Broker to Trading

Quantum computing vs classical computing

Classical computing is an established technology. The causal structure is well understood, which is to say using the flow of electrons to rapidly create and process logical statements that reflect the instructions of a computer program. It is, however, a technology which has improved dramatically over the years. Quantum computing is not an established technology. There is debate about whether the technology actually exists. One foundational issue is that the science it relies on, 'non-locality', is not established, in fact there are competing theories that try and explain this intriguing, counter-intuitive quantum phenomenon.

Non-locality and quantum computation

Non-locality is not, strictly speaking, a correct term, but it is useful to describe the phenomenon in a way that makes sense to humans living in a seemingly local universe. Non-locality is a feature of the quantum world where correlations occur such that it seems like event A is affecting event B instantaneously, even though they are separated by a distance. This is different from the electron flow model in classical computing, as it allows for a structure of indefinite complexity, that coalesces into a desired solution state. Instead of having to compute each logical statement, structures can be created that encompass possible solutions, without searching for them step-by-step.

Non-locality allows for a kind of depth of computation that would not be possible with classical computers. Non-locality seems counter-intuitive, but there are a number of theories which attempt to explain it. For example, Quantum Field Theory attempts to preserve locality by seeing the universe as consisting of fields, where particles are excitations of this field. The idea is that these fields allow for a correlation that means a particle can be 'affected' by another particle regardless of the distance, due to a property of the field.

The Many-Worlds Hypothesis seeks to preserve locality by allowing for each quantum event to branch into another universe, but maintaining the overall connection via a wave function that spans the multiverse.

There are other theories as well, but what this means is that there is a big unknown in terms of part of the mechanism behind the way quantum computers work, which may or may not matter. This may not be a problem, as the major apparent problem is that is it difficult to let the quantum computer compute, as it is highly sensitive to noise from the surrounding environment. However it is conceivable that the actual mechanism that underlies non-locality may affect the potential for quantum computers to compute. Quantum computers are reliant on entanglement, which uses non-locality to function, as well as superposition.

Superposition vs entanglement

Superposition and entanglement are different concepts, but both are used by quantum computers. In terms of computing, superposition is the capacity for a single bit to be in two states at the same time. This is radically different from classical computation where a bit can be in only one state at a given time: it can change to another state, and the speed at which this can happen is part of its capacity to compute.

Classical computers are founded on logic. Gates can be constructed, composed of bits in either on or off. Combinations of these gates provide a logical framework for software to program instructions (algorithms), to be executed by computer hardware. Quantum computers also have algorithms and logic, but the difference is that computations that can be executed only one step at a time on classical computers, can in theory be executed at the same time on quantum computers.

A quantum computer can be seen as a device that can compute multiple outputs at the same time. But it is necessary that the qubits encode multiple states, and the problem is that noise can change (decohere) the underlying computation, resulting in a problem that the quantum gates decohere faster than the computation. This one issue presents a major blockage in scaling up quantum computation, and there are other problems as well.

Entanglement, in terms of computation, is a way for qubits to be connected with each other. This allows many qubits to part of a computation, entangled together, without having to change each qubit, as is the case with classical computation on bits. This results in a kind of multiplier effect, from connected qubits. Entanglement is about the depth or complexity of the computations, while superposition is about parallelism. Entanglement is also subject to noise, as the same process that allows one qubit to entangle with another, also allows it to entangle with the external universe.

Bell's Theorem is a test of whether particles are entangled, where its violation demonstrated entanglement. Thus violation of Bell's Theorem can be used as a gauge as to whether a quantum computer is computing classically or using entanglement.

Case Study Example: D-Wave

There are a number of companies that are involved in quantum computing, as well as a vast array of research institutions. D-Wave is one of them. D-Wave makes products that use quantum effects to compute. Simulated annealing is a technique used in various ways, but in computing it is utilised by neural nets.

Problems can be considered as a solution space, where the lowest point represent the best solution, with the task being to get from the current state to the optimal low. When solving problems, the process used may fall into a solution that is not the best solution. Simulated annealing simulates the process of heating and cooling, to allow the system to exit these local minima and find more optimal configurations.

Quantum annealing utilises superposition to allow the search problem to be considered over many possible solutions, rather than trying one at a time. Quantum annealing uses another feature of the quantum world, tunnelling, to allow peaks that would otherwise obscure the solution to be passed through, without having to go over them. Quantum entanglement points towards correlations that make the path to the solution clearer (i.e. adding depth to the computation).

The effect of quantum computing on trading

Quantum computing is more like a set of specialized algorithms, to tackle some types of problems, but with significant hardware limitations. So quantum computing may have an effect on trading, in the future, by improving part of the computations that are involved in the trading process. But quantum computing has a long way to go.

What then are quantum computers (or at least a more robust future computer) good at? They are good at optimization problems. The example used above of quantum annealing is an optimisation process, as it seeks to find the 'best' solution in a landscape of solutions and obstacles. Optimisation is important to trading processes, for example portfolio optimisation or strategy optimisation, but it is also important for the underlying infrastructure supporting trading, for example routing algorithms in High Frequency Trading.

Quantum computers are potentially good at simulating, thus they could be useful for simulating markets, which is a most intriguing possibility. The reason they are good at simulating is because of the depth of representation available from entanglement, as well as superposition. But the practical problems apply of isolating the system to maintain entanglement and superposition in such a way as it continues to represent the problem and the path towards its solution, not the outside world.

There is also the additional issues of being able to have enough qubits to make fuller use of the potential for problem representation. Quantum computers are currently limited in the number of qubits they can use, like the way classical computer systems were limited by smaller bit sizes. As bit sizes have increased, the range and complexity of classical computing applications has also increased. As ways are found to increase the qubit count, then breakthroughs might be expected for quantum computing applications. For now, traders will have to make do with the technology available at trading providers. This technology is already cutting edge, as it involves the use of technology to leverage improvements in speed, but it is not yet technology utilising quantum computing. But the first signs of it, may be in the infrastructure supporting order routing, at some stage in the future, or even in intelligent chatbots.