A shorter-width post-selected circuit for a truncated-Taylor Hamiltonian-evolution segment
Listen
Radio episode about this paper
Transcript
Introduction to the show: ident: Quantum Radio. Generated commentary on the latest quantum physics and condensed matter papers.
Kai: We're looking at the paper: "A shorter-width post-selected circuit for a truncated-Taylor Hamiltonian-evolution segment".
Paper summary: Kai: So we're looking at this paper today about a shorter width post-selected circuit for a truncated Taylor Hamiltonian evolution segment. It seems the main point is that they found a way to simulate these things without needing as many ancilla qubits as usual, even while keeping the gate cost similar asymptotically.
Mira: Exactly. The thesis is that they use mid-circuit measurements and change multi-controlled gates into singly-controlled ones to achieve this reduction in qubit count. It’s about using fewer qubits overall while maintaining the same complexity for simulating time evolution operators using Taylor series expansions up to some order K, as mentioned in page one of THIS PAPER — A shorter width post-selected circuit for a truncated Taylor Hamiltonian-evolution segment <ref:2511.09461#pg1,up to some order K>.
Lev: From an engineering standpoint, I'm curious how this translates to actual hardware. They claim the total qubit count is ceiling of log K plus ceiling of log L plus n, where L is the number of terms in the Hamiltonian and n is your system qubit size. This suggests a logarithmic dependence on K and L for the ancillas, which sounds promising for scaling up if K or L get big.
Kai: It does sound good because it's exponentially reduced compared to what you might expect from a standard implementation of that Taylor series expansion. They’re using two ancilla registers specifically for encoding the Taylor series coefficients and the Hamiltonian coefficients, as page two of THIS PAPER — A shorter width post-selected circuit for a truncated Taylor Hamiltonian-evolution segment explains <ref:2511.09461#pg1>.
Mira: And they are employing mid-circuit measurements to handle the multiplication of block encodings without needing extra qubits. They suggest that every measurement indicates whether the right block is being projected, and if a failed projection happens, you just terminate and restart instantly.
Paper summary: Lev: That early abort and restart protocol is key for runtime analysis. They give an upper bound on the average runtime RTave of Wf as less than K times d e over pWf multiplied by one minus tau times alpha to the power of one minus p1, according to page three of THIS PAPER — A shorter width post-selected circuit for a truncated Taylor Hamiltonian-evolution segment <ref:2511.09461#pg1>.
Kai: So what they’re saying is that the first order reduction in runtime comes from that first failed postselection, which they say is a factor of K to the power of two times k equals zero beta e to two k plus one beta e to one minus p1 according to page three of THIS PAPER — A shorter width post-selected circuit for a truncated Taylor Hamiltonian-evolution segment <ref:2511.09461#pg1>.
Mira: Then they introduce the BLISS approach, which they suggest can be used to reduce the l1 norm of those Hamiltonian coefficients by exploiting symmetries and other invariant transformations. They link this back to optimizing the success probability of implementing WHe k using page three of THIS PAPER — A shorter width post-selected circuit for a truncated Taylor Hamiltonian-evolution segment <ref:2511.09461#pg1>.
Lev: If you look at the numerical evidence, they show that optimizing that l1 norm actually reduces the l1 norm of alpha and increases the success probability relative to an unmodified Hamiltonian, which is shown in table two of THIS PAPER — A shorter width post-selected circuit for a truncated Taylor Hamiltonian-evolution segment.
Kai: And when you look at how the resources scale with system size, figure three compares the success probabilities obtained by implementing Wf with varying number of shots against W in figure five of THIS PAPER — A shorter width post-selected circuit for a truncated Taylor Hamiltonian-evolution segment.
Mira: The paper notes that the resources utilized to execute Wf scale linearly with ceiling of log K plus one, and the scaling in two qubit gate operations from compiling the circuit W is linear with increasing K, page two of THIS PAPER — A shorter width post-selected circuit for a truncated Taylor Hamiltonian-evolution segment states <ref:2511.09461#pg1>.
Paper summary: Lev: For a specific application, they use the one dimensional Ising model Hamiltonian H equals J P n bracket i j Z i z j plus h P n i X i, and they show that the scaling in two qubit gate operations from compiling the circuit W is linear with increasing K according to page three of THIS PAPER — A shorter width post-selected circuit for a truncated Taylor Hamiltonian-evolution segment <ref:2511.09461#pg1>.
Kai: So if you only listen to the show, what this means is that for simulating these time evolution operators, we can get a circuit that uses fewer ancilla qubits while keeping the gate cost manageable and the overall complexity equivalent to previous methods.
