Weak imposition of Signorini boundary conditions on the boundary element method

Erik Burman; Matthew W. Scroggs; Stefan Frei

REVIEW 2 major objections 4 minor 23 references

Weak imposition of Signorini boundary conditions on the boundary element method

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 Signorini contact conditions can be imposed weakly in boundary element methods with no loss of convergence order.

desk verdict A genuinely new Nitsche-type BEM for Signorini contact, with sound analysis and honest numerics, but Lemma 5.3 has a repairable algebraic slip that must be corrected before the main theorems rest on it. read the letter →

arxiv 1908.05580 v2 pith:AYI5OEWZ submitted 2019-08-15 math.NA cs.NA

Erik Burman , Stefan Frei , Matthew W. Scroggs This is my paper · ORCID

classification math.NAcs.NA MSC 65N3865R2074M15

keywords NitschemethodboundaryelementSignoriniproblemCalderónprojectorweakimpositionvariationalinequalityaugmentedLagrangianapriorierrorestimate

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

Signorini contact conditions are inequality constraints on the boundary: the normal flux cannot exceed a limit, the gap cannot become negative, and the two can only be nonzero in a complementary way. This paper extends Nitsche-type weak imposition of boundary conditions, previously developed for boundary element methods on Dirichlet and Neumann data, to such unilateral constraints. It derives a boundary-only formulation based on the Calderón projector and an augmented-Lagrangian contact term, and proves that the nonlinear discrete problem has a unique solution and converges at optimal order. The practical payoff is a contact solver that discretises only the boundary, adds no Lagrange multipliers, and needs no smoothing of the inequality.

What carries the argument

The load-bearing object is the contact boundary operator $B_C$ of (4.7), built from the augmented-Lagrangian projection $P_\tau(u_h,\lambda_h)=\tau(u_h-g_C)-(\lambda_h-\psi_C)$ and its positive part $[\,\cdot\,]_+$. This term converts the inequality constraints into an equality-like weak form that is consistent with the exact solution. The proof of uniqueness and error estimates uses the distance functional $d_C$, which measures the error through a combination of the $V$-norm and a scaled $L^2$-norm of $\mu_h+[P_\tau(v_h,\mu_h)]_+$; the crux is showing that $d_C$ is controlled by the augmented form, so the nonlinear contact residual adds coercivity rather than destroying it. Choosing $\tau\asymp h^{-1}$ and $\beta_D$ bounded below independently of $h$ balances the competing trace norms.

What would settle it

Take the unit-cube test problem from Section 6, fix $\tau=0.5/h$, and solve (4.11) on successively refined meshes with two different initial guesses in Algorithm 6.1; the claim fails if two initial guesses converge to different solutions, or if the measured $V$-norm error systematically decays slower than $h^{\zeta-1/2}+h^{\xi+1/2}$ for smooth data.

Watch

Extended reading notes

Core claim

The central claim is that the mixed Dirichlet–Signorini Laplace problem can be solved by a boundary element method in which the contact conditions enter as a nonlinear term $\langle [P_\tau(u_h,\lambda_h)]_+, v_h+\tau^{-1}\mu_h\rangle_{\Gamma_C}$ in the variational form (4.11). With $\tau\asymp h^{-1}$ and a stable Dirichlet penalty $\beta_D$, the method is consistent, well-posed, and satisfies Theorem 5.6: in the trace norm $\|(\cdot,\cdot)\|_V$ the error is bounded by $h^{\zeta-1/2}|u|_{H^\zeta(\Gamma)}+h^{\xi+1/2}|\lambda|_{H^\xi(\tilde\Gamma)}$, where $\zeta$ and $\xi$ encode the polynomial orders and solution regularity. The same rate transfers to the $H^1(\Omega)$ error of the potential reconstructed from the boundary data via the representation formula. Thus the method reaches the same asymptotic accuracy as a Nitsche-type BEM for linear boundary conditions, while enforcing the Signorini complementarity condition weakly.

Load-bearing premise

Everything rests on the discrete coercivity of the nonlinear contact form: for every discrete trace pair the augmented form plus data term must remain positive and dominate the contact distance measure uniformly in $h$, and if that bound fails the well-posedness and error theorems collapse.

