如何評價"世界首台超越早期經典計算機的光量子計算機在我國誕生"?

世界首台超越早期經典計算機的光量子計算機在我國誕生

我來轉篇quora 的回答吧（我翻譯過了，有不對的地方歡迎指正）。各位自己看著辦。潘校是我科的教授，能做出這個工作還是很不容易的。對於做過科研的人，我覺得應該持保守態度。

另外：據我了解做成10比特的不是光量子計算機，而是超導量子比特系統。

https://www.quora.com/What-are-the-pros-and-cons-of-superconducting-Josephson-junctions-diamond-NV-centers-and-topological-materials-as-physical-implementations-of-qubits-Which-is-the-most-promising-physical-implementation-in-the-long-run

對於量子計算來說，超導約瑟夫結，金剛石NV-Center, 拓撲材料，那個是實現量子計算體系最有希望的？

Some of the key parameters to consider - the stability of physical qubits (time to decay) and the amount of noise in quantum evolutions, the necessary ambient conditions (esp. temperature, which determines infrastructure cost), amenability to fault-tolerant implementations, provisions for qubit coupling and necessary architectural constraints, the speed of quantum evolution/gates, provisions for measurement, ease of manufacturing and overall scalability. There"s more, but this is a good start.

有一些關鍵的幾個參數需要考慮：物理量子比特實體的穩定性(相干衰減時間) 和量子演化里的雜訊程度， 必要的外部條件（尤其是溫度，因為它決定了系統建造成本），容錯設計能力，量子-量子的耦合調控能力，量子門/演化的速度，測量精度，設備製造難度以及可擴展性(到多個量子比特)。還有其他問題，不過這些就夠說的了。

Superconducting qubits require serious cooling (dilution refrigerators), but otherwise can leverage advanced chip-making technologies, which is good for manufacturing and scalability. A lot is known about quantum control of such systems both in theory and in practice. They impose fewer complications on qubit coupling and communication than some other technologies. But qubit stability and overall noise remain serious challenges, as with practically every technology, especially as you try to scale to hundreds of qubits.

超導量子比特需要正兒八經的降溫（液He之類），但是這個技術可以最大化利用已有的高級晶元製造能力，對於設備製造和擴展性是有好處的。這種系統的量子調控在理論和實踐上已經知道的差不多了。這個技術對於量子比特的耦合以及整體變換比其他技術要強。比特的穩定性不夠和雜訊控制確實是相當的挑戰，當然其他技術也沒好到哪兒去，尤其是你想做幾百個比特的時候。

Topological structures have some promising properties with respect to quantum fault-tolerance, but the amount of science that needs to be done before they can be pitched as an engineering solution is enormous. In particular, their control is not understood well, AFAIK. Much of engineering is completely missing here.

拓撲材料應該很有前途，尤其是在量子容錯方面，可惜關於這個材料原理我們還沒怎麼全部搞明白哪，別提工程製造了，物理這方面的問題就夠多了（加：灌水好領域啊，養活了大批物理學家哈哈）。簡而言之，工程方面，這裡很缺失。

I know little about diamond NV centers, but my understanding is that they offer greater qubit stability, and at higher temperatures than superconductors, whereas qubit coupling can be very restricted and that can slow things down at the large scale. There are fewer opportunities to leverage industry-standard technologies and infrastructures, compared to superconducting qubits, which can be a problem when trying to scale to hundreds of qubits. In particular, implementing quantum fault-tolerance would be very expensive.

我對於金剛石-氮空心系統知道的不多，不過我理解的是這個系統有更強的比特穩定性（也就是退相干時間更長），在比超導溫度高得多的溫度下可以工作。然而這種量子-量子的耦合作用很受限制（指的是哈密頓量的形式應該很受限，所以得到一個量子門很不容易），會在大尺度上把量子計算速度降下來。這個系統可以到達工業級別製造量子計算機的機會更少，和超導比起來擴展到幾百個比特是很大的問題。特別是容錯設計的代價昂貴。

In the short run, superconductors seem the most "doable", which is why D-Wave is working in that direction. Topological materials are scientifically interesting, but their practical promise is anyone"s guess at this point. From the scaling perspective, NV centers seem to suffer from similar obstacles to scalability as conventional VLSI (the dominance of interconnect), and I haven"t seen compelling solutions to these problems.

短期來看，超導應該是最有前途的，這也是為什麼D-Wave在這方向死磕的原因。拓樸材料在科學上很有趣，但是它的實際用處我們還無從得知。從可擴展性這個角度來看，NV 系統在擴展性上有問題，就想傳統的集成電路一樣，我還沒有看到可以解決這個問題的途徑。

As far as measurement goes, I"ve seen some serious applied physicists working on technologies for which they don"t really know how to do q. measurement in practice. This is less of a problem for superconductors and more of a problem for topological materials.

在測量這個方面來看，我見過一些應用物理學家的工作，不過他們似乎也不知道該怎麼做量子測量。這在超導裡面問題小一點，在拓樸材料問題要大一些。

If you are serious about comparing different technologies for near-term implementations, you need to look at concrete numbers. Compare the reliability of qubits in a particular technology to the best known (estimate of) fault-tolerance threshold, check the costs of building and maintaining the largest devices possible today, as well as the time required to perform unitary evolutions and measurements, and also their fidelities.

如果你真的要比較這些最新技術的話，你需要看看具體的效能數據。比較量子比特的可實現數量，容錯的上限，建造和保養這些設備的開銷，以及做幺正量子變換以及測量的時間花費，還有測量結果的可信度。

At a "meta" level, a major positive aspect for superconducting qubits is that many of the people involved in the engineering effort know what they are doing and thought carefully about many aspects of the technology. I am not saying they have the solutions, but at least they know what the challenges are and have enough ideas to be optimistic. In comparison, research in topological materials primarily qualifies as basic/fundamental research - which some people value more than engineering, but that makes any engineering successes less likely in the near future. Basically, it"s a matter of faith. Research in NV vacancies seems somewhere in between those two cases.

形而上學的說，超導比特的一個主要的積極意義是很多在這個領域的工程師知道他們到底乾的是什麼，而且他們也很用心的思考過這個技術的方方面面。我沒說他們已經搞出來了，但是至少他們知道問題在哪兒，挑戰在哪兒，這樣我們才會有信心樂觀下去。相比較而言，拓撲材料的研究實際上是基礎研究--很多人覺得這個比工程牛逼的多，但是目前在工程上的任何成功可能性都不大,其實這就是個信仰問題。NV的研究似乎在上兩者的中間地帶。

這個消息值得慶祝，但應當關注國外，IBM計劃2017年實現50個量子比特的系統，順利的話今年就能完成。國外2015年就實現9個量子比特的系統了，2年沒動靜不覺得奇怪嗎？

http://www.linuxidc.com/Linux/2017-03/141515.htm

20170518更新，ibm造出兩台分別為16和17位量子比特的通用量子計算機。

http://www.leiphone.com/news/201705/4Q6U7rxled1sv7jd.html