The challenge of quantum memory

Quantum computers are considered by many to be the new “holy grail” of physics. Does this mean that like the original grail, when attained they will bestow god-like powers upon their owner? Perhaps it means that the quest for a quantum computer is as likely to succeed as the medieval quests for the Holy Grail? The jury is still out.


At the heart of an ordinary, classical computer lies its memory which is formed of ”bits”. A classical bit is the smallest unit of information,  carrying a single binary number, either 0 or 1. These bits are combined and processed using logical gates to produce all the amazing things computers can do today. Logic, as any Sherlock Holmes fan will tell you, is infallible and permanent. Rule out the impossible and you’re left with the solution. So when the only two possibilities are 0 and 1, if the bit is not 0, it must be 1. Hence classical computers are fully deterministic: given the same input, they will inevitably produce the same output. (Unless there is a serious bug somewhere, but let us assume the computer is working properly).

Quantum mechanics, however, does not ascribe to this deterministic world view. Apart from very specific and ideal situations, generally speaking, quantum mechanics deals with probabilities rather than certainties. The same way we cannot know for certain what will be the outcome of a dice roll, we cannot fully predict the state of a quantum system. A fair dice roll means that the probability to roll a 6 is 1/6 (as is the probability to roll any other number – thus making the dice fair). If our dice is loaded so that 6 is twice more likely than all the other numbers, then the probability to roll a 6 is no longer 1/6 but rather 2/7. Regardless, we cannot know for certain what the dice will show until we roll it, loaded or not. Accordingly, quantum memory replaces the ”bit” with its quantum mechanical counterpart, the ”qbit”. Instead of just 0 or 1, the qbit is allowed to be both 0 and 1, with different probability, meaning that when checked, it has a certain likelihood to be found to be 0 and a certain likelihood to be found to be 1. That doesn’t sound very useful: why would we want a computer that can’t decide if its bits are 0 or 1 ?

The answer is that though a single roll is similar to a roll of a dice, repeating the dice roll many times will give a distinct answer (eg. ”six is twice more likely”). In short: whereas a single quantum 1measurement is random, the statistics of many such measurements are predictable. Which brings us back to the original, ”classical” computer. So why bother at all with quantum memory if we eventually seek out the same predictability we have already achieved so well in its classical counterpart?

The main advantage that quantum computations offer is the possibility to create entangled states. One can think of two entangled qbits as two ends of a piece of string. They are not independent –  tugging one end will affect the other. The amazing thing is that the two qbits need not be physically connected and can be potentially very far apart from each other. This is a difficult concept to stomach, indeed Einstein himself had his misgivings. Yet quantum entanglement is real and could potentially be used for a huge and growing list of near-magical technologies: impossible-to-eavesdrop communication, unbreakable codes, unforgeable money, to name a few. Some, such as the prominent physicist and mathematician Roger Penrose, even go as far as to suggest that mimicking quantum processes in the brain is key to developing artificial intelligence. So if quantum memory and computation has such potential, what is holding us back?

The most serious obstacle we face is the problem of decoherence. Decoherence can be thought of as cutting the piece of string in half, thus disconnecting the two ends. The quantum state – the entanglement of two (or more) qbits – is very sensitive to outside noise. Such noise can come in many forms: eg. ambient temperature, mechanical vibrations, stray magnetic fields. Temperature especially, causes rapid decay of this unique quantum property. Over the years, many attempts have been made to insulate quantum circuits from the outside environment; none have been successful enough to provide a truly useful quantum memory. Indeed, the difficulty is not merely that of having better technology but is a fundamental issue that is unlikely to be solved by incremental technological advances. What we require is a scientific break-though, a novel approach which will allow us to ”cheat” nature and preserve the quantum state well enough to be able to carry out effective computations. So are there any ideas how to solve this challenge ?

One candidate approach involves ”braiding” the quantum state. Braiding is very much the equivalent of tying a knot in our piece of string. Every child knows that a knot can be unravelled only by very specific action; it is protected from random tugs and pulls. A braided quantum state is similar – the idea is that the fragile quantum state is not kept in a specific place but is rather kept in the knot (which involves the whole string) and is ”protected” from random tugs. This property is called topological protection and is the subject of much research in recent years. Although the idea is appealing, there are still significant difficulties in making viable quantum memories and circuits which rely on topological protection.

Can it be done? How to do it ? Nobody yet knows. This is the  challenge we face today.


Scientific Playground