STEP into and you step out of this world. Not quite literally, but there鈥檚 a lot within its walls that you won鈥檛 find outside its doors. On one day you might find his team playing with a quark-gluon plasma, a searingly hot substance that only naturally existed in the universe鈥檚 first split second. Drop by the next week, and they will be messing about on the surface of a neutron star. Sometimes they鈥檒l even be creating stuff that has never been made anywhere in the universe.
But there is more to this than physicists at play. The stuffs that the researchers at the Massachusetts Institute of Technology are studying have something in common, something they share with a lot of everyday materials: they are fiendishly, inscrutably, impossibly complex. Swipe the touchscreen of your mobile phone and, as an electrical pulse passes from your finger and brings it to life, a random mosaic of chemical bonds breaks and reforms. How does that work in detail? Or how exactly does that magnet keep your shopping list stuck to the refrigerator? Or what makes some materials lose all electrical resistance below a certain temperature and become superconductors? Get that to happen at close to room temperature, and our world would be revolutionised: we would have a way of storing energy for ever, for free.
If only. Trying to model the myriad effects and influences in all these situations soon leaves us floundering. 鈥淲e know the ingredients: they鈥檙e super-easy to write down,鈥� says Zwierlein. 鈥淏ut give that to a computer and it will explode.鈥�
Advertisement
That鈥檚 why his group and a few others dotted around the globe are working on plan B. They hope to get straight to the heart of the matter 鈥� by unleashing the full power of the quantum simulator.
Quantum simulation is an idea first floated in a by the revered US physicist Richard Feynman. He pointed out that conventional, classical computers are a window on the world with a restricted view. They demand ruthless logic and concrete instructions: if x, then y; move this binary digit here; add these two numbers; subtract; print, that kind of thing. Quantum physics, on the other hand, is all maybes and probablys. Particles have undefined locations or spins, and can be spookily influenced by a legion of other ghost particles around them.
How can we model such countless effects and avoid meltdown? Feynman answered in typically ebullient fashion: 鈥淣ature isn鈥檛 classical, dammit, and if you want to make a simulation of nature, you鈥檇 better make it quantum mechanical.鈥� Three decades later, though, we鈥檙e not particularly far on in our efforts to build a general computer that works according to the precepts of quantum physics.
But to simulate some quantum problems it turns out that you don鈥檛 actually need anything we would recognise as a conventional computer. In fact, you don鈥檛 need to actually calculate much at all. That鈥檚 because the complex rules and influences of quantum physics work in the same way for all particles 鈥� for atoms, electrons, whatever. To unlock the secrets of an intransigent quantum system, all you need is to make a more pliant, physical mimic.
Atomic egg boxes
There are several candidates: photons of light, or trapped ions, for example. Perhaps the most effective quantum simulator, though, is an ultracold atomic gas. A criss-cross of laser beams confines a million or so atoms in a landscape of neighbouring hollows that is rather like a quantum egg box. The action of the lasers cools down the atoms鈥� normal thermal jiggling so they have a temperature just billionths of a degree above absolute zero 鈥� cold enough for their behaviour to be cleanly described using the laws of quantum mechanics (see diagram).
It鈥檚 a fiddly technique that we鈥檙e just beginning to master. 鈥淭here鈥檚 an explosion of people with the ability to locate, measure and control individual atoms,鈥� says of the National Institute of Standards and Technology in Gaithersburg, Maryland. Once the atoms are becalmed, you can use further pulses of light to change their environment and behaviour, for example pushing some of them to sit in a 鈥渟uperposition鈥� between two neighbouring quantum states. 鈥淵ou can tickle them in all kinds of ways,鈥� says Zwierlein.
