New Algorithm Simulate Quantum Computers on Normal Computers Explained: Edinburgh Team Cuts Compute to 5%
Image: Unsplash (free Unsplash licence)
📌 KEY HIGHLIGHTS
- Edinburgh and EPCC team: simulations that once needed the world's most powerful supercomputers now run on standard computers, at around 5% of the previous computing power.
- The task is Gaussian boson sampling, the benchmark used to demonstrate photonic (light-based) quantum computers.
- In tests on systems with more than 100 optical modes, the classical simulation matched — and on several measures beat — data from real photonic hardware, including China's Jiuzhang experiments.
- Published in Science Advances (preprint arXiv:2511.14923), with Polytechnique Montréal; supported by the UK's EPSRC. This is a recent research explainer, not a product launch — nothing here is for sale.
What happened?
Quantum computers promise calculations far beyond today's machines, but today's devices are small and error-prone. To design bigger, more reliable ones, scientists first simulate how a device should behave — and until now, simulating a large photonic quantum computer demanded supercomputer time available at only a handful of facilities worldwide.
A team led by the University of Edinburgh and EPCC, the UK's national supercomputing centre, changed the maths instead of buying a bigger machine. Rather than trying to list every possible quantum state — a list that explodes in size — their algorithm works with statistical features of the results, called cumulants, and honestly models real-world imperfections such as photon loss. Because real experiments are noisy, and that noise simplifies the pattern of outcomes, a clever classical shortcut can reproduce the results faithfully. The research was published in the journal Science Advances.
New algorithm simulate quantum computers on normal computers explained — the numbers
| Measure | Before this algorithm | With the Edinburgh algorithm |
|---|---|---|
| Computing power needed | 100% — supercomputer scale | About 5% of that power |
| Hardware for many tasks | Supercomputer cluster | A single standard CPU or GPU for some tasks |
| Test scale | Limited by cost and access | Systems with 100+ optical modes tested |
| One headline test | — | 10 million samples in 115 minutes on eight Nvidia H100 GPUs |
| Memory for 144-mode precomputed data | Far larger in earlier approaches | About 2 GB |
| Result vs real quantum hardware | Classical simulation lagged behind | Matched or beat Jiuzhang data on several statistical measures |
Lead researcher Dr Raul Garcia-Patron Sanchez, of Edinburgh's School of Informatics, said the work brings big improvements in memory use, running time and energy cost, "making mainstream what was before only accessible to a handful of facilities in the world."
Image: Unsplash (free Unsplash licence)
Why it matters for UK and Europe readers
This is a UK research win with a practical payoff: testing a photonic quantum design no longer has to wait for supercomputer time. University labs and startups across the UK and Europe can check ideas cheaply, find where a device loses photons or drifts from theory, and fix the design before building expensive hardware. It also raises the bar for every "quantum advantage" headline: a quantum machine must now beat this smarter classical baseline, not yesterday's brute-force simulation. That stricter test is good science — and it is the same healthy scepticism we apply to AI claims in our NVIDIA Nemotron 3.5 Embed Rerank Parse open models explained coverage.
On the flip side, supercomputers are also being used to stress-test fault-tolerant quantum designs at scales no quantum chip can yet reach — see nvidia jupiter 104 qubit quantum simulation explained.
Should you care?
✅ If you study, build or fund quantum tech: yes — your benchmarking baseline just moved, and simulation just got far cheaper.
⏳ If you are a general reader: care about the principle, not a purchase. No quantum computer is coming to your desk because of this; what changed is how honestly and cheaply scientists can test the real ones. Developers keeping everyday systems safe should also see our Node.js October 2026 security release how to update guide.
FAQs
Can a normal computer simulate a quantum computer?
For this specific task, yes. The Edinburgh algorithm lets off-the-shelf computers run Gaussian boson sampling simulations that previously needed supercomputers, by modelling statistical patterns and real-world noise instead of every possible quantum state. That does not mean a laptop can simulate every quantum program.
What is Gaussian boson sampling in simple words?
It is a test where photons — particles of light — are sent through a network of optical parts, and detectors record which exits they emerge from. The pattern of exits follows rules that are extremely hard for ordinary computers to reproduce, which is why it is used to demonstrate photonic quantum computers.
Does this new algorithm mean quantum computers are useless?
No. It means one famous benchmark is easier to simulate than thought, especially because real hardware is noisy. Quantum research continues; the result simply forces future "quantum advantage" claims to clear a higher, better classical bar.
How much computing power does the Edinburgh quantum algorithm save?
The team says the simulations need around 5% of the computing power previously required — roughly a 95% saving — with some tasks running on a single standard CPU or GPU instead of a supercomputer cluster.
TAKEAWAY
A smarter classical algorithm just made photonic quantum simulation about 20 times cheaper and forced quantum advantage claims to prove themselves against a tougher opponent. Better baselines make better quantum computers.
Follow UK & Europe Tech on TechByCountry for the next hourly update.
Sources: University of Edinburgh — Software advance could aid quantum progress • Phys.org — Software advance brings complex quantum simulations to standard computers • Innovation News Network. Paper: Science Advances / arXiv:2511.14923. Checked 21 Oct 2026.
Comments
Post a Comment