Mira: And they are suggesting this shorter width approach, combined with mid-circuit measurements and symmetry exploitation via BLISS, offers a way to improve the success probabilities of their postselection protocol.
Lev: The limitation they state is that when measurements and gates operate on the same timescales, because the circuit might terminate dynamically upon a failed postselection, we expect reduced average runtime and gate resources according to page one of THIS PAPER — A shorter width post-selected circuit for a truncated Taylor Hamiltonian-evolution segment <ref:2511.09461#pg1>.
Kai: So in short they’ve proposed an alternative quantum circuit that uses fewer ancillas than before, achieves the same asymptotic gate cost, and relies on mid-circuit measurements and l1 norm optimization to improve its runtime performance.
Mira: And they suggest this method might be more practical for running these kinds of simulations on current or near-term hardware compared to the standard methods.
Lev: The implication is that if we can get those success probabilities up with the BLISS approach, and if measurements and gates overlap in time, we should see reduced average runtime as they predict according to page three of THIS PAPER — A shorter width post-selected circuit for a truncated Taylor Hamiltonian-evolution segment <ref:2511.09461#pg1>.
Kai: We’ll keep an eye on how these findings translate when we move from theoretical complexity to the actual fidelity of the circuits they propose, and that's where my focus lies.
Conclusion: Kai: So, we're wrapping up this look at the paper titled "A shorter-width post-selected circuit for a truncated Taylor Hamiltonian evolution segment."
Mira: Yeah, the authors are focusing on how they can build this circuit using fewer ancilla qubits while keeping it equivalent to what you used before.
Lev: From my side, I'm looking at how much overhead that reduction actually translates into when we try to run this on actual hardware.
Kai: The main thing here is that they managed to cut down the qubit count logarithmically with respect to the size of the Taylor series and the Hamiltonian terms.
Mira: That means they're using fewer qubits for encoding those complex time evolution operators, which sounds like a solid technical achievement on their own.
Lev: But I gotta ask, how much does that actually help when you're dealing with real-world noise and gate errors? We need to see if those savings in qubit count outweigh the complexity of the mid-circuit measurements they propose.
Kai: The paper suggests that by using mid-circuit measurements and optimizing their success probabilities with something called BLISS, they can get a better runtime.
Mira: It sounds like they're trying to address the high cost associated with those multi-controlled gates in the standard setup by trading some qubit count for smarter measurement strategies.
Lev: So it's not just about having fewer qubits; it's about making those measurements work more reliably so you don't waste time because of failed projections.
Kai: The authors show that this approach, when combined with their L1 norm optimization technique, can actually improve the success probability compared to a standard Hamiltonian setup.
Mira: It seems like the core idea is finding a way to make the simulation more robust by using these specific measurement protocols instead of just brute-forcing it with more ancillas.
Lev: If they can get that runtime reduction they claim when measurements and gates run at similar speeds, that’s a big deal for practical quantum simulations.
Kai: Exactly, so this paper is about finding a smarter way to build these circuits—using fewer resources and better techniques—to tackle those challenging time evolution problems.
Mira: It points toward needing more sophisticated measurement strategies when dealing with the complexity of Taylor series expansions in Hamiltonian dynamics.
Lev: So we've seen the proposed circuit structure and the theoretical improvements to runtime; now we have to see how much of this translates into a real, high-fidelity experiment on superconducting chips.
Michelle Wynne Sze, David Zsolt Manrique, David Muñoz Ramo, Nathan Fitzpatrick
Quantinuum
quant-ph
Submitted: 2025-11-12
Updated: 2026-11-17
Comments: In this version, the title and abstract are modified. Average runtime estimation for Select(V) in Eq.(19) is corrected. Supporting lemma and corollary are added in the Appendix
License: http://creativecommons.org/licenses/by/4.0/
The gist: Shorter width truncated Taylor series for Hamiltonian dynamics simulations presents an alternative quantum circuit that maintains equivalent asymptotic elementary gate cost but has an exponentially…
Key concepts
- Shorter-width circuit
- This is a quantum circuit design that uses fewer auxiliary (ancilla) qubits than previous methods to achieve the same computational goal. The authors use mid-circuit measurements and gate transformations to achieve this reduction while keeping the overall complexity similar to existing approaches.
- Mid-circuit measurements
- These are measurements performed during the execution of a quantum circuit, not just at the end. In this context, they are used as an intuitive way to perform block encodings without needing extra ancilla qubits. They allow for instant termination or restarting if a projection fails.
- Success Probability Optimization (BLISS)
- This involves optimizing the success probability of implementing certain unitary operations by reducing the l1-norm of Hamiltonian coefficients. The BLISS approach uses symmetries and other transformations to make this optimization more effective, leading to higher success rates for crucial steps in the simulation.