Editorial extensions

If this is right

The method works with standard discontinuous, facewise continuous, and dual-grid flux spaces, each with explicit convergence orders, so existing BEM codes need only add one nonlinear boundary term.
Because the error estimate transfers to $H^1(\Omega)$ through the representation formula, solving only on the boundary does not degrade volume accuracy.
The fixed-point iteration of Algorithm 6.1, with the nonlinear term treated explicitly, converges in the experiments with iteration counts that grow slowly as $h$ shrinks.
The same Calderón-plus-Nitsche template is stated to extend to other inequality boundary conditions, including frictionless contact in linear elasticity.
For spaces that are not an inf-sup stable pair, the predicted order is still observed numerically but conditioning degrades, which motivates the dual-grid pairings as the robust choice.

Reading between the lines

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

The distance functional $d_C$ used in the analysis could be read as a residual-based a posteriori error indicator for adaptive refinement, since it simultaneously measures the boundary residual and the contact complementarity violation.
If the monotonicity of the positive-part operator survives time discretisation, the same weak contact operator should extend to dynamic and frictional contact problems, a case the paper leaves open.
Theorem 5.6 exposes the flux approximation on the contact face as the limiting term; this suggests that mesh grading or enriching the flux space near contact-boundary edges would recover full order when $\lambda$ is only piecewise smooth.
The scaling $\tau\asymp h^{-1}$ makes the contact term act like a boundary stabiliser, so the same operator might regularise ill-conditioned BEM systems for other variational inequalities.

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.

Desk Editor's Note

read the letter

Worth a look. This is a clean extension of the authors' weak-boundary-condition BEM framework to Signorini contact. The formulation uses the Calderón projector with Nitsche-type penalty terms, and the analysis gives existence and uniqueness via Brouwer plus optimal a priori error estimates. The numerical section is honest: it shows the expected order for the inf-sup stable pair, reports the order loss for the unstable pair, and blames conditioning rather than hiding it. That is the right way to report numerics.

The novel piece is the nonlinear contact term inside a boundary integral formulation. It is not a paradigm shift—the tools are adapted from Chouly–Hild–Renard and from the authors' earlier paper—but the combination is not in the literature and the derivation is careful. The citation pattern is standard; leaning on [2] for coercivity of A+B_D is self-citation of a published, reviewed result, not a problem.

The soft spot is Lemma 5.3. As printed, the first equality drops the 2τ^{-1}⟨µ_h,[Pτ]_+⟩ term, and the next line writes u_h where it should be v_h. Lemma 5.3 is load-bearing: Theorems 5.4 and 5.6 both need it. However, this is not a hidden flaw in the method. Keep the cross term and use the fact that [Pτ]_+ and [Pτ]_− have disjoint support, so ⟨[Pτ]_+,Pτ⟩ = ||[Pτ]_+||². The intended identity B_C[(v_h,µ_h),(v_h,µ_h)] − ⟨[Pτ]_+, g_C − τ^{-1}ψ_C⟩ = τ^{-1}||µ_h + [Pτ]_+||² then follows. So the lemma is true; the proof in the manuscript needs a corrected chain, not a new idea.

One smaller caveat: the fixed-point iteration is shown to work numerically, but there is no convergence proof for Algorithm 6.1. That is clearly not the paper's main claim, so it is a minor gap, not a serious one.

My verdict: the mathematical core is sound, the presentation is competent, and the advertised results are credible once Lemma 5.3 is corrected. This deserves a real referee report, with the request to fix the lemma and check the subsequent signs. I would use the method and would cite the paper after revision.

Referee Report

2 major / 4 minor

Summary. The paper derives and analyzes a boundary element method for the Signorini problem for the Laplacian in three dimensions. Starting from the Calderón projector and the authors' earlier framework for weakly imposed Dirichlet conditions in [2], the method adds an augmented-Lagrangian-type boundary operator B_C on the contact boundary, formulated through the positive part [Pτ(uh,λh)]_+. The resulting nonlinear discrete problem is (4.11). The main theoretical results are the discrete coercivity Lemma 5.3, existence and uniqueness in Theorem 5.4, the best-approximation Lemma 5.5, and the a priori error estimates in Theorem 5.6, giving ||(u−uh, λ−λh)||_V ≲ h^{ζ−1/2}|u|_{H^ζ(Γ)} + h^{ξ+1/2}|λ|_{H^ξ(Γ_tilde)} and the corresponding H^1(Ω) estimate for the reconstructed potential. Numerical experiments on the unit cube use P1×DUAL0 and P1×DP0 discretizations and report convergence rates consistent with the theory for the first pairing, with a conditioning-related order loss for the second.

Significance. If the results hold, this is a useful contribution: it appears to be the first Nitsche-type boundary element method for Signorini contact conditions, extending the unified Calderón framework from [2] to unilateral constraints. The analysis is nontrivial and uses a quasi-distance d_C, a discrete coercivity lemma, and the Lipschitz properties of the positive part, all in a nonstandard boundary integral setting. The claimed convergence rates are optimal in h and the reconstructed H^1(Ω) estimate is a useful consequence of the Calderón representation. The numerical section also gives an honest discussion of the conditioning difficulties for P1×DP0, which is informative for users. The main weakness is that two displayed algebraic steps in the analysis, one in Lemma 5.3 and one in the continuity proof inside Theorem 5.4, are incorrect as printed; both are localized and appear repairable without changing the main results.

major comments (2)

[Section 5, proof of Lemma 5.3] The displayed algebraic chain in the proof is not an identity. The first equality drops the cross term 2τ^{-1}⟨µ_h,[Pτ(v_h,µ_h)]_+⟩, and the second line uses u_h where the statement requires v_h. Since Lemma 5.3 is invoked in Theorem 5.4 and, through Lemma 5.5, in Theorem 5.6, the central existence and error theorems are not proven by the printed proof. The gap is repairable: carrying the cross term through gives B_C[(v_h,µ_h),(v_h,µ_h)] − ⟨[Pτ(v_h,µ_h)]_+, g_C − τ^{-1}ψ_C⟩ = τ^{-1}‖µ_h + [Pτ(v_h,µ_h)]_+‖²_{Γ_C}, which together with Lemma 3.2 yields the claimed bound. The proof must be rewritten so that the printed equalities are correct.
[Section 5, proof of Theorem 5.4, continuity of F] The displayed identity for ⟨F(v1_h,µ1_h)−F(v2_h,µ2_h),(w_h,η_h)⟩ is algebraically incorrect. The difference of the two B_C terms is 1/2⟨µ1_h−µ2_h,w_h⟩ + τ^{-1}⟨µ1_h−µ2_h,η_h⟩ − 1/2⟨v1_h−v2_h,η_h⟩ + ⟨[Pτ(v1_h,µ1_h)]_+−[Pτ(v2_h,µ2_h)]_+, w_h+τ^{-1}η_h⟩, not the expression printed with 1/2⟨µ1_h−µ2_h,w_h+τ^{-1}η_h⟩ − 1/2⟨v1_h−v2_h,µ1_h−µ2_h⟩. The subsequent inequality is therefore not justified by the preceding line. The continuity of F can nonetheless be recovered from the Lipschitz property (5.4), the continuity of A+B_D, and finite-dimensional norm equivalence, so this is a repairable proof defect rather than a false claim, but the proof as printed needs correction.

minor comments (4)

[Section 5, Lemma 5.3 proof] In addition to the missing cross term, the proof writes u_h in the second displayed line where the variable of the lemma is v_h; this notation should be unified throughout the proof.
[Section 6, Figure 3 and surrounding text] For V_h = P1_h(Γ)×DP0_h(Γ), the measured convergence order is below the predicted 1.5, and the authors attribute this to ill-conditioning and the ineffective mass-matrix preconditioner. Since the theorem concerns the exact discrete solution and the experiments use GMRES with a fixed tolerance, this does not by itself refute Theorem 5.6, but the numerical support for that pairing would be stronger if the solver residual or an alternative stable pairing were reported.
[General] There are a few typographical issues, including 'analagous' in the sentence preceding Lemma 5.3, 'equivalance' in the proof of Theorem 5.4, and 'averge' in the caption of Figure 3; these should be corrected.
[Section 4, equations (4.1)-(4.2)] The equivalence of the augmented-Lagrangian identities (4.1) and (4.2) with the Signorini conditions is stated briefly with a reference to [7]; a one-sentence verification would improve readability, especially because the signs of the positive and negative parts are central to the method.

Circularity Check

0 steps flagged · score 0.0 of 10

No circularity: the Signorini BEM derivation is new content built on external augmented-Lagrangian identities; prior self-citations provide independent linear coercivity, not the target result.

full rationale

The paper's central claim—existence, uniqueness, and optimal-order convergence for the Nitsche-type BEM (4.11) with Signorini conditions—does not reduce to its inputs. The nonlinear contact operator BC and data LC are constructed from the Alart–Curnier identities (4.1)–(4.2), cited to external work [1,7,8], and the consistency of (4.11) is checked directly against the continuous solution, not assumed. The error bound of Theorem 5.6 follows from the best-approximation Lemma 5.5, standard approximation estimates, and the coercivity Lemmas 5.2–5.3; the new coercivity estimates involve the contact nonlinearity [Ptau]+ and are proven in this paper, not imported. The citation to the authors' earlier paper [2] supplies Lemma 2.1 and Lemma 3.2 (coercivity and continuity of A+B_D for the linear Dirichlet problem); that is a separately published, parameter-free result whose assumptions do not include the Signorini conclusion, so it is independent support rather than circular self-citation. No fitted constant is renamed as a prediction: the numerical choice tau=0.5/h is motivated by Figure 2 and Theorem 5.6 but is an implementation choice, not a fitted input to the error analysis. A separate, non-circularity concern: as printed, the proof of Lemma 5.3 drops the 2*tau^{-1}<mu_h,[Ptau]_+> cross term when replacing ||[Ptau]_+||^2 by <[Ptau]_+,Ptau>, so the displayed chain from tau^{-1}||[Ptau]_+||^2+2*tau^{-1}<mu_h,[Ptau]_+>+tau^{-1}||mu_h||^2 to tau^{-1}<[Ptau]_+,Ptau>+tau^{-1}||mu_h||^2 is not an identity; this is a proof gap to correct, but it is not a circular dependency.

Assumptions & free parameters 2 free parameters · 5 assumptions · 0 invented entities

The central claim rests on standard boundary integral operator theory, one prior paper by the same group, and a coercivity lemma whose proof has an algebraic gap. No new physical entities are introduced.

free parameters (2)

tau (Nitsche penalty parameter) = 0.5/h in numerical experiments; otherwise any tau asymptotic to h^{-1} per Theorem 5.6
Controls the strength of the contact penalty. Figure 2 sweeps tau to pick 0.5/h; the analysis fixes only the asymptotic scaling, not the constant.
beta_D (Dirichlet penalty parameter) = 0.01 in numerical experiments
Weight for weak Dirichlet imposition. Analysis requires beta_min < beta_D <= C h^{-1}; the specific value 0.01 is chosen, not derived.

assumptions (5)

standard math Lions-Stampacchia existence and uniqueness for the continuous variational inequality (1.1).
Invoked in Section 1 via reference [12].
standard math Continuity and coercivity of the multitrace form A and of A+B_D for beta_D=0 or beta_D>beta_min, imported from Betcke-Burman-Scroggs [2].
Used in Lemmas 2.1 and 3.2 without proof in this paper.
domain assumption The Alart-Curnier augmented Lagrangian relations (4.1)-(4.2) are equivalent to the Signorini complementarity conditions for all tau>0.
Standard contact mechanics identities cited to [1,7]; basis for the nonlinear boundary term.
domain assumption Solution regularity u in H^{3/2+epsilon}(Omega) and traces (u,lambda) in W = H^{1+epsilon}(Gamma) x H^epsilon(Gamma_tilde) for some epsilon in (0,1/2]; lambda is only piecewise regular across face edges.
Assumed in Sections 1 and 5 to obtain convergence rates; not proven for the Signorini problem.
ad hoc to paper Discrete coercivity of A+B_D+B_C as stated in Lemma 5.3.
The proof contains an invalid algebraic equality; the lemma functions as an unverified premise for Theorems 5.4 and 5.6.

how reviews work

0 comments

Cite this review

Pith. "Pith review of Weak imposition of Signorini boundary conditions on the boundary element method." pith.science (2026). https://pith.science/paper/AYI5OEWZ

@misc{pith2026190805580,
  author       = {Pith},
  title        = {Pith review of: Weak imposition of Signorini boundary conditions on the boundary element method},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/AYI5OEWZ}},
  note         = {Machine review of arXiv:1908.05580}
}

read the original abstract

We derive and analyse a boundary element formulation for boundary conditions involving inequalities. In particular, we focus on Signorini contact conditions. The Calder\'on projector is used for the system matrix and boundary conditions are weakly imposed using a particular variational boundary operator designed using techniques from augmented Lagrangian methods. We present a complete numerical a priori error analysis and present some numerical examples to illustrate the theory.

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

23 extracted references · 23 canonical work pages

[2]

Betcke, E

T. Betcke, E. Burman, and M. W. Scroggs , Boundary element methods with weakly imposed boundary conditions, SIAM Journal on Scientiﬁc Computing, 41 (2019), pp. A1357– A1384

work page 2019
[1]

Alart and A

P. Alart and A. Curnier , A mixed formulation for frictional contact problems prone t o Newton like solution methods , Computer Methods in Applied Mechanics and Engineering, 92 (1991), pp. 353–375

work page 1991
[3]

Betcke, M

T. Betcke, M. W. Scroggs, and W. ´Smigaj, Product algebras for Galerkin discretizations of boundary integral operators and their applications . submitted to ACM Transactions on Mathematical Software, 2018

work page 2018
[4]

Buffa and S

A. Buffa and S. H. Christiansen , A dual ﬁnite element complex on the barycentric reﬁne- ment, Mathematics of Computation, 76 (2007), pp. 1743–1769

work page 2007
[5]

Burman, P

E. Burman, P. Hansbo, and M. G. Larson , The penalty-free Nitsche method and noncon- forming ﬁnite elements for the Signorini problem , SIAM Journal on Numerical Analysis, 55 (2017), pp. 2523–2539

work page 2017
[6]

Chernov, M

A. Chernov, M. Maischak, and E. Stephan , A priori error estimates for hp penalty BEM for contact problems in elasticity , Computer Methods in Applied Mechanics and Engineering, 196 (2007), pp. 3871–3880

work page 2007
[7]

Chouly and P

F. Chouly and P. Hild , A Nitsche-based method for unilateral contact problems: nu merical analysis, SIAM Journal on Numerical Analysis, 51 (2013), pp. 1295–13 07

work page 2013
[8]

Chouly, P

F. Chouly, P. Hild, and Y. Renard , Symmetric and non-symmetric variants of Nitsche’s method for contact problems in elasticity: theory and numer ical experiments, Mathematics of Computation, 84 (2015), pp. 1089–1112

work page 2015

Show all 23 references

[9]

Gimperlein, F

H. Gimperlein, F. Meyer, C. ¨Ozdemird, and E. P. Stephan , Time domain boundary ele- ments for dynamic contact problems , Computer Methods in Applied Mechanics and Engi- neering, 333 (2018), pp. 147–175. WEAK IMPOSITION OF SIGNORINI BOUNDARY CONDITIONS ON BEM 17

2018
[10]

Han , A direct boundary element method for Signorini problems , Mathematics of Computa- tion, 55 (1990), pp

H. Han , A direct boundary element method for Signorini problems , Mathematics of Computa- tion, 55 (1990), pp. 115–128

1990
[11]

Han, The boundary ﬁnite element methods for Signorini problems , in Numerical Methods for Partial Diﬀerential Equations, Y.-I

H.-d. Han, The boundary ﬁnite element methods for Signorini problems , in Numerical Methods for Partial Diﬀerential Equations, Y.-I. Zhu and B.-Y. Guo, eds., Springer, 1987, pp. 38–49

1987
[12]

J. L. Lions and G. Stampacchia , Variational inequalities , Communications on Pure and Applied Mathematics, 20 (1967), pp. 493–519

1967
[13]

Maischak and E

M. Maischak and E. P. Stephan , Adaptive hp-versions of BEM for Signorini problems , Applied Numerical Mathematics, 54 (2005), pp. 425 – 449

2005
[14]

J. Nitsche , ¨Uber ein Variationsprinzip zur L¨ osung von Dirichlet-Prob lemen bei Verwendung von Teilr¨ aumen, die keinen Randbedingungen unterworfen s ind, Abhandlungen aus dem Mathematischen Seminar der Universit¨ at Hamburg, 36 (1971 ), pp. 9–15

