ANU College of Science and Medicine
Source: https://science.anu.edu.au/news-events/news/how-tell-if-your-quantum-measurement-any-good Parent: https://science.anu.edu.au/
How to tell if your quantum measurement is any good
How to tell if your quantum measurement is any good
Friday, 13 Feb 2026
PhD researcher Aritra Das
New research from the Department of Quantum Sciences and Technology (DQST) within the ANU Research School of Physics is helping to define the quality of quantum measurements.
“A collaborator asked us to help work out how good their measurements were, and to our surprise we realised no one had explored the theoretical limits,” said Aritra Das, PhD student in the Quantum Information Group within DQST.
“People had thought about the quantum states they were measuring and had come up with the best sort of ruler to measure them with.
“But the reverse problem, of quantifying how good their rulers themselves were, was left unexplored. Everyone just assumed a perfect ruler, which is often not a realistic assumption,” said Mr Das, who was the lead author on the paper.
As well as defining theoretical limits on measurements, the team successfully tested their theory on IBM’s Sherbrooke quantum computer.
“We wanted to show that this limit was not just theoretical – and we successfully managed to reach the limit in practice.”
Led by ANU researchers, the team included international partners from the Korea Institute of Science and Technology (KIST) and the Agency for Science, Technology and Research (A*STAR), Singapore. The work is published in Nature Communications.
As well as defining theoretical limits, the team make recommendations for how to extract the best precision from a realistic measurement device affected by quantum noise.
This knowledge will have benefits in many contexts, said senior author Dr Lorcan Conlon, from A*STAR.
“For the current generation of noisy quantum computing devices, it could help researchers accurately identify and characterise the noise sources affecting their measurement systems.
“For example, it could improve quantum cryptography by enhancing the security of key distribution and enhance the performance of low-signal sensors widely used in photonic circuitry.
“Specifically, our technique enables the precise calibration of single-photon devices, which is an essential step for advancing high-performance optical quantum systems, including those being developed by companies like Xanadu and PsiQuantum,” Dr Lorcan said.
The work is timely as the US National Institute for Standards and Technology has recently issued its own recommendations for how to calibrate single-photon devices.
“The NIST advice is an experimental procedure, which involves averaging across many input states, whereas our theory shows how to choose the best input state for calibrating a measurement device,” Mr Das said.
Mr Das drew an analogy with calibrating a ruler against a standard sized object.
“Calibration is going to work a lot better if you use an object shaped, say, as a rod, rather than a sphere.”
“And for a cylindrical rod you would want to calibrate against its length dimension, rather than using its crosswise circular dimension,” Mr Das said.
The team’s findings include ways to leverage entanglement to enhance detector calibration and show how quantum resources like squeezed light can give significant boosts to precision.
In the example of quantum states of light, Dr Jie (Sophie) Zhao, leader of the ANU Quantum Information Group, explained how the input state’s shape in phase space affects precision of measurement calibration.
“Existing approaches to calibrate optical detectors use classical coherent light, which is circular shaped. While these states are easy to generate experimentally, they do not offer the best precision,” said Dr Zhao, who supervised the project.
“In contrast, nonclassical light such as squeezed light, shaped like a highly squished ellipse, can also be generated, albeit more challenging, and can offer more precise calibration.”
The team’s starting place was information theory pioneered by statistician Ronald Fisher in the 1920s, who estimated the total amount of information contained in a classical state. Subsequent researchers calculated the Fisher Information contained in a quantum state, and in the evolution of such a state.
“But no one had looked at measurement,” Mr Das said.
“You need to consider that a measurement contains some quantum information, and your task is to extract as much of this information as possible.”
“It’s a quirk of quantum information, which can be quite abstract. So seeing it work out in practice on the IBM computer was a pleasant surprise.”
The Sherbrooke quantum computer experiment worked better than the team had hoped thanks to a stroke of luck: Sherbrooke’s quantum states are well defined, with a precision of one hundredth of a per cent, whereas the detector precision was 100 times worse, at only one per cent accurate.
“That imbalance played to our favour, because we could take the measurement precision to its limit,” Mr Das said.
Better characterisation will be a benefit for quantum computing, with noise presenting a major challenge on the route to scaling them up.
The new guidelines constitute a missing piece for the study of quantum information, said co-author Dr Syed Assad, from A*STAR.
“Quantum states and the evolution of these states were both well studied in this aspect, but quantum measurement calibration was missing”
“In developing a comprehensive framework for it, we find new properties that fit in perfectly with known results for state and process calibration,” he said.
“It’s kind of like extending one leg of a wobbly three-legged stool so that now it is perfectly balanced.”
Subscribe to receive our best science stories every month
(function($) { window.fnames = new Array(); window.ftypes = new Array(); fnames[0] = 'EMAIL'; ftypes[0] = 'email'; fnames[1] = 'FNAME'; ftypes[1] = 'text'; fnames[2] = 'LNAME'; ftypes[2] = 'text'; fnames[3] = 'ADDRESS'; ftypes[3] = 'address'; fnames[4] = 'PHONE'; ftypes[4] = 'phone'; fnames[6] = 'MMERGE6'; ftypes[6] = 'text'; fnames[11] = 'MMERGE11'; ftypes[11] = 'text'; fnames[12] = 'MMERGE12'; ftypes[12] = 'url'; fnames[13] = 'MMERGE13'; ftypes[13] = 'text'; fnames[15] = 'PDF'; ftypes[15] = 'text'; fnames[5] = 'MMERGE5'; ftypes[5] = 'text'; }(jQuery)); var $mcj = jQuery.noConflict(true);