2013年3月25日 星期一

Testing a New Class of Speedy Computer

量子計算機將帶來速度革命

Kim Stallknecht for The New York Times
洛克希德·馬丁公司購買了一款D-Wave量子計算機,計劃把它升級到商用規模。

不列顛哥倫比亞溫哥華——我們的數字時代是由位元構建的。一個位元就是一個0和1的特定狀態,由此產生出所有現代計算機編碼。
但是現在,一家大型軍工企業即將推出一種強大的新型商用電腦,它將把計算機技術帶進量子力學的神奇亞原子世界。那是一個奇妙的微觀世界,在那裡,常識邏輯好像都不再適用。1可以是1,也可以是1和0,或者同時處於介乎二者之間的任意狀態。
這聽起來可能有些荒誕,尤其是對那些習慣於非是即否的傳統計算技術的人。但是學院研究員和微軟(Microsoft)、IBM、惠普(Hewlett-Packard)等企業的科學家的確是在開發量子計算機。
洛克希德·馬丁公司(Lockheed Martin)兩年前曾經購買了加拿大D-Wave系統公司(D-Wave Systems)的一款早期量子計算機。如今,它對這種技術已經有了足夠的信心,決定把量子計算機升級為商用型,從而成為首家經營量子計算機業務的公司。
也有人並未信服,說D-Wave公司還不能對外界的科學家證明,它已經解決了量子計算的眾多難題。
但如果新的計算機像洛克希德和D-Wave期望的那樣運轉,那麼,它將可以極大地提升哪怕是如今最強大的計算機系統的運算能力,其處理科技和商務問題的速度,會比今天快數百萬倍。
洛克希德公司的技術主管雷·約翰遜(Ray Johnson)說,公司將利用量子計算機開發和測試複雜的雷達、航天和航空器系統。這種計算機將可以完成許多任務,比如即時得出一個有幾百萬行代碼的衛 星網絡管理軟件遇到太陽黑子爆發或核爆炸衝擊波時會作何反應。這類任務用今天的技術,即使可以做到,也要幾周的時間。
“這將是一場革命,就像早期計算機革命一樣,”他說,“它會改變我們對計算機的理解。”還有很多機構會發現D-Wave量子計算機的新用途。研究癌 症的科學家認為,可以用它快速處理海量基因數據。量子計算還可能用於研究人類基因組中蛋白質的行為。這是一項比基因組測序更艱巨的工作。谷歌 (Google)的研究人員也開始與D-Wave合作,利用量子計算機識別汽車和地面標誌物,這將是管理自動駕駛交通工具的關鍵一步。
量子計算比傳統計算技術快得多,是因為亞原子粒子的神奇特性。自從第一台計算機問世以來,我們用0和1的精確狀態表達數據;而亞原子粒子則同時處於 多種狀態,量子計算利用的就是亞原子的這種特性。粒子之間不同的互動關係也可以同時並存。所有這些可能的粒子狀態可以被縮小範圍,在近乎無限多的可能性中 得出一個最佳結果,從而實現快速解決某些類型的計算問題。
D-Wave公司總部在溫哥華,已有12年的歷史。目前,公司得到了多方投資,包括亞馬遜(亞馬遜運行着世界最龐大的計算機系統之一)的創始人傑 夫·貝佐斯(Jeff Bezos)、投資銀行高盛集團(Goldman Sachs)和投資公司In-Q-Tel,一家與中央情報局(Central Intelligence Agency)等政府部門關係密切的公司。
D-Wave首席執行官維恩·布勞內爾(Vern Brownell)說:“我們現在做的,是一種平行於過去70年來的計算機技術的開發工作。”
布勞內爾於2009年來到D-Wave。2000年以前,他一直是高盛的技術主管。“那時候,我們有5萬台服務器,僅僅是為了進行模擬運算”來決定交易策略,他說,“現在的運算規模肯定比那時更大了。但是我們將可以只用一台機器,成本也低很多。”
研究人員致力於開發量子計算技術已經30多年了。但是要實現這個目標一直很困難。他們的思路是利用物質在量子狀態下的一種特性。這種狀態被稱作疊加態。在疊加態下,量子計算機的基本單位,量子位元,可以同時承載極大量的數值。
科學家有多種方法可以創造達到疊加態和另一種被稱作量子糾纏的狀態所需的條件。量子疊加和量子糾纏都是量子計算所必需的。研究人員已經做到了使磁場中的離子懸浮,捕獲光子,或者在硅中操縱磷原子。
洛克希德購買的D-Wave計算機使用的數學方法和競爭對手有所不同。在D-Wave計算機的系統裡面,由極細的超導線網構成的量子處理器被冷卻到接近絕對零度。然後,把一組數學公式加載到超導線網上,對處理器進行編程。
然後,處理器檢索幾乎無限多的可能性,判斷出構建那兩種粒子關係所需的最少能量是多少。這個被視為最佳結果的狀態,就是問題的答案。
這種方法被稱為絕熱量子計算,很可能在計算蛋白質摺疊等領域大有可為。D-Wave的設計師還說,這種方法可能還可以用來評估複雜的金融策略或龐大的物流問題。
然而,公司的科學家還沒有發表任何科學數據,證明他們的系統運算速度比傳統的二進制計算機快。植物也在利用類似的亞原子特性,在幾微秒(百萬分之一 秒)甚至幾奈秒(十億分之一秒)的瞬間,把太陽光轉化為光合能。但是對D-Wave量子計算持批評態度的人說,那根本不是量子計算,而是一種典型的熱能行 為。
John Markoff自舊金山對本文有報道貢獻。
翻譯:黃少傑


Testing a New Class of Speedy Computer

Kim Stallknecht for The New York Times
Lockheed Martin brought a version of D-Wave’s quantum computer and plans to upgrade it to commercial scale.

