?
Hey Simon, what do you think of the potential of neutral atoms as a base architecture for building general quantum computers vs superconducting qubits?
to Mopic3d: for ion traps indeed minimising the impact of stray fields (or fluctuations in the applied fields) is significant for extending the inherent decoherence time and achieving 50+seconds. This is sometimes called the "memory lifetime" since it is the lifetime of a qubit when nothing is happening to it. However, the more significant type of decoherence comes from the small imperfections that happen whenever we tell the qubit to do something i.e. its lifetime while it is computing. Here we measure the lifetime not in seconds but rather in number of operations, and we use this concept of 'fidelity' e.g. 99.9% fidelity means, in essence, you'll get about 1000 operations done before the qubit is hopelessly messed up. Since operations are fast, say a few microseconds, this kind of decoherence happens far faster than the memory-type decoherence.
This is a great science communication video.
Below I tried to summarize some of the topics discussed:
01:12 Qubit fidelity
13:00 Quantum error correction
23:27 Quantum computing at Oxford
29:13 Schrödinger cats, decoherence, and ion chips
37:37 Scalability: Engineering vs science
40:40 Different qubit platforms
46:22 Quantum supremacy and the need for near-term applications
54:53 Network scalability in a modular architecture
01:08:20 Challenges of tech transfer
For quantum computing and quantum tech news and analysis, look for my Quantum Tech newsletter, hosted also on Medium.
This podcast continues to make me smarter. Sooo good!
😇 Thank you for your amazing important video, this is very much valued and I always value your hard work !👍
Terrific explanation, thankyou
Good material, I was able to catch >80% of information. Thanks!
Also: There is a video about the ion trap quantum computer here -- ua-cam.com/video/aV1wL5jsfRU/v-deo.html
Very interesting.
Thanks for the great interview. As I understand, the first QC was proposed at Oxford by David Deutsch in 1985 (please correct me if I'm wrong)-37 years ago. A 37-year span from the transistor's 1947 invention lands us in the mid-1980s when the VLSI technology, Intel's 80386, and Moore's Law had already set the pace. Today, Nvidia's H100 boasts 80 billion gates! However, QC's progress seems significantly slower and the field is very much undecided on the ideal hardware platform. What explains the huge disparity between the rates of progress of QC hardware vs. the silicon evolution?
Were you able to demonstrate small QCs connected together optically, as you mentioned in the interview?
Can the qubits not just be laid out in a line (as Simon said would be ideal) because a 1d loop has positive total curvature or something?
35:20 - 50 secs at room temperature. Triggered X___X
Amazing eh? Achieved back in 2014, see arxiv.org/abs/1403.1524