I write for a business audience, if you are a Physics boffin I suggest you tuck straight into this paper here. For the rest of us…
A team at Google Research (Quantum AI) has introduced a clever new quantum computational trick called Decoded Quantum Interferometry (or DQI if you’re mates).
It’s a fresh quantum algorithm that could - just, maybe, solve some of the toughest problems much faster than regular computers ever could.
What’s the Big Idea?
Normally, when quantum computers tackle optimisation, they play around with energy landscapes (like climbing hills and valleys).
That’s what methods like QAOA and adiabatic optimisation do.
But DQI says, “Hold my quantum beer,” and uses interference patterns instead.
This is quite confusing so lets break it down using non technical language.
Quantum Energy Landscapes: Explained
Imagine you’re at a giant theme park.
- There are hills, valleys, tunnels, and crazy slides everywhere.
- Your job is to find the deepest valley, because somewhere down there is a hidden pot of gold, with that pot of gold representing the best solution to an optimisation problem.
Now, you have three choices:
1. Adiabatic Optimisation:
You start at the top of the park, and you slowly stroll downhill, being very careful not to trip.
This method hopes that if you go slowly enough, you’ll land at the best (lowest) point without getting stuck on a small hill.
2. QAOA (Quantum Approximate Optimisation Algorithm):
Instead of strolling, you jump around on a pogo stick! 🦘
It’s faster but riskier - you might land in a slightly smaller valley by accident, but if you tweak your jumps just right, you get close to the pot of gold!
- Both methods are about moving over a landscape made of peaks and dips (representing good and bad solutions).
- They use energy (height) as a clue: lower = better.
- But they do it differently — slow and steady stroll (adiabatic) vs. bouncy, tuned jumping (QAOA).
And now with DQI, instead of hiking or bouncing, you just make the right valley glow using quantum light tricks — and walk straight there without breaking a sweat!
Think of it like a light show at a concert.
DQI arranges the light (or probability amplitudes) to shine brightest on the answers that are really, really good.
All the rubbish answers?
They get lost in the dark.
Instead of climbing hills, DQI plays with waves, making good solutions stand out naturally.
How DQI Works
- Start with a superposition: the quantum computer starts in a state where all possible answers are explored at once. (it’s a bit Hogwarts but this is how they work.)
- Use the Quantum Fourier Transform: this reshuffles the quantum state to make patterns where good answers shine brighter.
- Apply clever polynomials: A polynomial is just a fancy maths formula made up of numbers, x’s, and powers of x added together, the polynomial acts a bit like a quantum sound equaliser, turning up the volume on the best (good solutions) and muting the bad ones (rubbish solutions).
- Decode: the last step is like solving a code-breaking puzzle. If you can decode it correctly, boom—you get the better solutions more often.
Easy, right? Well, at least easier than it sounds in quantum language. Of course thats not exactly how it works, but should serve as a picture-gram for the core concepts.
Quick Side Quest: What’s Max-XORSAT?
Imagine you have a massive list of yes/no questions, and your job is to answer as many correctly as possible.
Max-XORSAT is basically that, but each question is tied up with a weird little maths equation. t’s really hard, it’s an NP-hard problem- meaning it takes forever for classical computers to solve when the problem gets large
DQI can help solve these puzzles by cleverly boosting the odds of getting more right answers.
Where DQI Shines the Brightest
DQI is especially powerful when the optimisation problems are naturally sparse - meaning only a few bits of important information are hiding among lots of noise. It’s like trying to find a few gold coins in a giant field of grass.
It turns out many real-world problems — like machine learning, coding theory, and even logistics — have this hidden structure. Cybersecurity certainly has these options in translating data to actual threats which is why all CISO’s are nuts (in a good way).
So, DQI could make a real difference there.
One example where DQI flexes its muscles is something called the Optimal Polynomial Intersection (OPI) problem.
Without getting lost in the maths jungle, it’s about finding a curve that passes through as many good spots as possible.
DQI can solve this better than any classical algorithm known so far — and it does it faster.
Potential quantum speedup? You bet.
Wait, Is This Quantum Supremacy?
Not quite… but it’s getting close, in fact every day we are getting closer and closer.
DQI shows that for some problems, a quantum computer can beat classical ones by a lot.
But for others (especially when things get too messy or dense), classical computers still hold their own — for now*.*
It’s like comparing a sports car to a tractor.
Each wins on its own track.
The Nitty-Gritty: What Makes DQI So Cool?
Sparse Problems: DQI loves it when the problem doesn’t have too much clutter.
Quantum Fourier Magic: This gives a neat way to make good solutions “shine” more in probability.
Coding Tricks: It taps into classic error-correcting codes like Reed-Solomon codes to solve parts of the problem.
And here’s the really juicy bit: for the polynomial fitting problem (OPI), classical methods can only solve it up to 55% accuracy unless they use crazy amounts of time.
DQI gets to about 72% — using far fewer resources.
In quantum terms, that’s basically like doing a high-five while everyone else is still tying their shoelaces.
What This Means for the Future
If DQI continues to develop, which it undoubtably will, we could see quantum computers helping to:
- Speed up machine learning - AI is Cool But Quantum ML is Bonkers
- Solve massive logistics problems
- Improve telecommunications - See Quantum for 6G
- Decode complicated error-correcting codes faster
It won’t replace classical computing altogether - but it could supercharge certain industries where finding a ‘good enough’ solution faster means winning billion-pound races.
And who knows? Maybe one day DQI could help in designing better medicines, fixing broken supply chains, or even sorting out the quantum internet..
Currently just one tiny mistake in DQI’s decoding step can ruin everything.
It’s a bit like trying to bake a cake and accidentally swapping salt for sugar.
This is hard stuff to get right.
Bottom Line
Decoded Quantum Interferometry is a fresh, exciting idea in quantum computing. Instead of brute-forcing answers, it uses the elegance of wave interference and clever maths to sniff out good solutions faster.
[edit] ” The goal, in quantum computing, is always to choreograph a pattern of interference such that, for each wrong answer, some of the contributions to its amplitude are positive and others are negative (or, they point every which way in the complex plane), so on the whole they interfere destructively and cancel each other out. …” Alain Chancé
Note that, if it weren’t for interference, then we might as well have just used a classical computer with a random-number generator, and saved the effort of building a quantum computer. In that sense, all quantum-computational advantage relies on interference.” — Scott Aaronson”
It’s early days yet - but if DQI lives up to its promise, the world of optimisation (and maybe even broader business applications) could change faster than a toddler on a sugar high.
No tractors were harmed in the making of this quantum algorithm.
Sources and references:
- Stephen P. Jordan et al., Optimization by Decoded Quantum Interferometry, Google Quantum AI .
- Further technical references available in the original paper.
With great thanks to my friend Alain Chancé who highlighted this paper for me, it took me 24 hours to break down but was well worth it. Thank you Alain.