VANCOUVER, British Columbia — Our digital age is all about bits, those precise ones and zeros that are the stuff of modern computer code.
But a powerful new type of computer that is about to be commercially deployed by a major American military contractor is taking computing into the strange, subatomic realm of quantum mechanics. In that infinitesimal neighborhood, common sense logic no longer seems to apply. A one can be a one, or it can be a one and a zero and everything in between — all at the same time.
It sounds preposterous, particularly to those familiar with the yes/no world of conventional computing. But academic researchers and scientists at companies like Microsoft, I.B.M. and Hewlett-Packard have been working to develop quantum computers.
Now, Lockheed Martin — which bought an early version of such a computer from the Canadian company D-Wave Systems two years ago — is confident enough in the technology to upgrade it to commercial scale, becoming the first company to use quantum computing as part of its business.
Skeptics say that D-Wave has yet to prove to outside scientists that it has solved the myriad challenges involved in quantum computation.
But if it performs as Lockheed and D-Wave expect, the design could be used to supercharge even the most powerful systems, solving some science and business problems millions of times faster than can be done today.
Ray Johnson, Lockheed’s chief technical officer, said his company would use the quantum computer to create and test complex radar, space and aircraft systems. It could be possible, for example, to tell instantly how the millions of lines of software running a network of satellites would react to a solar burst or a pulse from a nuclear explosion — something that can now take weeks, if ever, to determine.
“This is a revolution not unlike the early days of computing,” he said. “It is a transformation in the way computers are thought about.” Many others could find applications for D-Wave’s computers. Cancer researchers see a potential to move rapidly through vast amounts of genetic data. The technology could also be used to determine the behavior of proteins in the human genome, a bigger and tougher problem than sequencing the genome. Researchers at Google have worked with D-Wave on using quantum computers to recognize cars and landmarks, a critical step in managing self-driving vehicles.
Quantum computing is so much faster than traditional computing because of the unusual properties of particles at the smallest level. Instead of the precision of ones and zeros that have been used to represent data since the earliest days of computers, quantum computing relies on the fact that subatomic particles inhabit a range of states. Different relationships among the particles may coexist, as well. Those probable states can be narrowed to determine an optimal outcome among a near-infinitude of possibilities, which allows certain types of problems to be solved rapidly.
D-Wave, a 12-year-old company based in Vancouver, has received investments from Jeff Bezos, the founder of Amazon.com, which operates one of the world’s largest computer systems, as well as from the investment bank Goldman Sachs and from In-Q-Tel, an investment firm with close ties to the Central Intelligence Agency and other government agencies.
“What we’re doing is a parallel development to the kind of computing we’ve had for the past 70 years,” said Vern Brownell, D-Wave’s chief executive.
Mr. Brownell, who joined D-Wave in 2009, was until 2000 the chief technical officer at Goldman Sachs. “In those days, we had 50,000 servers just doing simulations” to figure out trading strategies, he said. “I’m sure there is a lot more than that now, but we’ll be able to do that with one machine, for far less money.”
Quantum computing has been a goal of researchers for more than three decades, but it has proved remarkably difficult to achieve. The idea has been to exploit a property of matter in a quantum state known as superposition, which makes it possible for the basic elements of a quantum computer, known as qubits, to hold a vast array of values simultaneously.
There are a variety of ways scientists create the conditions needed to achieve superposition as well as a second quantum state known as entanglement, which are both necessary for quantum computing. Researchers have suspended ions in magnetic fields, trapped photons or manipulated phosphorus atoms in silicon.
The D-Wave computer that Lockheed has bought uses a different mathematical approach than competing efforts. In the D-Wave system, a quantum computing processor, made from a lattice of tiny superconducting wires, is chilled close to absolute zero. It is then programmed by loading a set of mathematical equations into the lattice.
The processor then moves through a near-infinity of possibilities to determine the lowest energy required to form those relationships. That state, seen as the optimal outcome, is the answer.
The approach, which is known as adiabatic quantum computing, has been shown to have promise in applications like calculating protein folding, and D-Wave’s designers said it could potentially be used to evaluate complicated financial strategies or vast logistics problems.
However, the company’s scientists have not yet published scientific data showing that the system computes faster than today’s conventional binary computers. While similar subatomic properties are used by plants to turn sunlight into photosynthetic energy in a few million-billionths of a second, critics of D-Wave’s method say it is not quantum computing at all, but a form of standard thermal behavior.
John Markoff contributed reporting from San Francisco.

沒有留言: