Three new papers suggest ways to establish trust in quantum computer results even when classical simulation fails. Israeli company Kedma participated in one of the studies, but the papers have not yet been peer-reviewed and comparisons to classical computing are ongoing.
IBM and research groups from the University of Chicago, the Israeli Society Progress (Qedma) and Algorithmiq have published three papers in which they present demonstrations of Quantum advantage — a calculation that reaches into a realm where leading classical methods have difficulty providing a reliable answer — along with mechanisms designed to check whether the quantum result can be trusted.
The main innovation is not just increasing the number ofqubits Or the depth of the circuits. The researchers are trying to tackle a fundamental question: If the calculation is too difficult for a classical computer to reproduce, how can you know that the quantum computer didn't return an incorrect result due to noise and hardware errors?
IBM defines quantum advantage as a calculation that not only exceeds the capabilities of classical computing, but also allows for rigorous verification of its correctness. The three studies offer different answers to the verification problem. However, all have been published on arXiv at this stage, before peer review, and IBM itself notes that the claims will continue to be tested against new classical simulation methods.
A difficult-to-calculate circuit that includes a test mechanism
In the first study, researchers from IBM and the University of Chicago developed structured quantum circuits, based on Clifford circuits to which T-gates were added. The addition makes the calculation very difficult for classical simulation, while the basic structure allows for error tracking using a code that extends in space and time.
In the experiment, the researchers operated a 70-depth circuit with 70 qubits and 468 T-gates. The computation was encoded using 97 physical qubits. After filtering out runs in which error symptoms were detected, a tenfold reduction in the effective error rate was obtained.
The researchers calculated a lower bound of 0.284 for the state fidelity, at a 95% confidence level. This does not mean that the calculation was error-free, but that the mathematical structure and control measurements allowed a quantitative limit to be placed on the quality of the result without computing the full quantum output on a classical computer.
Advances have used the quantum computer as a tool for studying quantum matter
The second study, in which Kidma, IBM, RIKEN, BlueQubit, and other research groups participated, dealt with the Fluke dynamics of a multi-particle quantum system — a system that receives periodic excitation and can maintain organized oscillations for a long time before reaching equilibrium.
The researchers ran a processor IBM Heron r3 Circuits of up to 74 qubits, designed to simulate an Ising magnet in heavy-hex geometry. Using software QESEM of progress they made Error reduction And measured the development of magnetization with an accuracy of a few percent.
According to the article, at a certain point two leading classical simulation methods no longer provided a stable answer: tensor networks did not converge, while the Pauli-path method remained highly dependent on the level of truncation, even though the calculations were performed using advanced graphics processors and the Fugaku supercomputer in Japan.
To verify that the measured fluctuations were not a product of computer noise, the researchers used several independent error reduction methods, tested the noise model, and repeated some of the measurements on Quantinuum's trapped-ion quantum computers. The agreement between the tests provided, they say, evidence that the measured behavior reflects the physics of the system.
The scientific significance of the experiment is that a noisy quantum computer, which does not yet correct all its errors, may serve as a quantitative instrument for studying quantum matter under conditions where classical methods begin to lose reliability.
Verify the calculation process instead of the answer
In the third study, researchers from Algorithmiq, IBM, and other institutions developed a method for estimating a magnitude called the Loschmidt echo operator. This magnitude describes how information propagates in complex quantum systems.
The experiments were conducted in 56-qubit circuits and reached a point where several groups using leading classical simulation methods obtained results that differed from each other and from the quantum computer.
Rather than comparing the quantum result to an unavailable classical reference answer, the researchers tested its stability across several quantum computers with different noise characteristics. They also showed how bias-free estimates and quantitative error ranges can be obtained, provided the computer's noise model is accurately described.
Thus, the question of verification is shifted from the question "Can a classical computer reproduce the result?" to the question "Can the calculation process and its error model be sufficiently tested?".
Quantum advantage—but not yet a final decision
Jay Gambetta, IBM's research director, said the company is now "firmly in the quantum advantage era." However, even IBM's advertising Clarifies that the declaration of quantum advantage does not close the debate but rather opens the results to more rigorous scrutiny.
The three studies do not show that a quantum computer is currently superior to a classical computer at any task, nor do they demonstrate proven commercial application in areas such as drug development, logistics, or finance. They address specific physical and computational problems, chosen in part because they are suitable for testing the limits of classical and quantum computing.
In the past, there have been declarations of quantum superiority or advantage, followed by the development of classical algorithms that greatly improved the comparison performance. That is why the community operates the Quantum Advantage Tracker, which allows researchers to present quantum advantage claims and attempt to refute or strengthen them using new classical and quantum methods.
The current achievement is important because it shifts the emphasis from the number of qubits alone to the reliability of the result. If the methods stand up to further testing, quantum computers may begin to contribute to scientific research even before the advent of large computers with full error correction.
Questions and Answers
What is quantum advantage?
A situation in which a combination of a quantum computer and classical computing performs a task with precision, cost, or time that is not practically achievable using leading classical methods alone.
Why is quantum computation difficult to verify?
When the calculation is too large for a classical computer to reproduce, there is no simple reference answer to which the quantum output can be compared.
What was the role of Israeli progress?
Qedma's QESEM software was used to reduce errors and verify measurements in an experiment that simulated the dynamics of a quantum system with up to 74 qubits.
Has it been definitively proven that quantum computers are superior to supercomputers?
No. The studies make three strong claims on certain issues, but they have not yet been peer-reviewed and comparison with classical methods is ongoing.
For the original publication: Opening the original publication
More on the subject on the science website
- IBM quantum computer helped calculate how tritium binds to molten salt in fusion reactors
- Wolf Prize winner Jainendra Jain in a special interview: How entangled electrons became composite fermions
- Brain tissue more than 300 million years old preserved in a tiny fish fossil
- Not a time machine: This is how researchers made a quantum system look like it was going backwards
- US and IBM establish first dedicated quantum chip factory
