Scientists and engineers at the University of Sydney and Microsoft Corporation have developed a device that can handle thousands of qubits. To put things in perspective, the current state-of-the-art quantum computer can control only 50 qubits at a time.

Scaled-up quantum computers require control interfaces to manipulate or readout a large number of qubits, which usually operate at temperatures close to absolute zero (1 Kelvin or -273 degrees celsius).

The complementary metal-oxide-semiconductor (CMOS) technology has its limitations due to high thermodynamic dissipation, leading to heating of the fragile quantum bits. Overheating of quantum bits compromises its quantumness, the property of being in two states at the same time (also called superposition).

The current architecture uses multiple connections as every qubit is controlled by external circuitry with a separate electrical connection, generating a lot of heat.

The scientists from the University of Sydney built a CMOS interface between the qubits and the external circuitry, in such a way that the CMOS chip can generate control pulses for multiple qubits, with just four low-bandwidth wires, at 0.1 Kelvin, a temperature 30 times colder than deep space, with ultralow power dissipation.

The interface consists of four low-bandwidth wires at room temperature to provide input signals to the chip, which then configures 32 analogue circuit blocks to control the qubits that use dynamic voltage signals.

Analogue circuit boards use the low leakage of the transistors to generate dynamic voltage signals for manipulating qubits, consuming significantly less power.

Quantum computers are at a similar stage that classical computers were in their 40s when machines needed control rooms to function.

However, this chip, according to the scientists, is the most advanced integrated circuit ever built to operate at deep cryogenic temperatures.

The quantum computers that we have now are still lab prototypes and are not commercially relevant yet. Hence, this is definitely a big step towards building practical and commercially relevant quantum computers, said Mr Viraj Kulkarni, But I think that we are still far away from it.

This is because of the Error Correction. Any computing device always has errors in it and no electronic device can be completely perfect. There are various techniques that computers use to correct those errors.

Now the problem with quantum computing is that qubits are very fragile. Even a slight increase in temperature, vibrations, or even cosmic rays can make qubits lose their quantumness, and this introduces errors. So the key question of whether we can really control these errors is still relevant.

Nivedita Dey, research coordinator at Quantum Research and Development Labs, said the qubit noise is still a roadblock in developing quantum computers.

One of the biggest challenges in implementing a quantum circuit in this Noisy Intermediate Scale Quantum (NISQ) era is qubit noise, which causes hindrance in commercial availability of fault-tolerant full-scale quantum computers, said Ms Dey.

This approach can be well suited for practical quantum applications and might reduce the number of error-correcting qubits to be associated with noisy qubits, she added.

If quantum computing does prove to be commercially viable, it will open up completely new avenues.

A plane is not just faster than a car, it can also fly, said Mr Kulkarni, drawing an analogy between quantum computers and conventional computers. The idea is that quantum computers are not just faster, but at the same time will provide us with solutions that are better, especially in AI.

Hence, many applications in AI including complex mathematical equations, drug discovery by enabling chemical simulations, or building financial applications to come up with a better strategy will be solved in a faster and efficient way.

In the end its a tool, so any function a conventional computer can achieve, quantum computers will be able to do it faster and better.

Original post:

Microsoft Scientists Build Chip That Can Handle Thousands Of Qubits - Analytics India Magazine