Quantum-assisted h{\lambda}-adaptive finite element method

H. A. Shynkarenko; R. H. Drebotiy

REVIEW 4 major objections 5 minor 23 references

Quantum-assisted h{\lambda}-adaptive finite element method

T0 review · 4 major / 5 minor · reviewed 2026-08-12 · deepseek-v4-flash

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

Pith's one-line read The paper proves an a posteriori error bound for a λ-regularized adaptive finite element method and reports that it reaches 0.2% estimated relative error with 127 accumulated degrees of freedom, versus 541 for classical h-adaptivity.

desk verdict A genuinely new hλ-adaptive idea, honestly presented, but Theorem 1 has a factor-α error that must be corrected before the error-control claims can be trusted. read the letter →

arxiv 2411.12687 v1 pith:VQUZHV4B submitted 2024-11-19 math.NA cs.NA

R. H. Drebotiy , H. A. Shynkarenko This is my paper · ORCID

classification math.NAcs.NA MSC 65N3065N1565N5035B25

keywords finiteelementmethodadvection-diffusion-reactionsingularlyperturbedproblemh-adaptiveschemeaposteriorierrorestimateregularizationparameterquantumlinearsystemSWAP-test

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 establish a practical route to accurate finite element approximations of singularly perturbed advection-diffusion-reaction problems, whose solutions have sharp boundary layers that make standard methods oscillate. Its proposal is to regularize the variational problem by adding a λ-weighted Sobolev inner product tied to the reduced Cauchy-problem solution, choose λ by minimizing a heuristic oscillation-loss function H(λ), and then run classical h-adaptive refinement under a new a posteriori error estimate. If the Theorem 1 estimate is correct, the hλ-scheme gives automatic error control and, in the reported 1D experiment, reaches 0.2% estimated relative error with 127 accumulated degrees of freedom, fewer than the 541 used by classical h-adaptivity. The quantum-assisted part is that λ-selection could in principle be done by an HHL quantum linear-system solver combined with a SWAP-test, though all numerics here were run classically.

What carries the argument

The load-bearing objects are the regularized variational problem (8), the oscillation loss function $H(\lambda)$ of Eq. (16), the a posteriori error estimate of Theorem 1 (Eq. 26), and the local error indicator $\eta_K$ of Eq. (34). The regularization replaces the original bilinear form by $a(u,v)+\lambda(u,v)_V = l(v)+\lambda(u_0,v)_V$, where $u_0$ solves the reduced advection-reaction Cauchy problem. $H(\lambda)$ is a normed scalar product between second divided differences of the finite element solution and the alternating pattern $(1,-1,1,\dots)$, designed so its minimum marks the λ at which parasitic oscillations disappear; it is the quantity that a quantum computer would estimate via HHL and SWAP-test. Theorem 1 splits the total error into a known λ-term and the regularized energy-norm error, which the residual estimator bounds element-wise, giving $\eta_K$ for mesh refinement.

What would settle it

Run the scheme on a 1D singularly perturbed advection-diffusion-reaction problem with an initial mesh below the threshold mentioned in Remark 8 and record the graph of $H(\lambda)$: if it shows two local minima of comparable depth, or if the minimizer changes discontinuously with a small change in the initial mesh, the uniqueness claim fails. Alternatively, compute the true error $\|u-u_h^{\lambda^*}\|_V$ on a fine reference mesh while varying λ; if the minimum of $H(\lambda)$ and the minimum of the true error occur at substantially different λ, the heuristic selection is not a reliable proxy.

Watch

Extended reading notes

Core claim

The paper's central claim is Theorem 1 (Eq. 26): for the regularized finite element solution $u_h^\lambda$ of the singularly perturbed advection-diffusion-reaction problem, the error against the true solution $u$ in the Sobolev norm is controlled by two computable terms, $\|u - u_h^\lambda\|_V^2 \le \frac{2\lambda^2}{\alpha^2}\|u_0 - u_h^\lambda\|_V^2 + \frac{2(\alpha+\lambda)}{\alpha}\|u_\lambda - u_h^\lambda\|_{E,\lambda}^2$, where $u_0$ is the reduced Cauchy-problem solution and the second norm is the energy norm of the regularized problem. Because every quantity on the right is known or estimable, this estimate supports a practical adaptive loop: compute the stabilized approximation, check the relative error indicator, bisect elements where the local indicator $\eta_K$ is large, and repeat. The paper further claims that the loss function $H(\lambda)$, a divided-difference oscillation detector, selects a λ that removes high-frequency oscillations, and reports that this hλ-adaptive scheme reaches 0.2% estimated relative error in 5 iterations with 127 accumulated degrees of freedom, while classical h-adaptivity needs 8 iterations and 541, and an hp-scheme does not converge monotonically on this test.

