A fidelity-driven approach to quantum circuit partitioning via weighted hypergraphs for noise-resilient computation

AbdelMehsen Ahmad; Awad Wehbe; Safiya Al Khatib

REVIEW 4 major objections 6 minor 14 references

A fidelity-driven approach to quantum circuit partitioning via weighted hypergraphs for noise-resilient computation

T0 review · 4 major / 6 minor · reviewed 2026-08-07 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗

Pith's one-line read Fidelity-aware hypergraph partitioning cuts estimated SWAP gates by 77.3–100% and raises estimated circuit fidelity by up to about 250% across the tested NISQ benchmarks.

desk verdict A plausible error-aware partitioning scheme whose headline fidelity gains don't survive its own error model for two of three benchmarks; the core idea is worth a look, the evaluation needs major rework. read the letter →

arxiv 2506.06867 v2 pith:H4ROBEJN submitted 2025-06-07 quant-ph

Awad Wehbe , Safiya Al Khatib , AbdelMehsen Ahmad This is my paper

classification quant-ph MSC 68Q1205C65 PACS 03.67.Ac03.67.Lx

keywords quantumcircuitpartitioningfidelity-awarehypergraphNISQcompilationSWAPgatereductionerrorratesqubitmappingMt-KaHyPar

verification ladder T0 review T1 audit T2 compute T3 formal

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

The paper tries to show that quantum circuit partitioning for noisy devices should be guided by gate error rates, not just by circuit structure. It introduces Fidelipart, which turns each gate into a node of a hypergraph and adds two kinds of weighted hyperedges: one for multi-qubit interactions and one for the sequence of gates on each qubit, with weights derived from inverse error rates so that cutting a CNOT or splitting a qubit's timeline is costly. Fed to the Mt-KaHyPar partitioner, this representation produces fewer, more cohesive subcircuits. On three benchmark circuits (6, 10, and 24 qubits), the authors report 77.3–100% fewer estimated SWAP gates, up to 52.2% fewer cut qubits, and estimated fidelity gains from about 27% to over 250% (for example, 0.1724 to 0.5916 on the 6-qubit circuit). The price is a modest 8–13% increase in partitioning time and, in some cases, a deeper critical path within partitions.

What carries the argument

The central object is a fidelity-aware hypergraph. Each gate is a node with weight 10×(1/error) for CNOTs and 1×(1/error) for others; each multi-qubit gate gets a spatial hyperedge containing only its own node with weight 100×k×(1/error), and each qubit's timeline gets a temporal hyperedge over all its gate nodes with weight 100×⌊m/2⌋×(1/ϵ_H). These weights, normalized to a maximum of $10^{6}$, are fed to the Mt-KaHyPar partitioner with the k−1 (km1) metric and 5% imbalance, so that cutting a high-error or temporally dense connection is expensive. The paper argues the very large temporal weights (e.g., ~200,000 vs 4,000 for a CNOT context) intentionally preserve qubit coherence and thereby implicitly protect CNOTs. Subsequent stages merge partitions sharing many qubits and build a dependency DAG to fix execution order.

What would settle it

Re-run the comparison with QuickPartitioner configured to produce the same number of partitions (k=2,3,4) as Fidelipart on Circuits S, M, and L, keeping the same local re-mapping and error model; if the SWAP and fidelity gaps largely vanish at matched k, the claimed advantage comes from partition count and merging rather than error-aware weights.

Watch

Extended reading notes

Core claim

The paper's central claim is that fidelity-aware hypergraph partitioning improves NISQ circuit compilation: if a circuit's gates become nodes and its multi-qubit interactions and per-qubit gate histories become weighted hyperedges whose weights grow as gate error rates fall (so cutting a CNOT costs much more than cutting a Hadamard), then a hypergraph partitioner selecting cuts with minimum total weight will place partition boundaries around, rather than through, error-prone operations. On three benchmark circuits (6-qubit/22-gate, 10-qubit/55-gate, 24-qubit/88-gate) under a linear topology with identical local re-mapping, the authors report 77.3–100% fewer estimated SWAP gates (logical realignments), up to 52.2% fewer cut qubits, and estimated fidelity improvements such as 0.1724 to 0.5916 for the small circuit. The cost is an 8–13% increase in partitioning time and, in some circuits, an increase in maximum partition depth.

Load-bearing premise

The results credit fidelity-aware weighting for the gains, but Fidelipart also produces far fewer partitions (2–4) than QuickPartitioner (6–22) and applies a merging step, and the paper does not equalize partition counts to rule out a partition-count effect.

Editorial extensions

If this is right

Estimated SWAP gates drop by 77.3–100% and cut qubits by up to 52.2% on the three tested circuits, which would directly reduce communication-induced error under the paper's fidelity model.
Estimated fidelity improves from 0.1724 to 0.5916 for the 6-qubit circuit (about 243%), and the large SWAP reductions on the 10- and 24-qubit circuits imply comparable relative fidelity gains under the same error model.
The method costs 8–13% more partitioning time and can increase maximum partition depth, so it trades compile-time and depth for fewer inter-partition realignments.
Because the hypergraph model is topology-agnostic, the same weighting scheme can be applied to 2D grid or device-specific connectivities, with the SWAP model adjusted to physical distances.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

A natural next test is equalizing partition counts (k) between Fidelipart and QuickPartitioner, since the paper acknowledges the methods differ in number of partitions (2–4 vs 6–22) and does not control for this in the comparison.
An ablation that reduces or removes temporal hyperedge weights would reveal whether the gains come from preserving qubit timelines or from the CNOT penalty itself, given that temporal weights dominate the cost model.
The reported SWAP counts are lower bounds derived from local-index misalignment rather than full linear-chain routing distances, so actual physical SWAP overhead could be larger than the 77–100% reduction suggests.
The greedy merge heuristic is not globally optimal, so a threshold sweep or a different merge objective could shift the cut-qubit/SWAP trade-off, especially for densely connected circuits.

Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Desk Editor's Note

read the letter

Fidelipart is a reasonable incremental idea—weight hypergraph cuts by gate error rates before partitioning—but the paper's headline fidelity numbers don't survive contact with its own Eq. 1. For Circuits M and L, the reported fidelities imply ~190x and ~640x relative gains from SWAP differences, yet Table 1 shows 2.8x and 3.5x. The Figure 9 caption concedes the M/L absolute values 'would be lower' under epsilon_CNOT=0.05, which means those columns were not computed with the stated model. The abstract's '27.3% to over 250%' fidelity gain range is therefore unsupported for two of the three benchmarks.

What's genuinely useful: the hypergraph formulation itself (spatial single-node hyperedges for multi-qubit gates, temporal chains per qubit, weighted by assumed error rates) is a sensible encoding, and the detailed Circuit S walkthrough is transparent enough that the numbers check out—there, the 0.1724 to 0.5916 improvement is consistent with Eq. 1 given 8 vs 0 SWAPs. The authors also ship code on GitHub, which is more than many partitioning papers do.

The soft spots are real and load-bearing. First, besides the fidelity inconsistency, the comparison is confounded: Fidelipart outputs 2–4 partitions while QuickPartitioner outputs 6–22. The SWAP metric—logical misalignments between partitions—will mechanically favor fewer partitions. The authors acknowledge this in Section 5.4.1 but never control for it by, say, running QuickPartitioner at comparable k. Second, the weighting scheme has at least six hand-chosen constants (the 10x CNOT multiplier, C_g, C_t, merge threshold, block size, teleportation waiver probability) with no sensitivity analysis. The method might be robust, but nothing here demonstrates it. Third, three synthetic circuits and a simplified error model can't support broad NISQ claims.

Who is this for? Compiler researchers wanting a detailed example of error-aware partitioning might read it, but the quantitative section needs a rewrite. I'd send it to peer review because the core idea is testable and the code exists, but I'd expect major revision: recompute fidelity with Eq. 1, equalize partition counts, and add sensitivity analysis. As written, the central claims are not trustworthy.