The gentle laser massaging allows you to conjure phenomena beyond a classical computer鈥檚 ken 鈥� such as what happens in some extreme astrophysical environments. 鈥淲ith the right combination of lasers you can create scenarios like those in neutron stars,鈥� says , who runs a quantum simulation lab at the University of Birmingham, UK. Neutron stars pack in more than the mass of our sun in a ball just 10 to 15 kilometres across. They form when the cores of certain stars collapse in on themselves during their final supernova explosion. The remnant鈥檚 super-strong gravity and magnetic fields make for extremely complex interactions 鈥� far too complex for any conventional computer to model in detail.
Last February, Zwierlein鈥檚 group recreated a neutron star within a gas of ultracold lithium-6 atoms. Thanks to the value of their quantum-mechanical spin, these atoms, like neutrons, belong to the class of particles known as fermions. Quantum rules forbid fermions from getting too close to each other. The atomic simulation showed how this prevented the atoms condensing beyond a certain point 鈥� reproducing how the collapse of a neutron star stops beyond a certain density ().
We can get similar lab-based insights on the quark-gluon plasma, the hugely energetic state of matter that existed shortly after the big bang and which gave rise to today鈥檚 particles. in New York state, and latterly at CERN鈥檚 Large Hadron Collider near Geneva, Switzerland, have succeeded in recreating the real thing by bashing together ions of heavy atoms such as lead with such force that they disintegrate.
Zwierlein and his team鈥檚 鈥淟ittle Fermi Collider鈥� aims to create exactly the same physics, but using more manageable particles at a fraction of the temperature and price. Their first results, published in 2011, were encouraging. Magnetic forces drove together two clouds, each of around 100,000 lithium atoms. They first repelled, then repeatedly bounced off each other and finally merged 鈥� at an extremely slow rate precisely matching that predicted by theoretical models of the quark-gluon plasma ().
Hunting the Higgs
And how about the grandest particle prize of all, the Higgs boson? In experiments last year of the Max Planck Institute for Quantum Optics in Garching, Germany, and his team observed a wave of energy in a cloud of ultracold rubidium-87 atoms that exactly mimicked the mathematics of a Higgs boson (). If that sounds a little weak, bear in mind that no one at CERN has ever actually seen a Higgs boson either, just the energy signature of particles thought to derive from it. In effect, Bloch鈥檚 team had spotted an ultracold analogue of the elusive particle months before the real thing was spotted 鈥� probably 鈥� at CERN in July last year.
These first successes suggest that quantum simulation really does, as Feynman hoped, provide a viable way of teasing out the behaviour of any quantum system, however complex. 鈥淨uantum simulation is beginning to come of age,鈥� says Spielman. And that means it might get to grips with some far less esoteric quantum problems.
One of those is magnetism. Magnetism arises when the spins of atoms inside crystals of metals such as iron align in particular ways, but the details are complex, hindering the search for new magnetic materials. at the University of Maryland, Baltimore, are working on simulating the transition to a magnetic state using trapped ions. Their first experiments have not involved more than a handful of ions and have not gone beyond what is calculable with a classical computer 鈥� but with more ions much more should be possible ().
With luck the lessons can be applied to other practical quantum systems 鈥� that pattern of bond formation in your touchscreen, for example. The interest in magnetism is just a stepping stone to something bigger, too. That鈥檚 because of clues contained in a set of equations known as the Hubbard model. This describes how materials move between insulating and conducting behaviour, and it suggests a key to the enigma of high-temperature superconductivity. 鈥淭he model predicts a magnetic phase that is supposed to kick in before the superconductivity sets in,鈥� says Bongs.
Superconductors are materials that lose all resistance to electricity when cooled below a certain temperature. Materials that pull off this trick close to absolute zero have been known about for a century, and pretty well understood for half of that. But materials that do the same thing at higher temperatures remain a mystery. At the moment the record stands at 138 kelvin, slightly under halfway between absolute zero and room temperature.
