IBM researchers have announced their intention to overcome the largest challenge in Quantum Computing and introduce the first large-scale fault-tolerant machine by 2029.
This groundbreaking research showcases new error correction technologies that have the potential to create a system 20,000 times more powerful than any existing quantum computer.
In two recent studies published on the Preprint Arxiv server on June 2nd and June 3rd, researchers unveiled innovative error mitigation and correction techniques that effectively manage these errors and enable more efficient hardware scaling.
The new system, dubbed “Starling,” comprises 200 logical qubits made up of approximately 10,000 physical qubits. Following this, a machine known as “Blue Jay” is set to utilize 2,000 logical qubits by 2033.
A new study, yet to be peer-reviewed, details IBM’s quantum low-density parity-check (LDPC) code. According to the research, this paradigm shift in fault tolerance suggests that quantum computer hardware can exceed previous limitations. Jay Gambetta, IBM’s Vice President of Quantum Operations, stated to Live Science, “The science has been solved” for extended fault-resistant quantum computing. He further indicated that scaling Quantum Computers will shift from being a scientific obstacle to an engineering challenge.
Related: Google’s “Willow” quantum chip has tackled the problem of taking significantly longer to surpass the best supercomputers in the universe’s history.
Today’s quantum computers exist but can only outperform classical computer systems (which rely on binary calculations) for specially crafted problems designed to showcase their capabilities.
One of the major barriers to achieving quantum dominance lies in scaling the Quantum Processing Unit (QPU).
As scientists add more qubits to a processor, computational errors also increase. This happens because qubits are inherently “noisy,” leading to frequent errors compared to classical bits. Consequently, research efforts in this domain primarily emphasize Quantum error correction (QEC).
The Road to Fault Tolerance
Error correction poses fundamental challenges for all computing systems. In classical computers, binary bits can erroneously switch from 1 to 0 and vice versa, causing calculations to fail or yield incomplete results.
The qubits employed in quantum calculations are significantly more susceptible to errors than classical bits due to the intricacies of Quantum Mechanics. Unlike binary bits, qubits also carry “phase information.”
This allows for computations using quantum information but complicates the task of error correction significantly.
Until now, scientists have been uncertain about how to effectively scale quantum computers from the hundreds of qubits utilized in today’s models to the hundreds of millions that are theoretically necessary for practical applications.
However, Gambetta noted that the advancements in LDPCs and their successful implementation in existing systems will initiate substantial change.
The LDPC code involves a series of checks to identify and rectify errors, reducing the need for individual qubit checks; each check encompasses fewer qubits than in prior models.
A significant benefit of this technique is the markedly enhanced “encoding rate,” representing the ratio of logical qubits to physical qubits required for protection. By leveraging LDPC code, IBM plans to significantly decrease the number of physical qubits necessary for scaling the system.
This innovative approach, developed by IBM, executes error mitigation roughly 90% faster than previous methods according to the study. IBM intends to integrate this technology into the Loon QPU architecture, which will succeed the Heron architecture currently in use.
Transitioning from Error Measurement to Error Correction
Starling is anticipated to support 100 million quantum manipulations using 200 logical qubits, roughly equating to 10,000 physical qubits. In contrast, Blue Jay is theorized to facilitate 1 billion quantum operations with 2,000 logical qubits.
The current model operates with about 5,000 gates (analogous to 5,000 quantum operations) using 156 logic qubits. An IBM representative stated that advancements from 5,000 operations to 100 million can only be accomplished through technologies like LDPC. Other solutions, including those used by companies like Google, do not scale adequately to achieve the necessary size for fault tolerance.
To optimize the functionalities of Starling in 2029 and Blue Jay in 2033, IBM requires algorithms and programs tailored for quantum computing. Recently, IBM launched Qiskit 2.0, an open-source development kit designed for running quantum circuits on IBM hardware.
“The objective is to progress from error mitigation to error correction,” stated Blake Johnson. “Quantum computing is evolving from a domain where researchers experiment with quantum hardware to one where utility-scale quantum computing tools are accessible,” added IBM’s quantum engine lead in an interview with Live Science.
Source: www.livescience.com