Load-bearing premise

The method depends on the unproven assumption that the oscillation-loss function $H(\lambda)$ has one clear minimum and that the λ at that minimum removes the spurious oscillations; the paper's own Remark 8 concedes that the parameter choice fails when the starting mesh is too coarse.

Editorial extensions

If this is right

On the reported 1D test, the hλ-scheme reaches the 0.2% tolerance in 5 iterations with 127 accumulated degrees of freedom, versus 8 iterations and 541 for classical h-adaptivity.
The first hλ iteration already gives a qualitatively usable approximation, whereas the first iterations of the h- and hp-schemes are strongly oscillatory.
The Theorem 1 estimate holds for problems of type (1) and (8) in any space dimension, so the error-control mechanism is not tied to the 1D implementation.
The number of quantum solver calls needed for λ selection is bounded by a constant depending only on the estimation precision and $\lambda_{\max}$, not on the mesh size.
Because refinement is performed separately inside and outside the boundary layer, the scheme avoids excessive refinement of the layer itself.

Reading between the lines

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

Beyond the paper: the numerical comparison reports accumulated degrees of freedom, not runtime or quantum circuit depth; a cost comparison that counts actual linear solves or wall-clock time could change the apparent advantage.
Beyond the paper: the uniqueness of the minimum of $H(\lambda)$ is asserted heuristically; testing whether coarse initial meshes of the kind mentioned in Remark 8 produce multiple local minima would decide whether the λ-selection step needs a more robust global optimizer.
Beyond the paper: the same oscillation-loss idea could be applied to time-dependent or nonlinear problems where spurious high-frequency oscillations also appear, but the a posteriori estimate would need to be re-derived for each new setting.
Beyond the paper: the loss function excludes the boundary layer under the assumption that its location is known; an automatic layer detector for general 2D domains would be needed before the scheme can be used without user input.

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

Colleague,

The honest headline: the paper has a real, modestly new idea, but the central a posteriori theorem as displayed is wrong, and the numerical evidence is thinner than the claims. The new pieces are the hλ-adaptive loop (Algorithm 2), the generalized divided-difference loss H(λ) for non-uniform meshes (16), and the local indicator η_K (34). Those are genuine extensions of the authors' earlier work, and they are presented clearly. The citation pattern is fine; the paper leans on their own prior regularization scheme, which is reasonable.

What is done well: the a posteriori framework is a standard residual-estimator structure, the 1D specialization is concrete, and the paper is honest about limitations—Remark 8 concedes the λ-selection fails on too-coarse initial meshes, and Remark 4 states that the boundary layer is excluded from error control. That honesty is real.

The soft spot is not minor. Theorem 1 (26) as written has a factor α error. Substituting (24) into (23) gives second coefficient 2(α+λ)/α², not 2(α+λ)/α. For the singularly perturbed regime, α = min{μ,σ} can be small, and the stated inequality can be violated by a large factor. This is an internal inconsistency in the central theorem. The saving grace is that the practical bound (33) and the indicators built from it use the corrected coefficient, so the intended error estimator does not inherit the error. That makes the theorem fixable, but the paper must be revised before the error-control guarantee is taken seriously.

Numerics: one 1D test case, no true-error comparison, and the h-/hp- comparisons use different estimators and different error scales, so the 127-vs-541 d.o.f. comparison is not apples-to-apples. It is suggestive, not evidence of superiority. The quantum-assisted part is all classical in the experiment; HHL+SWAP is speculative.

Who is it for: readers working on stabilized adaptive FEM for singular perturbation problems. If the theorem is corrected and validation broadened, it could be a useful tool. As is, I would not cite it for the error bound or the efficiency claim.

Recommendation: it deserves a serious referee, with explicit instruction to check Theorem 1. A competent referee will find the factor α quickly. Not a desk reject, but condition on correction.

Referee Report

4 major / 5 minor

Summary. The paper proposes an hλ-adaptive finite element method for singularly perturbed advection-diffusion-reaction equations. The idea is to regularize the variational problem with an additional term λ(u0−u) in the V-scalar product, solve the regularized problem by linear finite elements, select λ by minimizing a heuristic oscillation-detection loss function H(λ), and drive h-refinement by a local residual-type error indicator. The authors state a general a posteriori error estimate (Theorem 1, Eq. (26)) and use it to formulate Algorithm 2. They report one 1D numerical experiment in which the hλ-scheme reaches an estimated 0.16% relative error with 127 accumulated degrees of freedom, versus 541 for a classical h-adaptive scheme.

Significance. If the a posteriori bound and the λ-selection heuristic were rigorously justified, the core idea—combining regularization with adaptive mesh refinement and possibly using a quantum computer (HHL plus SWAP-test) only to estimate the stabilization parameter—would be interesting and potentially useful for singularly perturbed problems. The paper is transparent about many of its limitations, including the heuristic nature of the loss function and the failure on too-coarse initial meshes. However, the central theorem contains an algebraic coefficient error, and the error estimator explicitly excludes the boundary layer from the error control. As a result, the headline numerical comparison (127 versus 541 accumulated degrees of freedom) is not supported by a valid global error estimate. The significance of the contribution is therefore limited until these issues are resolved.

major comments (4)

[§5, Eq. (26) and Eq. (33)] The proof of Theorem 1 is not valid as written. From Eq. (23) and the norm relation in Eq. (24), the second coefficient on the right-hand side of Eq. (25) should be (1+λ/α)(α+λ)^(−1/2) = √(α+λ)/α, whose square is (α+λ)/α². The displayed Theorem 1 in Eq. (26) instead has 2(α+λ)/α, which is a factor of α too large. The same error propagates into Eq. (33), where the residual term should have coefficient 4/α² rather than 4/α, and then into the stopping criterion (37). Since α = min{μ,σ} is small in the singularly perturbed regime targeted by the paper, the claimed a posteriori bound is not established, and the reported 0.16% error is not a valid consequence of the proof.
[§6, Remark 4 and Eq. (37)] The error estimator (37) and the refinement loop in Algorithm 2 use only the elements in the set Left (x ≤ x_layer); the boundary layer region (Right) is explicitly excluded from the error. Consequently, the computed 'Error' is not a bound on ||u−u_h^λ||_V over the whole domain Ω. For a singularly perturbed problem, the boundary layer is precisely where the largest gradients and potentially the largest approximation error occur. The abstract and conclusions claim automatic error control, but the method as presented controls only the smooth part of the solution outside the layer. This limitation must be stated in the main claims, or the estimator and stopping criterion must be amended to cover the full domain.
[§4, Eq. (16) and Remark 8] The selection of the regularization parameter λ* is load-bearing for the method's efficiency, but it is heuristic. The paper asserts that the loss function H(λ) in Eq. (16) has a unique local minimum and that this minimum corresponds to the elimination of high-frequency oscillations, citing the authors' previous work [8] and numerical experiments. No proof of uniqueness, nor a quantitative link between H(λ) and the true approximation error, is provided in this manuscript. Remark 8 explicitly concedes that the parameter-estimation procedure fails when the initial mesh is too coarse. Given that the subsequent numerical claims depend on the chosen λ*, this gap should be addressed, at minimum by a sensitivity study over λ and over the initial number of elements N.
[§7, Tables 1–3] The numerical validation is too narrow to support the general claims. Only one 1D test problem is considered, and the reported 'Error' is the estimator in Eq. (37), not the true global error on the whole domain. The comparison with h- and hp-adaptivity is also not apples-to-apples: the hλ error includes an extra λ-term and excludes the boundary layer, while the other schemes use different estimators. In addition, the quantum part is performed by classical simulation, so the stated numbers of quantum solver calls do not demonstrate quantum-hardware performance. A more thorough validation, including the true error in the V-norm over the entire domain and more than one test case, is needed before the 127-versus-541 degrees-of-freedom claim can be accepted.

