Reducing quantum resources for observable estimation with window-assisted coherent QPE
Harriet Apel, Cristian L. Cortes, Jessica Lemieux, Mark Steudtner · 2025
Read Reducing quantum resources for observable estimation with window-assisted coherent QPE by Harriet Apel, Cristian L. Cortes, Jessica Lemieux, Mark Steudtner (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2508-06677.txt.
Introduction
Harriet Apel, Cristian Cortes, Jessica Lemieux, and Mark Steudtner cut the cost of a core quantum routine. Their paper, from twenty twenty-five, improves quantum phase estimation. The routine reads out a phase angle, but finite registers make it fail at times even with perfect hardware. A classical trick, windowing the register, softens that flaw.
They study windowing inside a larger task: measuring molecular observables. A reflection step built on the routine is the costly part. With windows plus better data loading, gate counts drop by more than a hundredfold. That saving could bring chemistry targets within reach sooner. The analysis also models gains for other algorithms that reuse the routine.
Catalog record
| CREATOR | Harriet Apel, Cristian L. Cortes, Jessica Lemieux, Mark Steudtner |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 102 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2508-06677
Browse the full library · About the Tower of Babel
Full text
The opening of the work, from the library's hosted text:
Reducing quantum resources for observable estimation with window-assisted coherent QPE Harriet Apel, Cristian L. Cortes, Jessica Lemieux, Mark Steudtner Abstract Quantum Phase Estimation (QPE) routines are known to fail probabilistically even with perfect gates and input states. This effect stems from an incompatibility of finite-sized quantum registers to capture a phase within QPE with phase angles of infinite precision, and the effect extend even beyond what would be reasonably expected from rounding. This effect can be partially mitigated by biasing the phase register with a window, or taper state, from classical signal processing. This paper focuses on how windowing a coherent QPE used as a subroutine can improve the accuracy of the overall algorithm. Specifically we study the quantum task of estimating observables where window-assisted coherent QPE is used as a subroutine to implement a reflection about an eigenstate. Quantum resource estimates show over 2-orders-of-magnitude reduction in Toffoli counts over the previous costed techniques – also assisted by the use of improved block encoding techniques – demonstrating an encouraging decrease in resources for quantum computation of molecular observables. Since QPE, as one of only a few quantum building blocks, appears as a subroutine in many algorithms; this analysis also provides a model for understanding how window functions propagate to an improved error in composite algorithms. Quantum phase estimation (QPE) is a seminal routine in quantum algorithms, leveraging the quantum Fourier transform (QFT) to learn the eigenphase of a unitary with respect to a particular eigenstate. In addition to the qubits required to describe the system, the routine requires a phase register that is measured to extract the eigenphase. In practice, the phase register consists of a finite number of qubits and introduces a ‘bit discretisation’ effect. This does not only result in the coarsening of data into bins of finite precision, but distorts the probability distribution of the measured phase state. Therefore, even given an exact eigenstate as input, with non-zero probability the final phase register measurement obtains an outcome far from the true eigenphase – this failure probability can be as high as 19\% . This affects the success probability of the routine, necessitating repetitions to learn the phase with confidence. Bit discretisation can be well understood from the vantage of classical signal processing, as this distortion has the same origins as performing a Fourier transform on a time-limited signal Sanders2020 . It was also noted that preparing the register that holds the phase estimate in a superposition with non-uniform amplitudes can reduce the effect of bit discretisation. A uniform superposition is akin to sampling the signal from a rectangular window with sharp discontinuities at the periodic boundary, whereas a tapered window function dampens these discontinuity and reduces the distortion of the probability distribution. While being ubiquitous in classical signal processing, window functions have been used in quantum both to asymptotically reduce query counts Berry ; Coupledoscillator23 and to practically outperform quantum signal processing (QSP) in small instances greenaway2024case . There are two important quantities to assess the utility of an algorithm: the success probability describing the likelihood that a given run of the algorithm produces a “correct result”; and the error describing the maximum deviation from the true value that we are willing to accept as a “correct result”. For example, if the true value is x_{0} , we allow an error of \epsilon and the algorithm produces the result x according to some distribution, then the success probability of the algorithm is P(\,\mathinner{\!\left\lvert x-x_{0}\right\rvert}\leq\epsilon) . In standalone QPE – where the phase register is directly measured – window functions can increase the success probability of the algorithm for a given error but increasing the accuracy can only be improved by increasing the number of phase qubits to achieve a lower error. This work studies window functions for coherent QPE subroutines within a larger algorithm. The coherent nature of the QPE indicates that the phase register is not measured at the end of the routine and instead further gates are applied using the full phase register state for computation. Therefore, instead of a failure probability, we consider a failure amplitude of the subspace spanned by \{\lvert x\rangle\} with \mathinner{\!\left\lvert x-x_{0}\right\rvert}>\epsilon . This introduces a mid-circuit error, hence improving the success amplitude using window functions in coherent QPE increases the accuracy of the overall algorithm and can help achieve a lower error tolerance. Improving the algorithmic error is more compelling than improving the success probability, since while the latter can be compensated by taking repeated measurements, improving the former requires resources such as additional gates and logical qubits which will determine whether such algorithms can be run on limited hardware at all. A notable application of coherent QPE is in estimating the expectation value of an operator with respect to a system’s ground state Knill2006 ; Obrien2020 ; seve . This is an interesting case study since the concatenated nature of the algorithm nicely demonstrates the propagation from the QPE success amplitude to the overall algorithmic error. Furthermore, it is one of the few fault-tolerant algorithms using a coherent QPE where the quantum resources have been numerically estimated seve . The natural context for this algorithm is the estimation of quantum observables on ground states of molecular systems, which has profound applications in computational chemistry. Expectation values of unperturbed wavefunctions may be used to estimate first-order molecular properties such as molecular multipole moments, forces on nuclei and stationary points on potential energy surfaces frenkelʹ1934wave ; Clusius1941EinfhrungID . Furthermore, they can also be used to compute higher-order properties such as polarizability and harmonic vibrational frequencies with additional post-processing Helgaker . These properties play a significant role in predicting reactivity and as well as understanding spectroscopic data. The finite phase register in the QPE subroutine introduces a bit discretisation that degrades the algorithmic error – reducing the precision to which we can learn these molecular properties. This work both analyses the error of this algorithm when the QPE phase register is unitarily prepared in a general state and demonstrates the practical benefit of choosing the Kaiser window Kaiser1980 as the phase register taper function. The error analysis demonstrates that coherent use of QPE is exactly where introducing window functions can have the most significant reduction in quantum resources. As is somewhat expected due to the complex nature of the question, our resource estimates still clearly place this algorithm in the regime requiring a large-scale fault-tolerant quantum computer. However, the advantage of using window functions over other techniques to combat bit discretisation in QPE (such as applying QSP to round phase values in QPE Rall2021fastercoherent ) is seen via order-of-magnitude reductions in the number of Toffoli gates required to estimate molecular observables. We also show that this reduction is further improved by using state of the art block-encoding methods Rocca2024 ; loaiza2023block ; Zak2024 as well as adapting recent tensor factorisation methods (BLISS-THC) to simultaneously optimise the Hamiltonian and observable rank truncation parameters. The technical background of both window-assisted QPE and the expectation value algorithm is first given in Section 1 . Slight modifications to previous expectation value algorithms Knill2006 ; Obrien2020 ; seve are required to capture the benefit of window-assisted QPE, so this is presented and analysed in Section 2