Terminology
Summary
Shorter width truncated Taylor series for Hamiltonian dynamics simulations presents an alternative quantum circuit that maintains equivalent asymptotic elementary gate cost but has an exponentially reduced number of ancilla qubits by utilizing mid-circuit measurements and transforming multi-controlled to singly-controlled gates
Circuit Construction and Complexity
The proposed circuit construction utilizes only two ancilla registers leading to fewer number of qubits while preserving the same asymptotic circuit depth and number of controlled-Hamiltonians The total number of qubits utilized is given by ⌈log(K)⌉ + ⌈log(L)⌉ + n, where L is the number of terms in the Hamiltonian and n is the system qubit size This results in a circuit gate complexity O(KT), which can be smaller on average because of the mid-circuit measurement protocol.
Implementation via Mid-Circuit Measurements
The method is realized by applying a parallelization method and midcircuit measurements Different parallelization methods can be exploited to tackle the expensive subroutine that uses multi-control operations in LCU. Midcircuit measurements, or incoherent computations, are an intuitive and conservative approach to perform multiplication of block-encodings without using extra ancillas Since every measurement indicates whether the right block of the larger unitary is being projected, the circuit is terminated and restarted instantly if a failed projection is detected. p. 0 Midcircuit measurements, or incoherent computations [7], is
an intuitive and conservative approach to perform multiplication of block-encodings without using extra ancillas. Since every measurement indicates whether the
right block of the larger unitary is being projected, the
circuit is terminated and restarted instantly if a failed
projection is detected.Where the paper says it
Success Probability Optimization
The success probability of implementing WHe k is given by pkpk−1…p2p1 = ⟨ψ0(He k)†He kψ0⟩. This is the same success probability if using more ancilla qubits with deferred measurements Amplifying the success probability by optimizing the l1-norm is crucial for implementing the first few controlled-Hamiltonian in Wf. p. 2 Amplifying the success probability by optimizing the
ℓ1-norm. – While the ℓ1-norm ∥α∥1
contributes trivially to the block encoding of Uτ,K, it is still crucial
for implementing the first few controlled-Hamiltonian
in Wf. A route to improve the post-selection success probability is to reduce the ℓ1-norm of the Hamiltonian
coefficients by exploiting symmetries and other invariant transformations [25, 26]. To demonstrate how such
symmetry based reductions can improve post-selection
efficiency, we apply the block-invariant symmetry shift
(BLISS) approach [25] to a few second-quantized molecular Hamiltonians, replacing H with the energetically
equivalent operator
H(ξ0, ξ) = H −
ξ0 +
X
N
i,j=0
ξij a
†
i
aj
Nˆ
e − Ne
, (14)
where a
†
i
(ai) are the electronic creation (annihilation)
operators, Nˆ
e =
PN
k=0 a
†
k
ak is the total electron number operator and Ne is the electron number.Where the paper says it
Where the paper says it
Runtime Analysis and Improvements
The total runtime of a successful experiment with mid-circuit measurements is given by d(1 + p1 + p1p2 + … + p1p2…pk−1). The upper bound to the average runtime RTave of Wf is given by RTave < Kde/pWf [ 1 − τ∥α∥1 (1 − p1)]. p. 4 With mid-circuit measurements, the first order or largest reduction in the total runtime of a successful implementation of Wf comes from the first failed
(controlled-UHe ) postselection, and is a factor of
K−1 X2
k=0
βe2k+1
βe
1
(1 − p1) = 1 − p1
βe
1
K−1 X2
k=0
(τ ∥α∥1
)
2k+1
(2k + 1)! Thus, an upper bound to the average runtime RTave of
Wf is given by
RTave <
Kde
pWf
1 −
τ ∥α∥1
βe
1
(1 − p1)
. If de is the average runtime
of a single controlled-UH, then the total runtime of K
controlled-UH with successful deferred postselections is
Kd/p e
Wf. With mid-circuit measurements, the first order or largest reduction in the total runtime of a successful implementation of Wf comes from the first failed
(controlled-UHe ) postselection, and is a factor of
K−1 X2
k=0
βe2k+1
βe
1
(1 − p1) = 1 − p1
βe
1
K−1 X2
k=0
(τ ∥α∥1
)
2k+1
(2k + 1)!Where the paper says it
Where the paper says it
Conclusion
We have presented an alternative circuit that implements an approximated Taylor series representation of the time evolution operator The circuit uses fewer number of ancillas that is logarithmic with respect to the maximum Taylor series order and number of terms in the Hamiltonian while achieving the same complexity as in Ref. [1]. We have proposed to employ the BLISS-based approach to improve the success probabilities of the mid-circuit post-selection protocol in our algorithm. p. 4 The circuit
uses fewer number of ancillas that is logarithmic with
respect to the maximum Taylor series order and number
of terms in the Hamiltonian while achieving the same
complexity as in Ref. While the applicability of
oblivious amplitude amplification with midcircuit measurements is reserved for future work, we have proposed
to employ the BLISS-based approach to improve the
success probabilities of the mid-circuit post-selection
protocol in our algorithm. While the applicability of
oblivious amplitude amplification with midcircuit measurements is reserved for future work, we have proposed
to employ the BLISS-based approach to improve the
success probabilities of the mid-circuit post-selection
protocol in our algorithm. In cases when measurements
and gates operate on the same timescales, since dynamic
termination of the circuit may occur upon a failed postselection, we expect reduced average runtime and gate
resources.Where the paper says it
Where the paper says it
Appendix Details
Algorithm 1 outlines the steps for Truncated Taylor series for the time evolution operator with mid-circuit measurements The final outcome of Wf is given by Wf¯0⟩k¯0⟩lψ⟩ = 1/βe 1 ¯0⟩k¯0⟩lUτ,Kψ» + vuut 1 − 1/βe 2 ⊥⟩k,l,ψ, where ⊥⟩k,l,ψ are the states in the subspace orthogonal to ¯0⟩k,i.e., ⟨¯0k ⊥⟩k,l,ψ = 0. The success probability of projecting to the ¯0⟩k state is given by pWf = ⟨ψU† τ,KUτ,Kψ» βe 2/1, where s =∥βe∥1. p. 2 2, Wf|¯0⟩k|¯0⟩ℓ|ψ⟩
is the state after Be†
is performed, given K successful
postselections on |·⟩ℓ-register, and before the postselection on the |·⟩k-register. Finally, the success probability
of projecting to the |¯0⟩k state is given by
pWf =
⟨ψ|U
†
τ,KUτ,K|ψ⟩
βe
2
1
, (13)
which is equivalent to the result of Ref.Where the paper says it
Numerical Evidence
The numerical data set is obtained by calculating the right-hand side of Eq. (13) via matrix multiplication. p. 3 The numerical data set
is obtained by calculating the right-hand side of Eq. (13) via
matrix multiplication. 782
Table II. Improvement in the success probability p of
UHe due to ℓ1-norm ∥α∥1 optimization of the Jordan-Wigner
encoded Hamiltonian coefficients. To numerically quantify the
gain, we combine BLISS with Jordan–Wigner encoding
and optimize the Hamiltonians of the molecules listed
in Table II. In every case, the optimized Hamiltonian
coefficients reduce ∥α∥1 and increase p relative to the
unmodified Hamiltonian. (16)
In Fig. 4, we quantify the average runtime in terms of
two-qubit gate count depth plus measurements for an
Ising model Hamiltonian with different number of lattice sites. (Observe the nsites = 3, 4
in the plot). The scaling in two-qubit gate operations
from compiling the circuit W is linear with increasing
K. (Observe the nsites = 3, 4
in the plot). The scaling in two-qubit gate operations
from compiling the circuit W is linear with increasing
K. The scaling in two-qubit gate operations
from compiling the circuit W is linear with increasing
K. Comparing the two sets of data, we observe that
the savings from performing mid-circuit measurements
becomes more apparent for higher K.Where the paper says it
Where the paper says it
Where the paper says it
Where the paper says it
Where the paper says it
Where the paper says it
Where the paper says it
Figure Interpretation
Figure 3 compares the success probabilities obtained by implementing Wf, with varying number of shots, and W in Fig. 5. Higher number of shots leads to faster convergence to the analytical results. p. 2 3, we compare the success probabilities obtained by
implementing Wf, with varying number of shots, and W
in Fig. Higher number of shots leads to faster convergence to the analytical results. 4, we quantify the average runtime in terms of
two-qubit gate count depth plus measurements for an
Ising model Hamiltonian with different number of lattice sites. The resources utilized in executing Wf are
linear with ⌈log(K + 1)⌉. Two-qubit gates utilized in the W (◦) and Wf (⋆) circuits as functions of system size nsites for highorder K. The advantage of employing mid-circuit measurements becomes more pronounced as the system size becomes
larger.Where the paper says it
Where the paper says it
Where the paper says it
Specific Application Results
The 1D-Ising model Hamiltonian, H = J Pn ⟨i,j⟩ ZiZj + h Pn i Xi, is used with varying lattice sites nsites are used. The average runtime of a single shot experiment is the weighted sum of all possible runtimes of early terminated events and a complete one. p. 3 On the other hand, with mid-circuit measurements
as in Fig. 1, the average runtime of a single shot experiment is the weighted sum of all possible runtimes of
early terminated events and a complete one, that is
k
X−1
j=1
(1 − pj )
j
Y−1
i=1
pijd
+ p1p2. 2 depends on the
weight of the controls, βek. If de is the average runtime
of a single controlled-UH, then the total runtime of K
controlled-UH with successful deferred postselections is
Kd/p e
Wf. With mid-circuit measurements, the first order or largest reduction in the total runtime of a successful implementation of Wf comes from the first failed
(controlled-UHe ) postselection, and is a factor of
K−1 X2
k=0
βe2k+1
βe
1
(1 − p1) = 1 − p1
βe
1
K−1 X2
k=0
(τ ∥α∥1
)
2k+1
(2k + 1)! Thus, an upper bound to the average runtime RTave of
Wf is given by
RTave <
Kde
pWf
1 −
τ ∥α∥1
βe
1
(1 − p1)
.Where the paper says it
Where the paper says it
Where the paper says it
Acknowledgements
We thank Marcello Benedetti, Ben Criger, Yuta Kikuchi and Gregory Boyd for helpful discussions, and the team behind Guppy programming language for the generous technical support. The paper is available on arXiv:2511. p. 0 5), and ours
as Wf. Spacetime complexity estimates of the two algoarXiv:2511. Spacetime complexity estimates of the two algoarXiv:2511. 09461v1 [quant-ph] 12 Nov 2025Where the paper says it
Where the paper says it
Green — the paper says it, and the page is printed. Red — no passage carries it, so treat that sentence as unverified. The same check runs on your own text: check a summary against its source, $5.
Improvements for AI systems
-
Mid-circuit measurements allow for early termination of circuit execution if a failed projection is detected, which reduces resources and runtime by avoiding idling qubits or unnecessary gate operations.
-
The proposed circuit construction Wf utilizes only two ancilla registers, leading to fewer qubits than the reference circuit W while preserving the same asymptotic circuit depth and number of controlled Hamiltonians.
-
The BLISS-based approach can be applied to reduce the LCU success probability by exploiting symmetries, which increases post-selection efficiency when implementing mid-circuit post-selection protocols.
-
Optimizing the l1-norm of Hamiltonian coefficients through symmetry based reductions, such as the block invariant symmetry shift (BLISS) approach, can increase the success probability p of a controlled UHe operation.
-
The system can be simulated for complex quantum many-body systems, such as 1D Ising models with varying lattice sites, to evaluate circuit performance and observe that the savings from mid-circuit measurements become more apparent for higher order K.
Abstract
We present an ancilla-efficient, post-selected circuit construction for a truncated-Taylor approximation to short-time Hamiltonian evolution. The construction combines a binary Taylor-order register with mid-circuit measurement, reset, and reuse of the Hamiltonian-index register. Conditioned on the prescribed all-zero outcomes, the circuit implements the truncated Taylor approximation while maintaining coherence among the Taylor-order branches while permitting early termination after a failed post-selection. For Taylor cutoff K, a Hamiltonian containing L terms, and an n-qubit system, the circuit uses (K+1) + L +n qubits, and executes O(K) controlled Hamiltonian blocks in a complete segment. We analyze its post-selection probability and expected per-segment cost, and implement and evaluate the construction in Guppy[1,2]. The present work focuses on the low-width implementation of individual short-time propagation segments; coherent composition into a deterministic long-time simulation requires an additional amplification or compression mechanism and is left for future work.
Sources
- Hamiltonian Simulation in the Interaction Picture
- Methods for Reducing Ancilla-Overhead in Block Encodings
- Parallel Quantum Computation and Quantum Codes
- Low-Overhead Parallelisation of LCU via Commuting Operators
- A randomized quantum algorithm for statistical phase estimation
- Generalized Quantum Singular Value Transformation
- Repeat-Until-Success: Non-deterministic decomposition of single-qubit unitaries
- Quantum Measurement for Quantum Chemistry on a Quantum Computer
- Hamiltonian dynamics simulation using linear combination of unitaries on an ion trap quantum computer
Related papers
- Fulqrum: a generalized framework for quantum subspace diagonalization
- Flux-Activated Resonant Control of a Bosonic Quantum Memory
- Machine learning the effects of many quantum measurements
- Barriers to dequantization of near-term molecular electronic structure methods
- BRIDGE: Lazy-Move Compilation for Dual-Species Neutral-Atom Quantum Computers
- Non-Hermiticity of an anomalous superradiant phase