minor comments (5)

[§5, Eq. (22)] Equation (22) is garbled in the manuscript and does not read as a mathematical statement; please rewrite it so that the triangle-inequality step leading to Eq. (23) is clear.
[§4, Eq. (16)] The text says 'we propose to use the following modification of a quantity (9)', but Eq. (9) is the closed-characteristic example from Remark 1. The intended reference is likely Eq. (14), the central-difference quantity on uniform meshes.
[§4, Eqs. (13)–(17)] The index conventions in the summation in Eq. (13) and the divided-difference definition in Eq. (17) are not fully specified; please state exactly which nodes are used and how the boundary-layer exclusion is reflected in the summation limits.
[§6, Remark 5] Remark 5 assumes that the optimal λ decreases on mesh refinement; this is stated as a theoretical expectation, but it is not proven. Since the algorithm may rely on this monotonicity when reusing the previous λ* as the new λ_max, the assumption should be labeled explicitly as a heuristic.
[§7, Tables 1–3] The 'accumulated d.o.f.' metric is the sum of the degrees of freedom over all iterations. This is not a standard complexity measure for adaptive algorithms and should be defined and justified, or replaced by the final mesh size and the computational cost per iteration.

Circularity Check

1 steps flagged · score 4.0 of 10

Partial circularity: optimal-λ selection is imported from the authors' own prior heuristic, while the a posteriori error bound itself is derived independently.

uniqueness imported from authors [Section 4, paragraph after Eq. (13); relied on by Algorithm 1 step 1 for H(λ) of Eq. (16)]
"This leads us to a function, which has unique local minimum and this minimum is reached for the optimal parameter value, for which high-frequency oscillations are excluded from the approximate solution, which is shown, in particular, by numerical experiments in [8]."

The parameter-selection step is load-bearing: Algorithm 1 obtains λ* by minimizing H(λ), and the reported efficiency (127 accumulated d.o.f. vs 541) depends on that choice. The uniqueness/oscillation-elimination property of the loss function is not proved here; it is asserted for F(13) on the basis of the authors' own [8] and supported there only by numerical experiments, while H(16) is introduced only as a 'direct generalization' with no new proof. Thus the choice that makes the method work is carried by self-citation rather than by the paper's error estimates. Not full circularity because Theorem 1 (26) holds for arbitrary λ and does not use H; the a posteriori control chain remains independent.

full rationale

Most of the derivation is self-contained: Theorem 1 (Eq. 26) follows from the variational equations, triangle inequality, and energy-norm equivalence, and the local indicators (34) and stopping criterion (37) are computable expressions derived from that bound, not quantities fitted to the target error. The numerical 'Error' reported in Table 1 is the estimator value used by the stopping test, which is a standard a posteriori practice and not circular by itself. The one genuine circularity-adjacent element is the λ-selection heuristic: the paper imports from the authors' prior [8] the claim that the loss function has a unique local minimum whose argmin removes oscillations, and Algorithm 2's success rests on that unproved, self-cited premise. Since the central error-control theorem does not depend on this heuristic, the circularity score is moderate (4), not higher. Note also that the displayed Theorem 1 appears to contain an algebraic factor error (the squared second coefficient should be (α+λ)/α² rather than (α+λ)/α); that is a correctness defect, not a circularity, and would need correction independently of this assessment.

Assumptions & free parameters 7 free parameters · 6 assumptions · 0 invented entities

The central claim rests on one tuned parameter λ, several hand-chosen algorithm constants, and five background assumptions. The most fragile assumptions are the well-posedness of the reduced problem (no closed integral curves) and the heuristic uniqueness of the minimum of H(λ). No new physical or mathematical entities are introduced; the quantum part is an implementation choice rather than a new entity.

free parameters (7)