Referee Report

4 major / 6 minor

Summary. The paper proposes Fidelipart, a quantum circuit partitioning framework that converts circuits into gate-node hypergraphs with error-rate-weighted spatial and temporal hyperedges, partitions them with Mt-KaHyPar, and then trims and optionally merges subcircuits. It compares against BQSKit's QuickPartitioner on three benchmark circuits (6, 10, and 24 qubits) under a linear topology and reports SWAP reductions of 77.3%–100%, cut-qubit reductions up to 52.2%, and estimated fidelity improvements, including an increase from 0.1724 to 0.5916 for the 6-qubit circuit. The central claim is that fidelity-aware hypergraph weighting is what drives these gains.

Significance. The problem of connectivity- and noise-aware circuit partitioning is timely, and the paper describes a complete pipeline with a source-code link, deterministic seeds, a detailed worked example for Circuit S, and an explicit list of limitations. If the central claims held, the idea of using weighted hypergraphs to bias cuts away from high-error operations would be a useful contribution to NISQ compilation. However, as written, the paper does not establish that claim: the implemented spatial weighting is inert under the km1 objective, the fidelity numbers for two of three circuits contradict the stated error model, and the comparison with QuickPartitioner is confounded by partition count.

major comments (4)

[§5.4.2, Table 1, Eq. (1)] The fidelity values for Circuits M and L are not outputs of the stated model. Because both methods conserve the original H and CNOT gates and differ only in attributed SWAP counts, Eq. (1) forces the fidelity ratio to be (1 − ε_CNOT)^{3(SWAP_Fidelipart − SWAP_Quick)}. For Circuit M the reported ratio 0.5583/0.1574 ≈ 3.55, but with ε_CNOT = 0.05 and SWAP counts 4 versus 46, the model requires (0.95)^{−126} ≈ 640. For Circuit L the reported ratio ≈ 2.79, while the model requires (0.95)^{−102} ≈ 190. The reported ratios instead correspond to ε_CNOT ≈ 0.01. Figure 9's caption acknowledges that the M/L absolute values 'would be lower' with ε_CNOT = 0.05, but Table 1 and Section 5.4.2 still present these entries as estimated fidelities under the stated consistent error model. The M/L fidelity claims must be recomputed with Eq. (1) and ε_CNOT = 0.05, or removed and replaced with the SWAP and cut-qubit metrics only.
[§3.2.3, Algorithm 4.1.3, Listing 2] The 'gate-level' spatial hyperedges for multi-qubit gates are singletons of the form {i}. Under the km1 objective that Mt-KaHyPar minimizes, a hyperedge whose pins all lie in one block is never cut, so the weight 4,000 assigned to each CNOT singleton can have no effect on the optimization. Consequently the paper's central mechanism—penalizing cuts through high-error CNOT gates—is not actually implemented. The only cut-relevant hyperedges are the temporal chains, whose weights in Algorithm 4.1.3 use ε_H uniformly and do not distinguish CNOTs from H gates. The authors should replace singleton hyperedges with hyperedges that connect multiple pins (e.g., the gate node together with its qubit-line nodes), or provide a formal argument explaining how singleton-edge weights affect the km1 solution.
[§5.4.1, Table 1] The comparison is confounded by the number of partitions. Fidelipart is run with k = 2, 3, and 4 partitions, while QuickPartitioner produces 6, 22, and 14 partitions for the same circuits. Since the SWAP metric counts logical misalignments across partition boundaries, the partition-count difference alone can explain much of the reported reduction in SWAPs and, through Eq. (1), in estimated fidelity. Section 5.4.1 acknowledges the partition-count difference but does not control for it. The authors should add experiments with QuickPartitioner constrained to a comparable k, with Fidelipart run at k matched to QuickPartitioner, and an ablation using uniform hyperedge weights, so that the gains can be attributed to fidelity-aware weighting rather than to partition granularity.
[§4.6.3, Eq. (1)] The SWAP metric contains an ad hoc 'teleportation heuristic': once a global qubit incurs more than three SWAP misalignments, subsequent SWAP requirements for that qubit are waived with probability 0.6. This rule is not derived from any concrete teleportation protocol or hardware cost model, and it can differentially reduce the baseline's SWAP count, especially because QuickPartitioner produces many small partitions. Since the SWAP counts enter directly into Eq. (1), the heuristic propagates into the headline fidelity gains. The authors should either remove the heuristic and report raw misalignment counts, or replace it with a physically justified teleportation cost model.

minor comments (6)

