Quantum Phase Estimation Explained: The Engine Inside Many Quantum Algorithms
The intuition first
Imagine a spinning wheel that you cannot watch directly. You can only ask it to spin once, twice, four times, eight times, and then peek at a clock hand that shows how far it has turned. If one spin moves the hand a tiny, unknown fraction of a lap, then the eight-spin version moves it eight times as far, and the difference is easier to see. Quantum phase estimation (QPE) is that trick, done with quantum superposition so that all the different spin counts are tried at once.
In formal terms: a quantum operation U has special states called eigenstates. When U acts on one, the state comes out unchanged except for a twist, written as the number e raised to 2 pi i times theta. The number theta is the phase. QPE reads theta out as a binary fraction. As one reference puts it, because eigenvalues of such operations have unit size, the algorithm can be described as retrieving either the phase or the eigenvalue itself.
The recipe, in four moves
- Two registers. One register of n qubits holds the answer. The other holds the state you care about.
- Superposition. Put the answer register in an even mixture of all its values using Hadamard gates (see gates and circuits).
- Controlled powers. Apply U, then U twice, then four times and so on, each controlled by a different answer qubit. This stamps the phase onto the answer register in a pattern.
- Inverse quantum Fourier transform, then measure. This converts the pattern into the best n-bit guess of theta.
What it costs
The key facts, as stated in the reference below: to reach error epsilon you need about log(1/epsilon) qubits in the answer register and about 1/epsilon uses of the controlled operation. The basic version succeeds with probability at least 4 over pi squared, about 0.405, and adding roughly log(1/epsilon) extra qubits pushes the success chance up to 1 minus epsilon. For success probability 1 minus delta, the cost is about log(1/delta) divided by epsilon uses, which the same reference describes as optimal.
The 1/epsilon cost is the interesting part. To get ten times more digits of precision you need ten times more repeated use of U. A classical statistician repeating a noisy measurement would need about 1/epsilon squared samples, so the quantum method is quadratically better on that axis. The catch is that those long runs of U must stay coherent, which is why QPE is a fault-tolerant algorithm. It sits in the camp that needs error correction rather than the noisy near-term machines of the NISQ era.
Where QPE shows up
- Shor's algorithm. Factoring reduces to finding the period of a modular multiplication, and period finding is QPE on that operation. See Shor's algorithm and why it threatens RSA.
- Quantum counting and linear systems. The same reference lists quantum counting and the HHL linear-system algorithm (covered here) as users of QPE.
- Chemistry energies. If U is the time-evolution of a molecule, its phases are the molecule's energies. That connects QPE to the lead use case in quantum simulation.
What is proven and what is assumed
The cost counts above are mathematical results, and they count uses of U as the unit of cost. That is an honest and useful way to measure, but it hides the cost of building U in the first place. In chemistry, building a good version of U is most of the work. There is also a second assumption in the textbook setup: you must be able to prepare an input state that overlaps well with the eigenstate you want. If your starting guess has little overlap, you pay with many repeats. How hard that is for real molecules is an open, debated question, which is part of why timelines for chemistry come with caveats.
A note on history
The reference below credits Alexei Kitaev with introducing the algorithm in 1995, and Cleve, Ekert, Macchiavello and Mosca with a 1998 analysis of the success probability and the extra-qubit boost. Mande and de Wolf (2023) are cited for tight bounds on the dependence on the failure probability. It is a mature idea with decades of scrutiny, which makes it a firm foundation compared with newer heuristics.
The takeaway
QPE is not a headline algorithm. It is the dependable gear in the machine, and that is a good thing. Its guarantees are clean, its cost is understood, and its requirement is clear: long, clean, error-corrected quantum circuits. When labs report progress on logical qubits (see Helios and logical qubits), part of what they are aiming for is the ability to run QPE-style routines for real problems. This is education only, not financial advice, and the QNT memecoin has no link to Quantinuum Ltd.
Sources and further reading
- Wikipedia: quantum phase estimation algorithm
- Gilyen, Su, Low, Wiebe: Quantum singular value transformation (arXiv)
- Low and Chuang: Hamiltonian Simulation by Qubitization (arXiv)
Reported as of 2026-10-09. Theory results are proven only under the stated assumptions, and experimental claims and classical rebuttals keep changing, so check the primary papers. Nothing here is financial advice and nothing here predicts the price of any asset. The QNT memecoin is independent of Quantinuum Ltd, the real company, and of every lab and researcher named on this page.
Frequently asked questions
What does quantum phase estimation do?
It estimates the phase, a number between 0 and 1, that a quantum operation applies to one of its eigenstates. Equivalent to reading out the operation's eigenvalue.
How expensive is it?
As summarized on Wikipedia, error epsilon needs about log(1/epsilon) answer qubits and about 1/epsilon controlled uses of the operation.
Does it run on today's noisy machines?
Large-scale QPE needs long coherent circuits, so it is generally considered a fault-tolerant algorithm, not a near-term one.
Who invented it?
Alexei Kitaev introduced it in 1995, according to the sources cited on this page.
Keep reading
- Shor's Algorithm Explained Step by Step
How does Shor's algorithm work? A plain English walk through period finding, why it breaks RSA and elliptic curves in theory, and what hardware it would need. - Quantum Algorithms Explained for Beginners
What is a quantum algorithm? Learn how Shor's, Grover's and other quantum algorithms work in plain English, and which ones matter for cryptography and crypto. - Quantum Simulation: Trotter, Qubitization and Why Chemistry Leads
Simulating quantum systems is the original quantum algorithm. Learn how Trotter steps and qubitization work and why chemistry and materials are the lead use case.
All Quantum computing guides | Back to top | Search the site
Main pages: Quantum computing explained | Quantum and crypto | Companies | Quantum news | Glossary