Regularization parameter λ = Not reported; chosen per iteration by minimizing H(λ) (Eq. 16)
Central tuning knob of the stabilization; its optimal value is found by numerical optimization, not derived.
λ_max upper bound = Formula (15): constant proportional to diam(Ω)||β||∞/n
Heuristic upper bound for the line search; hand-chosen constant.
Refinement threshold θ = 2/3
Hand-chosen marking threshold in Algorithm 2, step 6.
Error tolerance Tol = 0.2%
Hand-chosen stopping criterion in Algorithm 2.
Initial mesh count N = 15
Hand-chosen in Section 7; Remark 8 warns performance depends on N.
Boundary layer separation node layer_x = pre-last node
Assumes the boundary layer is at x=1 and excludes it from the loss and error functionals; ad hoc for the test problem.
Precision for parameter estimation = 1
From [8]; controls the number of quantum solver calls (Remark 9).

assumptions (6)

standard math Lax-Milgram conditions hold for the variational ADR problem (2).
Invoked in Section 2 to guarantee well-posedness and coercivity (20).
domain assumption The advection field β has no closed integral curves entirely inside Ω.
Stated in Section 3 to make the reduced Cauchy problem (7) well-posed; otherwise (7) is not correctly formulated (Remark 1).
ad hoc to paper The loss function H(λ) has a unique local minimum corresponding to the oscillation-free approximation.
Assumed in Section 4 and Remark 2, based on heuristic evidence in [8]; no proof is provided in this paper.
standard math The residual error estimator from [9] is valid for the regularized problem (27).
Used in Section 5, Eqs. (29)-(30) without re-derivation; a cited external result.
ad hoc to paper The boundary layer is located near x=1 and can be excluded from the loss and error functionals.
Remark 4 and Algorithm 2 assign layer_x to the pre-last node; the method does not handle inner layers (stated in the Introduction).
ad hoc to paper The optimal λ decreases on mesh refinement, so the previous λ can be used as the new upper bound.
Remark 5 states this as a heuristic optimization option, not a theorem.

how reviews work

0 comments

Cite this review

Pith. "Pith review of Quantum-assisted h{\lambda}-adaptive finite element method." pith.science (2026). https://pith.science/paper/VQUZHV4B

@misc{pith2026241112687,
  author       = {Pith},
  title        = {Pith review of: Quantum-assisted h\lambda-adaptive finite element method},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/VQUZHV4B}},
  note         = {Machine review of arXiv:2411.12687}
}

read the original abstract

We propose a novel finite element method scheme for singularly perturbed advection-diffusion-reaction problems, which combines certain quantum-assisted stabilization scheme with a classical h-adaptive approach to provide automatic error control and corresponding approximation refinement. Appropriate finite element a posteriori error estimates are proved. Described approach demonstrates the possibility to overcome singular perturbations by applying error-controlled smoothening to finite element approximation. Possible benefits of the proposed finite element scheme are discussed and the numerical comparison with the typical adaptive scheme is provided.

Figures

Figures reproduced from arXiv: 2411.12687 by the authors.

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

23 extracted references · 20 canonical work pages

[8]

Heuristic choice of the regularization parameter for optimal stabilization of the finite element approximations / R

Drebotiy R. Heuristic choice of the regularization parameter for optimal stabilization of the finite element approximations / R. Drebotiy, H. Shynkarenko // Mathematical Methods and Physicomechanical Fields. – Pidstryhach Institute for Applied Problems of Mechanics and Mathematics, National Academy of Sciences of Ukraine. – Lviv 2023. – Vol.66, №1-2, pp.206-221

work page 2023
[1]

Numerical Approximation of Partial Differential Equations / S

Bartels S. Numerical Approximation of Partial Differential Equations / S. Bartels. – Springer, 2016. – 541p

work page 2016
[2]

Variational Quantum Linear Solver / LaRose R., Cerezo M., Subasi Y., Cincio L., Coles P

Bravo-Prieto C. Variational Quantum Linear Solver / LaRose R., Cerezo M., Subasi Y., Cincio L., Coles P . // ISSN: 2521- 327X, Quantum 7, 1188 (2023)

work page 2023
[3]

The Mathematical Theory of Finite Element Methods

Brenner S., Scott L. The Mathematical Theory of Finite Element Methods. Springer, 2008, 3ed. - 404 p

work page 2008
[4]

Quantum Fingerprinting

Buhrman H., Cleve R., Watrous J., de Wolf R. Quantum Fingerprinting. Physical Review Letters. 87 (16):167902. arXiv:quant-ph/0102001

arXiv
[5]

