×

Registration

Profile Informations

Login Details

or login

First name is required!
Last name is required!
First name is not valid!
Last name is not valid!
This is not an email address!
Email address is required!
This email is already registered!
Password is required!
Enter a valid password!
Please enter 6 or more characters!
Please enter 16 or less characters!
Passwords are not same!
Terms and Conditions are required!
Email or Password is wrong!

Barclays demonstrates proof-of-concept quantum clearing algorithm

A team of researchers from Barclays’ chief technology and innovation office, in collaboration with IBM, have published a paper describing a proof-of-concept quantum optimised application.

The paper, Quantum algorithms for mixed binary optimization, published earlier in October, discussed an algorithm the team had applied to transaction settlement for ensuring the integrity of trades in securities.

The paper described the problem of transaction settlement as one that is difficult to optimise. This is because it requires a combination of both the legal constraints that must be satisfied when settling delivery-versus-payment transactions and the additional optionality introduced by what the authors describe as “collateralising assets and utilising credit facilities”.

Describing the transaction settlement process in an interview with Computer Weekly, Lee Braine, director of research and engineering at Barclays’ chief technology and innovation office, said: “I submit my trade details; you submit yours. They are then matched, which becomes a transaction that then sits in a queue.”

Braine said settlement can occur on a transaction-by-transaction basis or the securities trades can be pushed into a queue and processed in batches during a batch window. “The goal is to settle as many trades as possible during the batch window,” he said.

This is a computationally complex problem, said Braine, because of traders’ ability to tap into funds before their transaction has been cleared. “I can settle if there is funding available or if I can get access to funds in the form of a credit collateral facility,” he said.

The more trades that are involved, the more complexity increases, said Braine. “If only a small number of trades exist, the calculations can probably be done in your head,” he said. “Scale up to 10-20 and you end up using paper. Beyond this is the realm of classical computing architectures. When there are hundreds of trades, classical computer algorithms begin to experience limitations.”

Braine added: “As it scales up further, you need to use heuristics, which includes techniques like simulated annealing.” This uses an optimisation process to identify a sufficient subset of transactions that can be cleared, he said. These transactions are then actually settled. 

In cases where the securities trade involves 50,000 to 100,000 transactions, Braine said there would be chains of transactions, such as A to B to C to D to E, all of which must be executed in the correct order. “You also get cyclic dependency and collateralisation around credit lines.”

According to Braine, a classical computer would require a large number of binary bits to describe the securities transaction settlement problem. “We can easily describe the nature of the problem on a classical computer, but the challenge is that we need to run it for a certain period of time until we get a good enough result,” he said. “This is not necessarily the best result.”

In the paper, the researchers described how a system using a small number of qubits on a quantum computer could run algorithmically complex aspects  of securities settlement trading. In a classical computer, said Braine, “we would need tens of millions of bits to represent the values in trading. It is quite a complex beast of a problem, and would need to run on 200 CPUs with large amounts of memory”.

With a seven-qubit system, he said, the team explored the core optimisation problem in securities transaction trading. “In our case, we were able to identify certain features that were of sufficient complexity,” he said.

This was the optimisation that could then be run on IBM’s cloud-based quantum computer. “When looking at quantum optimisation, you’re not looking at whether it runs better for clearing three trades,” said Braine. “Of course not – you can do that in your head. Does it run it better for a few, say 10? No. What we are looking for is whether the quantum computing has the potential to handle tens of thousands of trades and whether we can solve the trade settlement problems and get a better result on a larger scale.”

Although the paper described a proof of concept, there are many barriers to overcome before such technology makes it in the commercial world. First, there are the skills, said Braine: “There are many people at the bank with physics and computer science backgrounds, but we need to partner to get the specialism in quantum computing that is currently required.”

At Barclays, one of the roles of the chief technology and innovation office is to assess potential scenarios, he said. “We need to think about the specialist hardware required, the specialist teams and the need for the quantum computer to be hosted by a third party, which raises potential jurisdiction challenges.”

Explaining how the quantum computer came up with its answer is another problem that needs to be overcome, said Braine. “Explainability in quantum computers is similar to AI. Someone needs to articulate how the algorithm works that can be understood by a regulator and risk assessors.”

Go to Source