[§3.4.1, §4.4.1] Both sections contain unresolved 'Section??' cross-references; these should be fixed before resubmission.
[§4.5] The pseudocode for build_dependency_graph is reproduced twice with only cosmetic differences; the duplicate block should be removed.
[Table 1] The table as typeset is very difficult to read because the columns are run together; it should be reformatted so that the reader can distinguish the Method, Target k, Actual k, Cut Qubits, SWAP Gates, Fidelity, Max Depth, and Time columns.
[§5.2.3, §5.3.2] The walkthrough merges the two Circuit S partitions into one partition but then proceeds with the unmerged partitions; the paper should state explicitly whether the reported Table 1 results use the merging stage, and if so, with what threshold.
[Listing 2] The caption says the hMETIS weights are normalized but the listed values appear to be pre-normalization; clarify which quantities are actually written to the hypergraph file.
[§5.4.3] Fidelipart increases maximum partition depth for Circuits S and L, but Eq. (1) does not include depth or decoherence; the paper should at least acknowledge that longer critical paths could offset fidelity gains on real hardware.

Circularity Check

0 steps flagged · score 0.0 of 10

No circularity: fidelity-aware weights and the fidelity metric share an error model, but the optimization objective and the headline SWAP/cut-qubit metrics are independent; the Table 1 fidelity inconsistencies are correctness issues, not circular reductions.

full rationale

Fidelipart's derivation chain is self-contained. Hypergraph node and hyperedge weights are fixed functions of declared error rates and structural counts (Sections 3.2.3, 4.1), with no parameter fitted to the benchmarks. The reported SWAP and cut-qubit metrics are computed by independent routines (Section 4.6.3) from the final partitions, not read off the Mt-KaHyPar objective. Using the same epsilon_CNOT=0.05 for weighting and for fidelity evaluation is modeling consistency rather than circularity: the partitioner minimizes the weighted km1 cut, which is not identical to the product-of-errors fidelity formula in Eq. (1), and the SWAP/cut reductions are logically prior to the fidelity estimate. The comparison is not controlled for partition count (Fidelipart generates 2-4 partitions versus 6-22 for QuickPartitioner), so attributing all gains to error-aware weighting is a confound; and the M/L fidelity columns in Table 1 are internally inconsistent with Eq. (1) given the stated SWAP counts — indeed, the Figure 9 caption admits that for M and L 'absolute values would be lower with epsilon_CNOT=0.05'. These are significant correctness and experimental-control concerns, but they are not cases where a claimed prediction reduces to its inputs by construction, by definition, or through a load-bearing self-citation. The citations to BQSKit, KaHyPar, and Mt-KaHyPar are external tools, not self-citations, so no self-citation chain forces the result.

Assumptions & free parameters 7 free parameters · 5 assumptions · 0 invented entities

The central claim rests on a large set of hand-chosen constants (weight multipliers, scaling factors, merge threshold, block sizes, teleportation waiver) and on a simplified error model. These parameters are not derived from first principles or tuned with sensitivity analysis. The assumptions about topology, fidelity model, and solver quality are stated but not independently verified.

free parameters (7)

CNOT node weight multiplier 10 = 10
Section 4.1.2: CNOT nodes weighted 10*(1/epsilon_CNOT); the 10x emphasis is hand-chosen.
Spatial hyperedge scaling factor C_g = 100
Section 3.2.3: multi-qubit hyperedge weight = 100*k*(1/error_rate); scaling factor is ad hoc.
Temporal hyperedge scaling factor C_t = 100
Section 3.2.3: temporal chain weight = 100*floor(m/2)*(1/epsilon_H); scaling factor is ad hoc.
Merge threshold = 3 (default)
Section 4.4.2: partitions sharing at least threshold qubits are merged; default set by invocation without analysis.
Block size B in k formula = 4, 6, 8
Section 5.1: block_size=4 for Circuit S, 6 for M, 8 for L; directly determines target partition count.
Teleportation waiver probability = 60%
Section 4.6.3: if a global qubit incurs more than 3 misalignments, subsequent SWAPs are waived with 60% probability; random.seed(42) is used.
Weight normalization scaling factor = 1e6
Section 4.1.4: hMETIS weights scaled to max 1e6; arbitrary and affects relative penalties.

