# Technology Focus: Will Quantum Computing transform optimisation of the rail network?

**Quantum Computing uses principles of quantum mechanics to perform multiple calculations in parallel. It has the potential to vastly increase the computational speed of particular processes.**

### Latest update: April 2019

## What is Quantum Computing?

Quantum Computing uses **principles of quantum mechanics**, such as **superposition** and **entanglement**, to perform calculations. Classical computers encode data in binary digits known as bits. Bits are able to hold a value of either 1 or 0 (on or off). Quantum computers encode data with **quantum-mechanical systems** **known as qubits** (quantum bits). Due to the principle of superposition, qubits are able to hold a value of either 1, 0 or some combination of the two. As a result, a **set of n qubits has a probability of being every single possible combination of bits simultaneously** (2n combinations). For example, 2 qubits can represent 4 times as many values as 2 classical bits. Additionally, the phenomenon of quantum entanglement links the states of qubits together (qubit states cannot be described independent of one another), which allows for improved speed in quantum algorithms. These properties mean that quantum computers are able to perform **multiple calculations in parallel** and vastly speed up certain computational processes. Although multiple calculations can be performed in parallel, only one possible result can be measured. As a result, only certain types of algorithm can be used to produce an answer that is significant. **Quantum computing will only be useful for some applications**; however, it may also be able to solve problems that classical computers cannot. The hypothetical point at which this may occur has been called ‘**Quantum Supremacy**’.

## What is the current state of R&D?

Broadly, there are two approaches to quantum computing, though they are not completely separate from one another:

**Quantum Annealing****Universal Quantum Computing**

**Quantum annealers** can be used to find a solution for **optimisation problems** and **sampling problems**. Qubits are manipulated to represent optimisation problems as energy minimisation problems, where the lowest energy state represents the best solution. The system naturally follows quantum principles to reach its lowest energy state (analogous to a ball rolling down a hill), hence finding the solution. This process can also be used to sample from many low-energy states to define energy landscapes (useful for Machine Learning applications). Universal quantum computers use quantum logic gates to perform calculations specified by the developer, similar to classical computing. Because of this, **universal quantum computers** can perform a wider variety of computing tasks. Universal quantum computers are more difficult to build than annealers but offer significant advantages.

D-Wave Systems have commercially released several quantum annealing systems. The latest system currently available is the D-Wave 2000QTM which has 2048 qubits, valued at $15 million USD. D-Wave Systems announced their PegasusTM system with over 5000 qubits (Feb 2019). It remains unclear whether quantum annealers will lead to an increase in computational speed over traditional computing. With regard to universal quantum computing systems, **IBM** have released the IBM Q System OneTM (Jan 2019) – a **20-qubit quantum computer** integrating hardware, firmware and also classical computation methods (to access the cloud). This is the world’s first commercially available quantum computer. Chips with up to **128 qubits** have been developed (Rigetti) but not released in a full-scale commercial computing system.

As quantum computing methods are very different from classical computing methods, a lot of research is focused on discovering applications. Organisations with the aim of creating quantum systems integrating hardware and software have given access to their systems through the cloud or released software developer kits, such as the Microsoft Quantum Development Kit. This is to encourage research into quantum applications and algorithm development.

## What are the potential applications of Quantum Computing?

It has been suggested that **quantum annealers** will solve optimisation problems within financial modelling, system optimisation, vehicle route management and healthcare. Defence contractor Lockheed Martin is evaluating if the D-Wave 2X can be used to verify and validate complex computer and information systems. Additionally, Volkswagen have used the D-Wave system to develop an **intelligent traffic management system** for transport service providers. There are proposals to test the system in Barcelona. Quantum annealers are also proposed for **machine learning** to train **neural networks** and **image recognition systems**. The Quantum Artificial Intelligence Lab is using a D-Wave 2X system to investigate this.

**Universal quantum computers** would be able to simulate annealing well and could perform additional computing tasks such as **Shor’s or Grover’s algorithm**. They may solve optimisation problems through different methods more efficiently, without having to map the problem into a state suitable for quantum annealing. If universal quantum computers are developed with enough qubits, they could replace annealers. Universal quantum computers can also be applied to problems that quantum annealers would be unsuitable. One application of interest is modelling complex molecular and chemical interactions, aiding in the **discovery of new materials and medicines**. Chemical interactions are naturally quantum-mechanical. Quantum computers could potentially be used to simulate them more easily than conventional computing.

## What uncertainties remain?

The greatest obstacle of quantum computers is the **coherence length** of qubits. This is the length of time a qubit can maintain quantum properties before they collapse to classical states. Longer coherence times allow for more operations to be performed on the qubit. Qubits are extremely sensitive. Environmental noise, such as temperature fluctuations, causes qubits to **decohere**, creating calculation errors. This may be analogous to wind blowing over a spinning coin. As qubits are small in scale, the slightest bit of interaction with the environment causes decoherence. Quantum computing developers are investigating quantum error correction to achieve fault-tolerant computation.

Decoherence also means it is **difficult to scale the number of qubits up within a system**. Quantum computing systems must be run at a temperature close to absolute zero and in extremely isolated environments to extend coherence time of qubits. Some researchers are exploring qubits that do not need to be extremely cold but the need for qubits to be extremely isolated results in extremely large and complex systems with high costs. The IBM Q System One has dimensions of 2.5 x 2.5 x 2.8 meters.

## How will it impact the rail industry?

Since the potential uses of a quantum computer are still being evaluated, there are many potential ways it could impact the railway. Quantum computers could run complex optimisation tasks with regards to infrastructure maintenance, scheduling and logistics. This could aid in timetabling and real time disruption recovery, hence minimising disruptions to train service whilst **optimising energy use**.

The potentially high computational power could be used to tackle big data analytics – **improving asset maintenance** or providing a **more customised customer experience**. It could also be used to develop artificial intelligence, facial recognition systems, etcetera. A universal quantum computer could also aid in the discovery of **new advanced materials **to be used in rolling stock of infrastructure.

Quantum computers could render some existing cryptography methods useless through use of Shor’s algorithm. This may leave the rail industry’s digital systems vulnerable to attack.

## What should the rail industry do?

Stakeholders could **make use of cloud quantum computing** to investigate methods to apply quantum algorithms to optimising the rail network. This would also build a skill base of workers who are familiar with quantum algorithm programming. Train operating companies could partner with other transport companies currently developing **traffic optimisation algorithms** to assist with this. Rail research groups could fund research opportunities studying advanced material design and algorithm development for traffic management systems. The rail industry could also invest in **encryption methods** that would not be vulnerable to quantum computing algorithms. Engagement with the quantum computing industry early on may allow the rail industry to develop vital skills for when quantum computing becomes a reality and help combat the quantum skills gap.