Algorithms called phantom codes could help quantum computers run complex programs without errors, overcoming a big hurdle for making the technology more broadly useful.

Early on, some physicists doubted that quantum computers would ever be useful because they expected these devices to be too prone to hard-to-correct errors. Today, several types of quantum computers exist and have already been used for scientific discovery and exploration. Yet, while progress has been made, researchers have not managed to fully curtail the error-making problem.

Many popular error-correcting programs enable quantum computers to store information without errors, yet struggle when it comes to computation, says Shayan Majidy at Harvard University.

In search of a remedy, he and his colleagues focused on calculations that include many computational steps, which makes them long and inefficient to run, and runs the risk of additional errors creeping in.

Quantum computers are made from physical units called qubits, but these computations involve logical qubits, or groups of qubits that share information to reduce error rates. To make computation error-proof, devices usually have to manipulate logical qubits – for example, by shooting lasers or microwaves at the physical qubits – in order to make two or more of them entangled or change their quantum properties.

Phantom codes enable many logical qubits to be entangled without any physical action being necessary – hence the name “phantom”. Practically, this means that the whole computation would require fewer such actions, adding to its efficiency and cutting down the number of ways errors can happen.

Majidy and his colleagues used computer simulations to test phantom codes on two tasks: preparing a special qubit state that is often used in computations and emulating a toy model of a quantum material. They found that because it required fewer physical manipulations, their approach provided up to 100 times more accurate results than more conventional error-correction programs.

Phantom codes cannot help with every single quantum computing program, says Majidy, but they excel in situations where a computation requires a lot of entanglement already. They don’t create entanglement from nothing, he says, but rather take advantage of what is already there. “It’s not a free lunch. It’s just a lunch that was already there and we weren’t eating it,” he says.

Mark Howard at the University of Galway in Ireland says choosing an error-correction code for a quantum computing task is like choosing a suit of armor – a plate armor suit might achieve more protection than chain mail, at the cost of being heavier and less flexible. Phantom codes offer flexibility, but just like chain mail they also have drawbacks, such as requiring more qubits than some traditional approaches, says Howard. Because of this, they could be used for some targeted quantum computing program subroutines but are unlikely to be a full solution to quantum computers’ error woes, he says.

Dominic Williamson at the University of Sydney in Australia says it is an open question as to how competitive phantom codes can be with other error-correction methods, part of which may hinge on future developments in quantum computing hardware.

Majidy says his team is already closely collaborating with colleagues who build quantum computers from extremely cold atoms. He expects the lessons learned from phantom codes, combined with insights into what a qubit can practically do, will to lead to a new strategy where quantum computing programs will be more specifically tailored to a particular task and implementation.