assumptions (5)

domain assumption Linear chain hardware topology, where only adjacent qubits interact directly.
Section 3.4.1; all benchmarks assume this topology and all SWAP cost estimates depend on it.
domain assumption Circuit fidelity is the product of per-gate survival probabilities with epsilon_H=0.001, epsilon_CNOT=0.05, and SWAP modeled as three CNOTs.
Section 4.6.3, Eq. 1; the evaluation metric is a simplified model not validated against hardware or simulation.
domain assumption Mt-KaHyPar with km1 objective and 5% imbalance produces partitions that are near-optimal for the weighted cut.
Section 4.2; the solver's heuristic quality is assumed without independent verification of optimality.
domain assumption Local contiguous re-mapping is an unbiased post-processing procedure for both methods.
Section 4.6; the fairness of the comparison rests on this assumption, which is not separately validated.
domain assumption Scaling hyperedge weights to a maximum of 1e6 preserves the rank ordering of desired penalties.
Section 4.1.4; the normalization is required for hMETIS format but its effect on partitioner behavior is not analyzed.

how reviews work

0 comments

Cite this review

Pith. "Pith review of A fidelity-driven approach to quantum circuit partitioning via weighted hypergraphs for noise-resilient computation." pith.science (2026). https://pith.science/paper/H4ROBEJN

@misc{pith2026250606867,
  author       = {Pith},
  title        = {Pith review of: A fidelity-driven approach to quantum circuit partitioning via weighted hypergraphs for noise-resilient computation},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/H4ROBEJN}},
  note         = {Machine review of arXiv:2506.06867}
}

read the original abstract

Effective circuit partitioning is critical for Noisy Intermediate-Scale Quantum (NISQ) devices, which are hampered by high error rates and limited qubit connectivity. Standard partitioning heuristics often neglect gate-specific error impacts, leading to suboptimal divisions with significant communication overhead and reduced fidelity. This paper introduces Fidelipart, a novel framework that transforms quantum circuits into a fidelity-aware hypergraph. In this model, gate error rates and structural dependencies inform the weights of nodes (gates) and hyperedges (representing multi-qubit interactions and qubit timelines), guiding an Mt-KaHyPar partitioner to minimize cuts through error-prone operations. We evaluated Fidelipart against BQSKit's QuickPartitioner on 6-qubit/22-gate, 10-qubit/55-gate, and 24-qubit/88-gate benchmarks under a linear topology with a consistent local contiguous re-mapping strategy. Results demonstrate Fidelipart's superior performance. It achieved SWAP gate reductions ranging from 77.3% to 100% and up to a 52.2% decrease in cut qubits. These physical improvements directly translated to estimated fidelity gains ranging from 27.3% to over 250%. While Fidelipart showed a modest runtime increase of 8-13% and variable effects on maximum partition depth, its substantial enhancement of circuit fidelity highlights the significant benefits of integrating detailed error-awareness into the partitioning process for more reliable NISQ computations.

Figures

Figures reproduced from arXiv: 2506.06867 by the authors.

**Figure 1.** Workflow of the Fidelipart quantum circuit partitioning framework. The input quantum circuit is first transformed into a fidelity-aware hypergraph, where nodes represent gates and hyperedges (capturing multiqubit interactions and temporal dependencies) are assigned error-aware weights that penalize high-error operations. This weighted hypergraph is then partitioned using Mt-KaHyPar, optimizing the km1 metric to m… view at source ↗

**Figure 2.** Hypergraph representation of a 5-qubit circuit [PITH_FULL_IMAGE:figures/full_fig_p007_2.png] view at source ↗

**Figure 4.** Subcircuits for a 5-qubit circuit with 10 gates, [PITH_FULL_IMAGE:figures/full_fig_p008_4.png] view at source ↗

Figures from the paper (7 more)

**Figure 5.** Figure 5: Conceptual visualization of a fidelity-aware hy [PITH_FULL_IMAGE:figures/full_fig_p024_5.png]

**Figure 6.** Figure 6: Dependency graph for the two partitions of [PITH_FULL_IMAGE:figures/full_fig_p025_6.png]

**Figure 7.** Figure 7: Comparison of Global Cut Qubits for Fidelipart vs. QuickPartitioner across benchmark circuits (“S” denotes Small, “M” Medium, and “L” Large). The reduction in cut qubits achieved by Fidelipart directly translates to a significant decrease in the estimated SWAP gate ove…