The idea is that, by setting a lattice of atoms to behave according to the Hubbard model鈥檚 equations, we can then twiddle the knobs of the simulation 鈥� by altering the magnetic field, how densely the atoms are packed, the ratio of atoms with different spins, and so on 鈥� and look for behaviours that resemble the onset of magnetism and superconductivity. Do that, and in theory it鈥檚 just a case of fiddling about with the same parameters to show us how we might produce the same effects at higher temperatures.
In practice, it鈥檚 not that simple. For a start, says Bloch鈥檚 colleague , the temperature of our existing ultracold simulators is still too high. That seems harsh for an environment that, at a few billionths of a degree above absolute zero, is the most frigid we know. This is not about absolute physical temperature, however. Rather, it is about 鈥淔ermi temperature鈥� 鈥� roughly speaking, the temperature at which half of a conducting material鈥檚 free electrons are free enough to wander about and transport energy. Every superconducting material has a set temperature when this happens, and it is usually many times higher than the temperature above which a material ceases to superconduct.
鈥淯ltracold simulators are still too hot 鈥� harsh as that seems for an environment a few billionths of a degree above absolute zero鈥�
A perfect simulation requires that the ratio of Fermi temperature to working temperature in the ultracold gas is about the same as the ratio of the material鈥檚 Fermi temperature to its superconducting temperature. To get the ratio right, the working temperature of the gas needs to drop by a further factor of 5 or so. That鈥檚 a tough ask. Lowering the temperature means the atoms must interact more. Like the neutrons of a neutron star, electrons are fermions, so they don鈥檛 like getting too up close and personal 鈥� and the same applies to a fermionic gas designed to simulate them.
That means workarounds are needed. 鈥淧eople are getting really creative,鈥� says Zwierlein. His team is currently testing an idea dreamed up by 鈥榮 team at the University of Innsbruck, Austria: mix fermions with more sociable bosons in the simulator. In such a scheme, the fermions stay bound in the egg-box lattice, while the bosons roam, interact and lose energy through collisions, reducing the overall temperature ().
Zwierlein鈥檚 gas of fermionic lithium-6 atoms also has some promising characteristics. The temperature at which the atoms begin mimicking superconductivity is a full 16 per cent of the gas鈥檚 Fermi temperature, rather than just a tiny fraction 鈥� meaning a superconductor with equivalent properties would be expected to superconduct at way above room temperature. That makes it a milestone in our quest for room temperature superconductors, says theorist of the Technical University of Munich in Germany.
Yet the true beauty of the simulator approach in the quest for new and better materials such as superconductors is its flexibility. Traditionally researchers have had to design a potentially promising material, grow it in the lab and measure its properties 鈥� only to find it is a dud. Mimic a superconductor with a gas of ultracold atoms, however, and you can scan through hundreds of slightly different atomic recipes at different simulated temperatures, all in a couple of days. 鈥淭hat鈥檚 a profound difference,鈥� says Spielman.
It also means that theorists are finally getting some real-world checks on their models. 鈥淭hey can鈥檛 just come up with a random number 鈥� for the first time, it has to be tested through experiment,鈥� says Zwierlein. Bongs is similarly enthusiastic. 鈥淐old atoms allow us to access something that鈥檚 hardly accessible in nature.鈥�
That same point makes Cirac sound a note of caution. When it comes to phenomena that have never been observed before, such as room-temperature superconductivity, we still won鈥檛 be 100 per cent sure that the quantum simulator is not sending us on a wild goose chase until we can actually make the equivalent material. 鈥淚f you knew the result, you wouldn鈥檛 need the simulator,鈥� he says.
Nevertheless, there is a whole new avenue of possibilities to explore. Feynman ended his talk on quantum simulation with the words, 鈥淏y golly it鈥檚 a wonderful problem, because it doesn鈥檛 look so easy.鈥� Three decades on, we鈥檙e under no illusions. But in Feynman鈥檚 terms, though it still doesn鈥檛 look so easy, it鈥檚 starting to look doable, dammit.
This article appeared in print under the headline 鈥淭he quantum sims鈥�
