Pith. sign in

REVIEW 2 major objections 4 minor 41 references

An Adaptive $s$-step Conjugate Gradient Algorithm with Dynamic Basis Updating

T0 review · 2 major / 4 minor · reviewed 2026-08-14 · deepseek-v4-flash

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

Pith's one-line read An adaptive s-step conjugate gradient algorithm that rebuilds its Newton or Chebyshev bases from incremental Ritz-value estimates meets user-specified accuracy with fewer global synchronizations than the previous adaptive s-step method.

desk verdict A genuinely useful algorithmic improvement with an honest experimental section, but the new automatic safety parameter rests on a bound that goes in the wrong direction. read the letter →

arxiv 1908.04081 v1 pith:J7F5RH2O submitted 2019-08-12 math.NA cs.DCcs.NA

Erin C. Carson This is my paper · ORCID
classification math.NAcs.DCcs.NA MSC 65F1065F5065Y0565Y20
keywords conjugategradients-stepmethodsKrylovsubspacecommunication-avoidingalgorithmsRitzvalueestimationNewtonbasisChebyshevfiniteprecisionaccuracy
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

Communication overloads parallel conjugate gradient iterations, and s-step variants cut synchronization points at the cost of worse finite-precision behavior because the Krylov bases they build can become ill-conditioned. This paper claims that the previous adaptive s-step CG algorithm can be improved by computing cheap incremental estimates of the largest and smallest Ritz values during the iteration and using them to refresh the Newton or Chebyshev polynomial basis parameters after each outer loop. The same estimates replace a previously heuristic constant that controls when the inner loop must stop, turning a user-supplied parameter into a computable quantity. In experiments on small symmetric positive definite test problems, the improved algorithm reached the requested accuracy in every configuration tested, while fixed s-step CG and the old adaptive version often stagnated or diverged for larger step sizes, and it did so with fewer outer-loop synchronization points in many cases.

What carries the argument

The object that carries the argument is the s-step Krylov basis matrix $\widehat{Y}_{k,s}$ and its Gram matrix $G_{k,s}=\widehat{Y}_{k,s}^T\widehat{Y}_{k,s}$, whose condition number governs the finite-precision residual gap through the bound (2.10). The mechanism that keeps that condition number under control is the incremental estimation of the extreme Ritz values from the bidiagonal factor $L_i$ of the Lanczos matrix, combined with the scalar recurrence for $\psi_{m+j+1}$ that produces the adaptive constant $c_{m+j+1}$ from (4.6). These ingredients set the recurrence coefficients $\theta$, $\gamma$, $\mu$ for the Newton or Chebyshev bases and determine when the inner loop should stop, using the largest residual norm seen so far in the current outer loop.

What would settle it

On a small symmetric positive definite matrix with an isolated smallest eigenvalue, record $\widetilde{\lambda}_{\min}$ from Algorithm 4 at every inner iteration before convergence and compare it with the true $\lambda_{\min}(A)$ from a high-precision eigensolver. If $\widetilde{\lambda}_{\min} > \lambda_{\min}(A)$ at any such iteration, the premise behind the bound (4.6) is violated; one can then check whether the residual gap exceeds the bound (2.10) or the algorithm fails to reach the requested tolerance while standard CG succeeds.

Watch

Extended reading notes

Core claim

The paper's central claim is that incremental estimates $\widetilde{\lambda}_{\min}$ and $\widetilde{\lambda}_{\max}$, obtained from the bidiagonal Cholesky factor of the Lanczos tridiagonal matrix as the CG recurrences run, are sufficient to keep the s-step bases well conditioned and to set the inner-loop stopping rule automatically. Concretely, the algorithm builds Newton parameters by (2.7) or simplified Chebyshev parameters by (2.8), and sets $c_{m+j+1}=\max\{1, \widetilde{\lambda}_{\max}(\psi_{m+j+1}/\widetilde{\lambda}_{\min})^{1/2}\}$ with $\psi_{m+j+1}=\psi_{m+j}/(\psi_{m+j}+\beta_{m+j})$, which approximates the exact ratio $\xi_{m+j+1}=\|A\|\,\|x-\widehat{x}_{m+j+1}\|/\|\widehat{r}_{m+j+1}\|$. With this choice the paper argues that the basis condition-number bound can be enforced without user-supplied spectral information or a heuristic constant, and the numerical experiments support the claim that the resulting algorithm attains the prescribed accuracy with fewer global synchronizations than the earlier approach.

Load-bearing premise

The load-bearing premise is that the incremental Ritz-value estimates stay accurate enough in finite-precision s-step arithmetic that $\widetilde{\lambda}_{\min}$ never overestimates the true smallest eigenvalue, because equation (4.6) treats $\widetilde{\lambda}_{\min}$ as a lower bound when setting the stopping constant.

Editorial extensions

If this is right

  • For the tested matrices, the improved algorithm converged to the prescribed accuracy in every case, while fixed s-step CG and the old adaptive algorithm frequently stagnated or diverged at $\sigma=10$ or $\sigma=15$.
  • The number of outer-loop iterations, a proxy for global synchronizations, decreased as the maximum step size $\sigma$ increased, and in several tests dropped by more than a factor of 12 relative to HSCG.
  • The total number of inner iterations needed by the improved algorithm stayed roughly constant as $\sigma$ grew, so choosing $\sigma$ too large no longer inflates the iteration count.
  • The heuristic choice of $c_k$ in the previous algorithm is replaced by an automatic estimate, removing one user-supplied parameter and the need for a priori spectral information.
  • Newton and Chebyshev bases both produced improved convergence behavior, but the paper does not claim either basis is uniformly better; the choice remains problem-dependent.

Reading between the lines

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

  • If the finite-precision reliability of the incremental Ritz estimates were proven inside the s-step inner loop, the same dynamic basis restart idea could be transferred to s-step GMRES, s-step BiCGSTAB, or s-step Lanczos eigensolvers, where basis conditioning plays the same role.
  • The adaptive formula for $c_{m+j+1}$ is essentially an estimate of the error-to-residual ratio; a direct test of its robustness would compare $\widetilde{\lambda}_{\min}$ from Algorithm 4 against the true smallest eigenvalue early in the iteration, because an overestimate would weaken the bound (4.6).
  • The observed near-independence of total iterations from $\sigma$ suggests that an autotuner could safely choose a large maximum step size and let the adaptive loop trim $s$, simplifying parameter selection.
  • A high-performance parallel implementation on a realistic machine is the natural next test: the paper's experiments are small and report synchronization counts as a proxy rather than wall-clock time.
Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

2 major / 4 minor

Summary. The paper proposes an improved adaptive s-step conjugate gradient algorithm for SPD linear systems. It uses incremental estimates of the largest and smallest Ritz values, following Meurant and Tichý, to dynamically update Newton or Chebyshev basis parameters and to replace the previously user-supplied constant c_k by an automatically updated quantity c_{m+j+1} based on Eq. (4.6). The authors present numerical experiments on several small SuiteSparse matrices showing that the improved algorithm often converges to the requested accuracy with fewer outer iterations (global synchronizations) than fixed s-step CG and the previous adaptive s-step CG, while also honestly reporting a difficult case (nos1).

Significance. If the claims hold, the contribution is practically relevant: it removes one heuristic parameter from adaptive s-step CG and improves basis conditioning without user-supplied spectral information. The paper's strengths include a clear experimental setup, comparison against both classical HSCG and the prior adaptive algorithm, a range of tolerances and matrix sizes, and an honest discussion of cases where the improved method is not beneficial. The main weakness is that the derivation of the safeguard c_{m+j+1} in Section 4 contains a bound-direction error, so the central reliability claim is not theoretically substantiated; the experiments are encouraging but cannot by themselves close this gap.

major comments (2)
  1. [Section 4, Eq. (4.6) and Algorithm 5, lines 20–25] The substitution mu = tilde_lambda_min is invalid for the stated purpose. The bound quoted from [33] requires 0 < mu <= lambda_min(A), but in exact arithmetic the smallest Ritz value lambda_min(T_i) of the Lanczos tridiagonal matrix satisfies lambda_min(T_i) >= lambda_min(A) by the Rayleigh–Ritz (interlacing) property. Hence tilde_lambda_min in (4.2) is an upper bound, not a lower bound, for lambda_min(A). Consequently the quantity tilde_xi in (4.6) can be smaller than the true ratio xi in (4.4), and the break test (4.7) can allow an s_k that violates the sufficient condition (3.1). The claim that the improved algorithm attains the user-specified accuracy is therefore not substantiated. The experiments do not test this safeguard directly, because Figure 2 plots tilde_xi against the global bounds 1 and kappa(A), not against the true xi.
  2. [Section 4 (Algorithms 3–4) and Section 5] The incremental Ritz estimates are computed from the alpha and beta coefficients generated inside the finite-precision s-step inner loop (Algorithm 5, lines 13–17). No finite-precision analysis is provided for these estimates in the s-step setting, where the computed coefficients and coordinate recurrences can be amplified by kappa(Y_hat_{k,s}). Since both the basis parameter update (line 29) and the safeguard c_{m+j+1} (line 23) depend on these estimates, the paper does not establish that the dynamic bases are better conditioned than monomial bases in finite precision, nor that (4.7) enforces (3.1). I would ask for an explicit perturbation analysis, even a preliminary one, or for a clear statement that the dynamic basis update is heuristic and outside the scope of the theoretical guarantee.
minor comments (4)
  1. [Section 5, Tables 2–7] The value of c_k used for the old adaptive s-step CG baseline is not stated; since this heuristic strongly affects the number of inner iterations, please specify it for reproducibility.
  2. [Algorithm 5, line 1] The initial value c_1 = epsilon^{-1/2} is not derived or explained; please justify this choice or label it as a default heuristic.
  3. [Section 4, before Eq. (4.6)] The phrase 'approximately bound' is ambiguous: a bound is either valid or not. Given the issue raised above, please rephrase this to say explicitly that the quantity is heuristic, or provide a certified upper bound.
  4. [Section 5.3] The statement that the total number of iterations is 'independent of sigma' is not strictly true for all reported cases; for example, in Table 7 (nos1, epsilon* = 1e-6, Newton basis) the totals are 1217, 1290, and 1086 for sigma = 5, 10, 15. Please qualify this claim.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: all claimed improvements are validated by external convergence benchmarks, and the Ritz-based parameter updates are algorithmic bootstrapping rather than fitted predictions.

full rationale

The paper's central claims—better numerical behavior and fewer synchronizations—are supported by experiments measuring true relative residuals against HSCG, fixed s-step CG, and the author's earlier adaptive algorithm. Nothing in the derivation is equivalent to its own inputs by construction. The new ingredients are (i) Newton/Chebyshev basis parameters updated from incremental extreme Ritz estimates via Meurant–Tichý, and (ii) automatic setting of c_{m+j+1} in (4.6) from λ̃max, λ̃min, and ψ. These quantities are produced by the CG/Lanczos recurrences, not by fitting a parameter to the target accuracy, so using them inside the algorithm is legitimate bootstrapping. Condition (3.1) is inherited from the author's published rounding-error analysis; citing one's prior work is normal cumulative research and is not load-bearing in the sense of being unverified or identical to the result. The skeptical issue that λmin(T_i) is an upper bound on λmin while (4.6) requires 0 < μ ≤ λmin is a genuine correctness/bound-direction concern, but it is not circularity: no equation reduces to a pre-supplied answer. The paper also states honest limitations (no demonstrated performance benefit over HSCG yet; unexplained behavior on nos1), consistent with a non-circular empirical study.

Assumptions & free parameters 3 free parameters · 5 assumptions · 0 invented entities

No new physical or mathematical entities are introduced. The algorithm relies on standard numerical linear algebra assumptions, the author's prior finite-precision bound, and the Meurant-Tichý Ritz estimation technique, all of which are treated as inputs from the literature. The free parameters are user-facing tuning knobs rather than fitted constants, with the exception of the heuristic initial value c_1.

free parameters (3)
  • initial c_1 = ε^{-1/2}
    Algorithm 5 line 1 sets c_1 = ε^{-1/2} without derivation; it affects the break criterion in the first inner loop before Ritz estimates exist.
  • maximum allowed block size σ = 5, 10, 15 in experiments
    User input; the paper notes in Section 6 that σ must be tuned to maximize performance.
  • maximum basis growth factor f = f = σ in experiments
    User input controlling how quickly the block size can grow; set to the maximum in all experiments, but still a tuning parameter.
assumptions (5)
  • domain assumption A is symmetric positive definite.
    The algorithm and all cited analysis are for SPD matrices; used in Section 2 and throughout.
  • domain assumption The residual gap bound (2.10) from [9] accurately models the amplification of local roundoff by κ(Y_hat_k,s).
    This bound motivates the adaptive step-size criterion (3.1) and the break condition (4.7); the paper does not re-derive it.
  • domain assumption The incremental norm estimates of Meurant and Tichý (Algorithms 3 and 4) give approximations of λ_max and λ_min that remain useful when computed inside finite-precision s-step CG iterations.
    Algorithm 5 uses these estimates to set basis parameters and c_{m+j+1}; their finite-precision behavior in the s-step setting is not analyzed.
  • domain assumption The quantity ~ξ in (4.6) is a valid replacement for the ratio ξ = ||A|| ||x̂ - x|| / ||r̂||, i.e., μ = ~λ_min is a lower bound for λ_min.
    The paper calls it an approximate bound; if ~λ_min overestimates λ_min, the computed c can be too small and allow ill-conditioned bases.
  • domain assumption Number of outer loop iterations is a meaningful proxy for the number of global synchronizations.
    Used throughout Section 5 to compare algorithms; reasonable but not a measurement of wall-clock time.

how reviews work

0 comments
Cite this review

Pith. "Pith review of An Adaptive $s$-step Conjugate Gradient Algorithm with Dynamic Basis Updating." pith.science (2026). https://pith.science/paper/J7F5RH2O

@misc{pith2026190804081,
  author       = {Pith},
  title        = {Pith review of: An Adaptive $s$-step Conjugate Gradient Algorithm with Dynamic Basis Updating},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/J7F5RH2O}},
  note         = {Machine review of arXiv:1908.04081}
}
abstract

The adaptive $s$-step CG algorithm is a solver for sparse, symmetric positive definite linear systems designed to reduce the synchronization cost per iteration while still achieving a user-specified accuracy requirement. In this work, we improve the adaptive $s$-step conjugate gradient algorithm by use of iteratively updated estimates of the largest and smallest Ritz values, which give approximations of the largest and smallest eigenvalues of $A$, using a technique due to Meurant and Tich{\' y} [G. Meurant and P. Tich{\' y}, Numer. Algs. (2018), pp.~1--32]. The Ritz value estimates are used to dynamically update parameters for constructing Newton or Chebyshev polynomials so that the conditioning of the $s$-step bases can be continuously improved throughout the iterations. These estimates are also used to automatically set a variable related to the ratio of the sizes of the error and residual, which was previously treated as an input parameter. We show through numerical experiments that in many cases the new algorithm improves upon the previous adaptive $s$-step approach both in terms of numerical behavior and reduction in number of synchronizations.

Figures

Figures reproduced from arXiv: 1908.04081 by the authors.

Figure 1
Figure 1. Test results for matrix nos1 with s, σ = 10 and ε ∗ = 10−6 . Plots on the left show the convergence trajectories for HSCG, fixed s-step CG, and improved adaptive s-step CG with Newton and Chebyshev bases. Plots on the right show the values of sk in the improved adaptive algorithm using the Newton basis (magenta) and the Chebyshev basis (green). The top row uses cm+j+1 = λ˜max/λ˜min, the middle row uses cm+j+1 = 1, a… view at source ↗
Figure 2
Figure 2. The value of cm+j+1 = ξ˜m+j+1 for improved adaptive s-step CG using the Newton basis (magenta) and the Chebyshev basis (green). The dashed horizonal lines show the bounds for ξm+j+1, 1 and κ(A). nicely highlights the benefits of the improved approach. When ε ∗ = 2.2 · 10−10, the improved algorithm outperforms the fixed s-step algorithm and the old adaptive s￾step algorithm in all cases, even for small σ. When ε ∗ = … view at source ↗
Figure 3
Figure 3. Convergence of the relative true residual 2-norm for the matrix 494bus. fixed s-step CG old adptv s-step CG impr. adptv s-step CG w/Newton basis impr. adptv s-step CG w/Chebyshev basis HSCG ε ∗ = 2.2e-10 s, σ = 5 131 (652) 131 (652) 86 (415) 86 (414) s, σ = 10 - [2e-08] 109 (879) 58 (416) 53 (414) 413 s, σ = 15 - - 57 (416) 51 (414) ε ∗ = 1e-6 s, σ = 5 82 (410) 82 (410) 84 (408) 84 (408) s, σ = 10 115 (1147) 115 (11… view at source ↗
Figures from the paper (1 more)
Figure 4
Figure 4. Figure 4: Convergence of the relative true residual 2-norm for the matrix bcsstk09. fixed s-step CG old adptv s-step CG impr. adptv s-step CG w/Newton basis impr. adptv s-step CG w/Chebyshev basis HSCG ε ∗ = 1.9e-12 s, σ = 5 51 (253) 48 (235) 59 (219) 54 (205) s, σ = 10 - [7e-09…

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

41 extracted references · 41 canonical work pages

  1. [33]

    Meurant and P

    G. Meurant and P. Tich´y, Approximating the extreme Ritz values and upper bounds for the A-norm of the error in CG , Numer. Algs., (2018), pp. 1–32

    work page 2018

  2. [1]

    Z. Bai, D. Hu, and L. Reichel , A Newton basis GMRES implementation , IMA J. Numer. Anal., 14 (1994), pp. 563–581

    work page 1994

  3. [2]

    Bouras and V

    A. Bouras and V. Frayss ´e, A relaxation strategy for inexact matrix-vector products for Krylov methods , Tech. Rep. CERFACS TR/PA/00/15, European Centre for Research and Advanced Training in Scientific Computation, 2000

    work page 2000

  4. [3]

    Matrix Anal

    , Inexact matrix-vector products in Krylov methods for solving linear sys- tems: a relaxation strategy, SIAM J. Matrix Anal. Appl., 26 (2005), pp. 660–678

    work page 2005

  5. [4]

    Calvetti, G

    D. Calvetti, G. H. Golub, and L. Reichel, An adaptive Chebyshev iterative method for nonsymmetric linear systems based on modified moments, Numerische Mathematik, 67 (1994), pp. 21–40

    work page 1994

  6. [5]

    Calvetti and L

    D. Calvetti and L. Reichel , On the evaluation of polynomial coefficients , Numer. Algs., 33 (2003), pp. 153–161

    work page 2003

  7. [6]

    Carson and J

    E. Carson and J. Demmel , A residual replacement strategy for improving the maximum attainable accuracy of s-step Krylov subspace methods, SIAM J. Matrix Anal. Appl., 35 (2014), pp. 22–43

    work page 2014

  8. [7]

    Carson and J

    E. Carson and J. W. Demmel , Accuracy of the s-step Lanczos method for the symmetric eigenproblem in finite precision , SIAM J. Matrix Anal. Appl., 36 (2015), pp. 793–819

    work page 2015

Show all 41 references
  1. [8]

    E. C. Carson, Communication-avoiding Krylov subspace methods in theory and practice, PhD thesis, U.C. Berkeley, 2015

    2015

  2. [9]

    E. C. Carson, The adaptive s-step conjugate gradient method, SIAM J. Matrix Anal. Appl., 39 (2018), pp. 1318–1338

    2018

  3. [10]

    Chronopoulos and C

    A. Chronopoulos and C. Gear,s-step iterative methods for symmetric linear systems, J. Comput. Appl. Math., 25 (1989), pp. 153–168

    1989

  4. [11]

    T. A. Davis and Y. Hu , The University of Florida sparse matrix collection , ACM Transactions on Mathematical Software (TOMS), 38 (2011), pp. 1:1–1:25

    2011

  5. [12]

    de Sturler, A parallel variant of GMRES(m) , in Proc

    E. de Sturler, A parallel variant of GMRES(m) , in Proc. 13th IMACS World Congress Comput. Appl. Math., vol. 9, IMACS, Criterion Press, 1991

    1991

  6. [13]

    De Sturler and H

    E. De Sturler and H. A. van der Vorst, Reducing the effect of global com- munication in GMRES(m) and CG on parallel distributed memory computers , Appl. Numer. Math., 18 (1995), pp. 441–459

    1995

  7. [14]

    Demmel, M

    J. Demmel, M. Hoemmen, M. Mohiyuddin, and K. Yelick, Avoiding com- munication in sparse matrix computations , in IEEE Int. Symp. Parallel Dist. Proc., IEEE, 2008, pp. 1–12. 24

    2008

  8. [15]

    Dongarra et al., The international exascale software project roadmap, Int

    J. Dongarra et al., The international exascale software project roadmap, Int. J. High Perf. Comput. Appl., 25 (2011), pp. 3–60

    2011

  9. [16]

    Dongarra, M

    J. Dongarra, M. A. Heroux, and P. Luszczek , High-performance conjugate-gradient benchmark: A new metric for ranking high-performance com- puting systems, Int. J. High Perf. Comput. Appl., 30 (2016), pp. 3–10

    2016

  10. [17]

    Erhel , A parallel GMRES version for general sparse matrices , Electronic Trans

    J. Erhel , A parallel GMRES version for general sparse matrices , Electronic Trans. Numer. Anal., 3 (1995), pp. 160–176

    1995

  11. [18]

    Gautschi, The condition of polynomials in power form , Math

    W. Gautschi, The condition of polynomials in power form , Math. Comput., 33 (1979), pp. 343–352

    1979

  12. [19]

    Ghysels, T

    P. Ghysels, T. J. Ashby, K. Meerbergen, and W. Vanroose , Hiding global communication latency in the GMRES algorithm on massively parallel ma- chines, SIAM J. Sci. Comput., 35 (2013), pp. C48–C71

    2013

  13. [20]

    Ghysels and W

    P. Ghysels and W. Vanroose , Hiding global synchronization latency in the preconditioned conjugate gradient algorithm , Parallel Comput., 40 (2014), pp. 224–238

    2014

  14. [21]

    Greenbaum, Behavior of slightly perturbed Lanczos and conjugate-gradient recurrences, Lin

    A. Greenbaum, Behavior of slightly perturbed Lanczos and conjugate-gradient recurrences, Lin. Alg. Appl., 113 (1989), pp. 7–63

    1989

  15. [22]

    Matrix Anal

    , Estimating the attainable accuracy of recursively computed residual meth- ods, SIAM J. Matrix Anal. Appl., 18 (1997), pp. 535–551

    1997

  16. [23]

    Gutknecht and Z

    M. Gutknecht and Z. Strako ˇs, Accuracy of two three-term and three two- term recurrences for Krylov space solvers, SIAM J. Matrix Anal. Appl., 22 (2000), pp. 213–229

    2000

  17. [24]

    Heroux, R

    M. Heroux, R. Bartlett, V. H. R. Hoekstra, J. Hu, T. Kolda, R. Lehoucq, K. Long, R. Pawlowski, E. Phipps, A. Salinger, H. Thorn- quist, R. Tuminaro, J. Willenbring, and A. Williams , An Overview of Trilinos, Tech. Rep. SAND2003-2927, Sandia National Laboratories, 2003

    2003

  18. [25]

    M. R. Hestenes and E. Stiefel , Methods of conjugate gradients for solving linear systems, J. Research Nat. Bur. Standards, 49 (1952), pp. 409–436

    1952

  19. [26]

    Hindmarsh and H

    A. Hindmarsh and H. Walker, Note on a Householder implementation of the GMRES method, Tech. Rep. UCID-20899, Lawrence Livermore National Lab., CA., 1986

    1986

  20. [27]

    Hoemmen, Communication-avoiding Krylov subspace methods , PhD thesis, U.C

    M. Hoemmen, Communication-avoiding Krylov subspace methods , PhD thesis, U.C. Berkeley, 2010

    2010

  21. [28]

    Imberti and J

    D. Imberti and J. Erhel , Varying the s in your s-step GMRES , Electronic Trans. Numer. Anal., 47 (2017), pp. 206–230

    2017

  22. [29]

    W. D. Joubert and G. F. Carey , Parallelizable restarted iterative methods for nonsymmetric linear systems. Part I: Theory , Int. J. Comput. Math., 44 (1992), pp. 243–267

    1992

  23. [30]

    Liesen and Z

    J. Liesen and Z. Strakoˇs, Krylov Subspace Methods: Principles and Analysis , Oxford University Press, 2013. 25

    2013

  24. [31]

    T. A. Manteuffel, Adaptive procedure for estimating parameters for the non- symmetric Tchebychev iteration , Numerische Mathematik, 31 (1978), pp. 183– 208

    1978

  25. [32]

    Meurant and Z

    G. Meurant and Z. Strakoˇs, The Lanczos and conjugate gradient algorithms in finite precision arithmetic , Acta Numerica, 15 (2006), pp. 471–542

    2006

  26. [34]

    Philippe and L

    B. Philippe and L. Reichel , On the generation of Krylov subspace bases , Appl. Numer. Math., 62 (2012), pp. 1171–1186

    2012

  27. [35]

    Reichel, Newton interpolation at Leja points , BIT, 30 (1990), pp

    L. Reichel, Newton interpolation at Leja points , BIT, 30 (1990), pp. 332–346

    1990

  28. [36]

    Saad, Iterative methods for sparse linear systems, SIAM, Philadelphia, Penn- sylvania, 2 ed., 2003

    Y. Saad, Iterative methods for sparse linear systems, SIAM, Philadelphia, Penn- sylvania, 2 ed., 2003

    2003

  29. [37]

    Simoncini and D

    V. Simoncini and D. Szyld , Theory of inexact Krylov subspace methods and applications to scientific computing , SIAM J. Sci. Comput., 25 (2003), pp. 454– 477

    2003

  30. [38]

    G. L. G. Sleijpen and H. A. van der Vorst , Reliable updated residuals in hybrid Bi-CG methods , Computing, 56 (1996), pp. 141–163

    1996

  31. [39]

    H. A. van der Vorst and Q. Ye , Residual replacement strategies for Krylov subspace iterative methods for the convergence of true residuals , SIAM J. Sci. Comput., 22 (1999), pp. 835–852

    1999

  32. [40]

    Van Rosendale, Minimizing inner product data dependencies in conjugate gradient iteration, in Proc

    J. Van Rosendale, Minimizing inner product data dependencies in conjugate gradient iteration, in Proc. IEEE Int. Conf. Parallel Proc., IEEE, 1983, pp. 44–46

    1983

  33. [41]

    Williams, M

    S. Williams, M. Lijewski, A. Almgren, B. Van Straalen, E. Carson, N. Knight, and J. Demmel, s-step Krylov subspace methods as bottom solvers for geometric multigrid , in Proc. IEEE Int. Symp. Parallel Dist. Proc. (IPDPS), IEEE, 2014, pp. 1149–1158

    2014

Pith tools

Reviewed August 14, 2026 · model on record in the stance chip above.