**Figure 8.** Figure 8: Comparison of Estimated SWAP Gates (Logical Realignments) for Fidelipart vs. QuickPartitioner across benchmark circuits (“S” denotes Small, “M” Medium, and “L” Large). 5.4.2 Estimated Circuit Fidelity The efficacy of a partitioning strategy is ultimately reflected in …

**Figure 9.** Figure 9: Comparison of Estimated Circuit Fidelity. For [PITH_FULL_IMAGE:figures/full_fig_p029_9.png]

**Figure 10.** Figure 10: Comparison of Maximum Partition Depth for Fidelipart vs. QuickPartitioner across benchmark circuits (“S” denotes Small, “M” Medium, and “L” Large). 5.4.4 Computational Performance: Partitioning Time The computational cost of the partitioning process itself is an imp…

**Figure 11.** Figure 11: Comparison of Partitioning Time for Fidelipart vs. QuickPartitioner across benchmark circuits (“S” denotes Small, “M” Medium, and “L” Large). 5.5 Summary of Key Findings The comparative analysis across the benchmark circuits—“Circuit S” (6Q, 22G), “Circuit M” (10Q, …

Discussion (0). Sign in to comment.

Reference graph

Works this paper leans on

14 extracted references · 13 canonical work pages

[1]

Quantum computation and quantum in- formation

Michael A. Nielsen and Isaac L. Chuang. “Quantum computation and quantum in- formation”. Cambridge University Press. (2010)

work page 2010
[2]

Quantum computing in the nisqera andbeyond

John Preskill. “Quantum computing in the nisqera andbeyond”. Quantum2, 79(2018)

2018
[3]

Ground-state energy estimation of the wa- ter molecule on a trapped-ion quantum computer

Yunseong Nam, Jwo-Sy Chen, Neal C. Pisenti, Kenneth Wright, Conor Delaney, Dmitri Maslov, Joel M. Smith, Ken- neth R. Kim, and Christopher Monroe. “Ground-state energy estimation of the wa- ter molecule on a trapped-ion quantum computer”. npj Quantum Information6, 33 (2020)

work page 2020
[4]

Noise-adaptive compiler mappings for noisy intermediate-scale quan- tum computers

Prakash Murali, Jonathan M. Baker, Ali Javadi-Abhari, Frederic T. Chong, and Mar- garet Martonosi. “Noise-adaptive compiler mappings for noisy intermediate-scale quan- tum computers”. In Proceedings of the 24th International Conference on Architec- tural Support for Programming Languages and Operating Systems (ASPLOS). Pages 1015–1029. (2019)

work page 2019
[5]

Tack- ling the qubit mapping problem for nisq-era quantum devices

Gushu Li, Yufei Ding, and Yuan Xie. “Tack- ling the qubit mapping problem for nisq-era quantum devices”. In Proceedings of the 24th International Conference on Architec- tural Support for Programming Languages and Operating Systems (ASPLOS). Pages 1001–1014. (2019)

work page 2019
[6]

On the qubit routing problem

Alexander Cowtan, Silas Dilkes, Ross Dun- can, Alexandre Krajenbrink, Will Simmons, and Seyon Sivarajah. “On the qubit routing problem” (2019)

work page 2019
[7]

Berkeley Quantum Synthesis Toolkit (BQSKit)

Ed Younis, Costin C. Iancu, Wim Lavrijsen, Marc Davis, and Ethan Smith. “Berkeley Quantum Synthesis Toolkit (BQSKit)”. Lawrence Berkeley National Labora- tory (2021). Computer software. Version v1. 35

work page 2021
[8]

High- quality hypergraph partitioning

Sebastian Schlag, Tobias Heuer, Lars Gottesbüren, Yaroslav Akhremtsev, Chris- tian Schulz, and Peter Sanders. “High- quality hypergraph partitioning” (2021)

work page 2021

Show all 14 references

[9]

Scalable shared-memory hyper- graph partitioning

Tobias Heuer, Lars Gottesbüren, Peter Sanders, Sebastian Schlag, and Christian Schulz. “Scalable shared-memory hyper- graph partitioning”. In Martin Farach- Colton and Sabine Storandt, editors, Pro- ceedings of the 2021 Workshop on Algorithm Engineering and Experiments (ALENEX...