1971
[15]

Schmit and G

H. Schmit and G. Schneider , Boundary element solution of the Dirichlet-Signorini prob lem by a penalty method , Applicable Analysis, 51 (1993), pp. 175–186

1993
[16]

M. W. Scroggs , Eﬃcient computation and applications of the Calder´ on proj ector, PhD thesis, University College London, 2019

2019
[17]

Spann , On the boundary element method for the Signorini problem of t he Laplacian , Nu- merische Mathematik, 65 (1993), pp

W. Spann , On the boundary element method for the Signorini problem of t he Laplacian , Nu- merische Mathematik, 65 (1993), pp. 337–356

1993
[18]

Steinbach , On a generalized L2 projection and some related stability estimates in Sobolev spaces, Numer Math, 90 (2002), pp

O. Steinbach , On a generalized L2 projection and some related stability estimates in Sobolev spaces, Numer Math, 90 (2002), pp. 775–786

2002
[19]

Steinbach , Numerical approximation methods for elliptic boundary val ue problems , Springer, 2008

O. Steinbach , Numerical approximation methods for elliptic boundary val ue problems , Springer, 2008. Finite and boundary elements

2008
[20]

Steinbach , Boundary element methods for variational inequalities , Numerische Mathe- matik, 126 (2014), pp

O. Steinbach , Boundary element methods for variational inequalities , Numerische Mathe- matik, 126 (2014), pp. 173–197

2014
[21]

Temam , Navier-Stokes equations: Theory and numerical analysis , vol

R. Temam , Navier-Stokes equations: Theory and numerical analysis , vol. 2 of Studies in math- ematics and its applications, North-Holland Publishing, 1 977
[22]

Zhang and X

S. Zhang and X. Li , An augmented Lagrangian method for the Signorini boundary v alue problem with BEM , Boundary Value Problems, 2016 (2016), p. 62

2016
[23]

Zhang and J

S. Zhang and J. Zhu , A projection iterative algorithm boundary element method f or the Sig- norini problem , Engineering Analysis with Boundary Elements, 37 (2013), p p. 176–181

2013

Pith tools

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

[1] [2]

Betcke, E

T. Betcke, E. Burman, and M. W. Scroggs , Boundary element methods with weakly imposed boundary conditions, SIAM Journal on Scientiﬁc Computing, 41 (2019), pp. A1357– A1384

work page 2019

[2] [1]

Alart and A

P. Alart and A. Curnier , A mixed formulation for frictional contact problems prone t o Newton like solution methods , Computer Methods in Applied Mechanics and Engineering, 92 (1991), pp. 353–375

work page 1991

[3] [3]

Betcke, M

T. Betcke, M. W. Scroggs, and W. ´Smigaj, Product algebras for Galerkin discretizations of boundary integral operators and their applications . submitted to ACM Transactions on Mathematical Software, 2018

work page 2018

[4] [4]

Buffa and S

A. Buffa and S. H. Christiansen , A dual ﬁnite element complex on the barycentric reﬁne- ment, Mathematics of Computation, 76 (2007), pp. 1743–1769

work page 2007

[5] [5]

Burman, P

E. Burman, P. Hansbo, and M. G. Larson , The penalty-free Nitsche method and noncon- forming ﬁnite elements for the Signorini problem , SIAM Journal on Numerical Analysis, 55 (2017), pp. 2523–2539

work page 2017

[6] [6]

Chernov, M

A. Chernov, M. Maischak, and E. Stephan , A priori error estimates for hp penalty BEM for contact problems in elasticity , Computer Methods in Applied Mechanics and Engineering, 196 (2007), pp. 3871–3880

work page 2007

[7] [7]

Chouly and P

F. Chouly and P. Hild , A Nitsche-based method for unilateral contact problems: nu merical analysis, SIAM Journal on Numerical Analysis, 51 (2013), pp. 1295–13 07

work page 2013

[8] [8]

Chouly, P

F. Chouly, P. Hild, and Y. Renard , Symmetric and non-symmetric variants of Nitsche’s method for contact problems in elasticity: theory and numer ical experiments, Mathematics of Computation, 84 (2015), pp. 1089–1112

work page 2015