Programming a computer is usually considered to be a fairly difficult process as it involves hours of coding, not to mention the strenuous work of debugging, testing, and documenting in order to ensure that it works properly.
However, things are in fact much tougher for a team of physicists from the Harvard-MIT Center for Ultracold Atoms and the California Institute of Technology.
Working in a Harvard Physics Department lab, a team of researchers headed by Harvard Professors Mikhail Lukin and Markus Greiner and Massachusetts Institute of Technology Professor Vladan Vuletic produced a distinct type of quantum computer, called a quantum simulator, that is in fact programmed by capturing super-cooled rubidium atoms with the help of lasers and then arranging them in a particular order. This is then followed by allowing quantum mechanics to do the required calculations.
It could be possible to use the system for shedding light on a host of complex quantum processes, which also includes the connection between material properties and quantum mechanics, and to examine new phases of matter and resolve difficult real-world optimization problems. A description of the system is available in a Nov. 30th paper featured in the journal Nature.
Researchers state that the combination of the system’s high degree of quantum coherence and large size make it a vital achievement. With over 50 coherent qubits, this is considered to be one of the largest quantum systems ever developed with individual measurement and assembly.
In the same issue of Nature, a team of researchers from the Joint Quantum Institute at the University of Maryland defined a correspondingly sized system of cold charged ions, also controlled with the helps of lasers. Considered together, these complimentary improvements establish a vital step toward large-scale quantum machines.
“Everything happens in a small vacuum chamber where we have a very dilute vapor of atoms which are cooled close to absolute zero,” Lukin said. “When we focus about 100 laser beams through this cloud, each of them acts like a trap. The beams are so tightly focused, they can either grab one atom or zero; they can’t grab two. And that’s when the fun starts.”
Researchers can use a microscope to take images of the captured atoms in real time, and they can then arrange them in random patterns for input.
“We assemble them in a way that’s very controlled,” said Ahmed Omran, a postdoctoral fellow in Lukin’s lab and a co-author of the paper. “Starting with a random pattern, we decide which trap needs to go where to arrange them into desired clusters.”
As researchers start feeding energy into the system, the atoms then start to interact with each other. Those interactions, according to Lukin, provide the system its quantum nature.
“We make the atoms interact, and that’s really what’s performing the computation,” Omran said. “In essence, as we excite the system with laser light, it self-organizes. It’s not that we say this atom has to be a one or a zero — we could do that easily just by throwing light on the atoms — but what we do is allow the atoms to perform the computation for us, and then we measure the results.”
According to Lukin and colleagues, those results could shed light on complex quantum mechanical phenomena that are just impossible to model using standard computers.
“If you have an abstract model where a certain number of particles are interacting with each other in a certain way, the question is why don’t we just sit down at a computer and simulate it that way?” asked Ph.D. student Alexander Keesling, another co-author. “The reason is because these interactions are quantum mechanical in nature. If you try to simulate these systems on a computer, you’re restricted to very small system sizes, and the number of parameters are limited.
“If you make systems larger and larger, very quickly you will run out of memory and computing power to simulate it on a classical computer,” he added. “The way around that is to actually build the problem with particles that follow the same rules as the system you’re simulating. That’s why we call this a quantum simulator.”
Even though it is possible to employ classical computers in order to model small quantum systems, the simulator produced by Lukin and colleagues makes use of 51 qubits, enabling it to be almost impossible to replicate using standard computing techniques.
“It is important that we can start by simulating small systems using our machine,” he said. “So we are able to show those results are correct … until we get to the larger systems, because there is no simple comparison we can make.”
“When we start off, all the atoms are in a classical state. And when we read out at the end, we obtain a string of classical bits, zeros, and ones,” said Hannes Bernien, another postdoctoral fellow in Lukin’s lab, and also a co-author. “But in order to get from the start to the end, they have to go through the complex quantum mechanical state. If you have a substantial error rate, the quantum mechanical state will collapse.”
According to Bernien, it is this coherent quantum state that permits the system to operate as a simulator, and it also makes the machine a potentially valuable tool that will help gain insight into complex quantum phenomena and ultimately perform useful calculations. The system already permits researchers to attain exclusive insights into transformations between varied types of quantum phases, known as quantum phase transitions. It may also enable shedding light on exotic and new forms of matter, Lukin said.
“Normally, when you talk about phases of matter, you talk about matter being in equilibrium,” he said. “But some very interesting new states of matter may occur far away from equilibrium … and there are many possibilities for that in the quantum domain. This is a completely new frontier.”
Lukin said, the researchers have already seen evidence of these states. In one of the first experiments performed with the new system, the researchers discovered a coherent non-equilibrium state that continued to be stable for an amazingly longer period of time.
“Quantum computers will be used to realize and study such non-equilibrium states of matter in the coming years,” he said. “Another intriguing direction involves solving complex optimization problems. It turns out one can encode some very complicated problems by programming atom locations and interactions between them. In such systems, some proposed quantum algorithms could potentially outperform classical machines. It’s not yet clear whether they will or not, because we just can’t test them classically. But we are on the verge of entering the regime where we can test them on the fully quantum machines containing over 100 controlled qubits. Scientifically, this is really exciting.”
Other co-authors of the study include visiting scientist Sylvain Schwartz, Harvard graduate students Harry Levine and Soonwon Choi, research associate Alexander S. Zibrov, and Professor Manuel Endres.
The National Science Foundation, the Center for Ultracold Atoms, the Army Research Office, and the Vannevar Bush Faculty Fellowship funded the research.