Convergence of an adaptive hp finite element strategy in one space dimension / W

Dorfler W . Convergence of an adaptive hp finite element strategy in one space dimension / W . Dorfler, V. Heuveline // ScienceDirect, Applied Numerical Mathematics. – 2007. – No.57. – P . 1108–1124

work page 2007
[6]

Comparison of error indicators and refinement criteria for hp -adaptation algorithm for finite element method / R

Drebotiy R. Comparison of error indicators and refinement criteria for hp -adaptation algorithm for finite element method / R. Drebotiy, H. Shynkarenko // Visnyk Odeskoho Natsionalnoho Universytetu. Matematyka i Mekhanika. – Odessa, 2014.– V.19, Is. 4(24).– Pp. 45–57

work page 2014
[7]

Fully parallel algorithm for solution of advection -reaction Cauchy problem in one finite element regularization scheme / R

Drebotiy R. Fully parallel algorithm for solution of advection -reaction Cauchy problem in one finite element regularization scheme / R. Drebotiy, V. Stelmashchuk, H. Shynkarenko // Manufacturing Processes. Actual Problems. – Politechnika Opolska.– Opole, 2023.– Vol. 1.– Pp. 35-44

work page 2023

Show all 23 references

[9]

hp -adaptive finite element method for 1D diffusion -convection-reaction boundary value problems / R

Drebotiy R. hp -adaptive finite element method for 1D diffusion -convection-reaction boundary value problems / R. Drebotiy, H. Shynkarenko // Manufacturing Processes. Actual Problems-2014. Politechnika Opolska. – Opole. – 2014. – Vol. 1. – P . 11–26

2014
[10]

On the application of the one hp-adaptive finite element strategy for nonsymmetric convection-diffusion- reaction problems / R

Drebotiy R. On the application of the one hp-adaptive finite element strategy for nonsymmetric convection-diffusion- reaction problems / R. Drebotiy, H. Shynkarenko // Journal of Numerical and Applied Mathematics, ISSN: 0868 -6912.– Kyiv, 2017.– Issue 3(126). Pp. 48-60

2017
[11]

Regularized finite element method for singular perturbed advection-diffusion-reaction models with nonuniform sources

Drebotiy R., Shynkarenko H. Regularized finite element method for singular perturbed advection-diffusion-reaction models with nonuniform sources. Visnyk of the Lviv University. Series Appl.Math.and Informatics,ISSN: 2078-5097. - Lviv, 2021, Issue 29. - P .27-36

2021
[12]

Recent Developments in Discontinuous Galerkin Finite Element Methods for Partial Differential Equations

Feng X., Karakashian O., Xing Y. Recent Developments in Discontinuous Galerkin Finite Element Methods for Partial Differential Equations. The IMA Volumes in Mathematics and its Applications, 2012, Vol.157

2012
[13]

G. I. Marchuk Mathematical Models in Environmental Problems, North Holland, 2012, ISBN 978-0444558244

2012
[14]

H.-G. Roos, M. Stynes, L. Tobishka Numerical Methods for Singularly Perturbed Differential Equations: Convection - Diffusion and Flow Problems. Springer
[15]

W ., Hassidim A., Lloyd S

Harrow A. W ., Hassidim A., Lloyd S. Quantum algorithm for solving linear systems of equations. Physical Review Letters. 103 (15): 150502. arXiv:0811.3171

arXiv
[16]

Tikhonov type regularization methods: History and recent progress

Lenzen F., Scherzer O. Tikhonov type regularization methods: History and recent progress. Proceeding Eccomas, 2004

2004
[17]

Logan J. D. Transport modeling in hydrogeochemical systems. New York: Springer, 2001. - 226 p

2001
[18]

Quantum algorithms and the finite element method, ISSN:2469 -9934, Physical Review A, Vol

Montanaro A., Pallister S. Quantum algorithms and the finite element method, ISSN:2469 -9934, Physical Review A, Vol. 93, No.3, American Physical Society (APS), 2016

2016
[19]

Quantum Computation and Quantum Information: 10th Anniversary Edition

Nielsen, M., Chuang, I. Quantum Computation and Quantum Information: 10th Anniversary Edition. Cambridge: Cambridge University Press. - 2010. doi:10.1017/CBO9780511976667