2021
[10]

Automated distribution of quantum circuits via hypergraph partitioning

Pablo Andrés-Martínez and Chris Heunen. “Automated distribution of quantum circuits via hypergraph partitioning”. Physical Re- view A100, 032308 (2019)

2019
[11]

Qiskit: An open-source framework for quantum computing

Qiskit Community. “Qiskit: An open-source framework for quantum computing” (2019)

2019
[12]

ScaffCC: A framework for compilation and analysis of quantum computing programs

Ali Javadi-Abhari, Shruti Patil, Daniel Kudrow, Jeff Heckey, Alexey Lvov, Fred- eric T. Chong, and Margaret Martonosi. “ScaffCC: A framework for compilation and analysis of quantum computing programs”. In Proceedings of the 11th ACM Confer- ence on Computing Frontiers. Pages 1...

2014
[13]

Ibm quantum experience - architecture

IBM Quantum Experience. “Ibm quantum experience - architecture” (2021). Accessed: 2025-05-25

2021
[14]

Rigetti quantum pro- cessor architecture

Rigetti Computing. “Rigetti quantum pro- cessor architecture” (2021). Accessed: 2025- 05-25. 36

2021

Pith tools

Reviewed August 7, 2026 · model on record in the stance chip above.

[1] [1]

Quantum computation and quantum in- formation

Michael A. Nielsen and Isaac L. Chuang. “Quantum computation and quantum in- formation”. Cambridge University Press. (2010)

work page 2010

[2] [2]

Quantum computing in the nisqera andbeyond

John Preskill. “Quantum computing in the nisqera andbeyond”. Quantum2, 79(2018)

2018

[3] [3]

Ground-state energy estimation of the wa- ter molecule on a trapped-ion quantum computer

Yunseong Nam, Jwo-Sy Chen, Neal C. Pisenti, Kenneth Wright, Conor Delaney, Dmitri Maslov, Joel M. Smith, Ken- neth R. Kim, and Christopher Monroe. “Ground-state energy estimation of the wa- ter molecule on a trapped-ion quantum computer”. npj Quantum Information6, 33 (2020)

work page 2020

[4] [4]

Noise-adaptive compiler mappings for noisy intermediate-scale quan- tum computers

Prakash Murali, Jonathan M. Baker, Ali Javadi-Abhari, Frederic T. Chong, and Mar- garet Martonosi. “Noise-adaptive compiler mappings for noisy intermediate-scale quan- tum computers”. In Proceedings of the 24th International Conference on Architec- tural Support for Programming Languages and Operating Systems (ASPLOS). Pages 1015–1029. (2019)

work page 2019

[5] [5]

Tack- ling the qubit mapping problem for nisq-era quantum devices

Gushu Li, Yufei Ding, and Yuan Xie. “Tack- ling the qubit mapping problem for nisq-era quantum devices”. In Proceedings of the 24th International Conference on Architec- tural Support for Programming Languages and Operating Systems (ASPLOS). Pages 1001–1014. (2019)

work page 2019

[6] [6]

On the qubit routing problem

Alexander Cowtan, Silas Dilkes, Ross Dun- can, Alexandre Krajenbrink, Will Simmons, and Seyon Sivarajah. “On the qubit routing problem” (2019)

work page 2019

[7] [7]

Berkeley Quantum Synthesis Toolkit (BQSKit)

Ed Younis, Costin C. Iancu, Wim Lavrijsen, Marc Davis, and Ethan Smith. “Berkeley Quantum Synthesis Toolkit (BQSKit)”. Lawrence Berkeley National Labora- tory (2021). Computer software. Version v1. 35

work page 2021

[8] [8]

High- quality hypergraph partitioning

Sebastian Schlag, Tobias Heuer, Lars Gottesbüren, Yaroslav Akhremtsev, Chris- tian Schulz, and Peter Sanders. “High- quality hypergraph partitioning” (2021)

work page 2021