2010 doi
[20]

-C., Alexander S., van den Berg E., Chapuran T

Scherer A., Valiron B., Mau S. -C., Alexander S., van den Berg E., Chapuran T. E. Concrete resource analysis of the quantum linear system algorithm used to compute the electromagnetic scattering cross section of a 2D target. Quantum Inf Process (2017) 16: 60. arXiv:1505.06552v2

2017 arXiv
[21]

Korzekwa, Ken Korzekwa Continuous Intestinal Absorption Model Based on the Convection– Diffusion Equation, Mol Pharm

Swati Nagar, Richard C. Korzekwa, Ken Korzekwa Continuous Intestinal Absorption Model Based on the Convection– Diffusion Equation, Mol Pharm. 2017 Sep 5; 14(9): 3069 –3086. https://www.ncbi.nlm.nih.gov/pmc/articles/PMC5627780/

2017
[22]

M., Shynkarenko H

Trushevsky V. M., Shynkarenko H. A., Shcherbyna N. M. Finite element method and artificial neural networks: theoretical aspects and application. Lviv: Ivan Franko National University of Lviv, 2014, ISBN:978 -617-10-0127-5 (in Ukrainian)

2014
[23]

Adaptive finite element methods

Verfürth R. Adaptive finite element methods. Ruhr-Universität Bochum: Lecture notes. – 2011. – 129 p

2011

Pith tools

Reviewed August 12, 2026 · model on record in the stance chip above.

[1] [8]

Heuristic choice of the regularization parameter for optimal stabilization of the finite element approximations / R

Drebotiy R. Heuristic choice of the regularization parameter for optimal stabilization of the finite element approximations / R. Drebotiy, H. Shynkarenko // Mathematical Methods and Physicomechanical Fields. – Pidstryhach Institute for Applied Problems of Mechanics and Mathematics, National Academy of Sciences of Ukraine. – Lviv 2023. – Vol.66, №1-2, pp.206-221

work page 2023

[2] [1]

Numerical Approximation of Partial Differential Equations / S

Bartels S. Numerical Approximation of Partial Differential Equations / S. Bartels. – Springer, 2016. – 541p

work page 2016

[3] [2]

Variational Quantum Linear Solver / LaRose R., Cerezo M., Subasi Y., Cincio L., Coles P

Bravo-Prieto C. Variational Quantum Linear Solver / LaRose R., Cerezo M., Subasi Y., Cincio L., Coles P . // ISSN: 2521- 327X, Quantum 7, 1188 (2023)

work page 2023

[4] [3]

The Mathematical Theory of Finite Element Methods

Brenner S., Scott L. The Mathematical Theory of Finite Element Methods. Springer, 2008, 3ed. - 404 p

work page 2008

[5] [4]

Quantum Fingerprinting

Buhrman H., Cleve R., Watrous J., de Wolf R. Quantum Fingerprinting. Physical Review Letters. 87 (16):167902. arXiv:quant-ph/0102001

arXiv

[6] [5]

Convergence of an adaptive hp finite element strategy in one space dimension / W

Dorfler W . Convergence of an adaptive hp finite element strategy in one space dimension / W . Dorfler, V. Heuveline // ScienceDirect, Applied Numerical Mathematics. – 2007. – No.57. – P . 1108–1124

work page 2007

[7] [6]

Comparison of error indicators and refinement criteria for hp -adaptation algorithm for finite element method / R

Drebotiy R. Comparison of error indicators and refinement criteria for hp -adaptation algorithm for finite element method / R. Drebotiy, H. Shynkarenko // Visnyk Odeskoho Natsionalnoho Universytetu. Matematyka i Mekhanika. – Odessa, 2014.– V.19, Is. 4(24).– Pp. 45–57

work page 2014

[8] [7]

Fully parallel algorithm for solution of advection -reaction Cauchy problem in one finite element regularization scheme / R

Drebotiy R. Fully parallel algorithm for solution of advection -reaction Cauchy problem in one finite element regularization scheme / R. Drebotiy, V. Stelmashchuk, H. Shynkarenko // Manufacturing Processes. Actual Problems. – Politechnika Opolska.– Opole, 2023.– Vol. 1.– Pp. 35-44

work page 2023