The Spectral Position of Neumann Domains on the Torus


Neumann domains of Laplacian eigenfunctions form a natural counterpart of nodal domains. The restriction of an eigenfunction to one of its nodal domains is the first Dirichlet eigenfunction of that domain. This simple observation is fundamental in many works on nodal domains. We consider a similar property for Neumann domains. Namely, given a Laplacian eigenfunction f and its Neumann domain \(\Omega \), what is the position of \(\left. f\right| _{\Omega }\) in the Neumann spectrum of \(\Omega \)? The current paper treats this spectral position problem on the two-dimensional torus. We fully solve it for separable eigenfunctions on the torus and complement our analytic solution with numerics for random waves on the torus. These results answer questions from (Band and Fajman in Ann Henri Poincaré, 17(9):2379–2407, 2016; Zelditch in Surv Differ Geom 18:237–308, 2013) and raise new ones.

This is a preview of subscription content, log in to check access.

Fig. 1
Fig. 2
Fig. 3
Fig. 4
Fig. 5
Fig. 6
Fig. 7
Fig. 8


  1. 1.

    The reason for this is that the boundary of a general Neumann domain might include a cusp and in general we do not have an explicit expression of the cusp.

  2. 2.

    We thank Michael Levitin for suggesting this experiment and pointing out FEM++ for this purpose, [*, [19]].

  3. 3.

    To apply the theory in [7, 28] for our case, we take the group to be \(C_{2}\times C_{2}\) (the direct product of two copies of the cyclic group, \(C_{2}\)) with its regular representation.

  4. 4.

    We thank John Hannay for pointing out this interesting geometrical meaning to us.

  5. 5.

    As a matter of fact, \(y_{0}=-b\) also gives a Neumann line, but it is connected to a different saddle point.


  1. 1.

    Abramowitz, M., Stegun, I.A.: Handbook of Mathematical Functions with Formulas, Graphs, and Mathematical Tables. U.S. Government Printing Office, Washington, D.C. (1964)

    Google Scholar 

  2. 2.

    Acosta, G., Armentano, M.G., Durán, R.G., Lombardi, A.L.: Nonhomogeneous Neumann problem for the Poisson equation in domains with an external cusp. J. Math. Anal. Appl. 310(2), 397–411 (2005)

    MathSciNet  Article  Google Scholar 

  3. 3.

    Agmon, S.: Lectures on Elliptic Boundary Value Problems. AMS Chelsea Publishing, Providence, RI (2010)

    Google Scholar 

  4. 4.

    Albert, J. H.: Topology of the nodal and critical point sets for eigenfunctions of elliptic operators. Thesis (Ph.D.)–Massachusetts Institute of Technology, ProQuest LLC, Ann Arbor, MI (1972)

  5. 5.

    Alon, L., Band, R., Bersudsky, M., Egger, S.: Neumann domains on graphs and manifolds. arXiv:1805.07612

  6. 6.

    Band, R., Fajman, D.: Topological properties of Neumann domains. Ann. Henri Poincaré 17(9), 2379–2407 (2016)

    MathSciNet  Article  Google Scholar 

  7. 7.

    Band, R., Berkolaiko, G., Joyner, C. H., Liu, W.: Quotients of finite-dimensional operators by symmetry representations. (2017) arXiv: 1711.00918

  8. 8.

    Band, R., Cox, G., Egger, S.: Spectral properties of Neumann domains via the Dirichlet-to-Neumann operator

  9. 9.

    Bandle, C.: Extremaleigenschaften von Kreissektoren und Halbkugeln. Comment. Math. Helv. 46, 356–380 (1971)

    MathSciNet  Article  Google Scholar 

  10. 10.

    Bandle, C.: Isoperimetric Inequalities and Applications, volume 7 of Monographs and Studies in Mathematics. Pitman (Advanced Publishing Program), Boston (1980)

  11. 11.

    Courant, R.: Nachr. Ges. Wiss. Göttingen Math Phys. Ein allgemeiner Satz zur Theorie der Eigenfunktione selbstadjungierter Differentialausdrücke K1, 81–84 (1923)

    Google Scholar 

  12. 12.

    Edmunds, D.E., Evans, W.D.: Spectral Theory and Differential Operators. The Clarendon Press, Oxford University Press, New York (1987)

    Google Scholar 

  13. 13.

    Elon, Y., Gnutzmann, S., Joas, C., Smilansky, U.: Geometric characterization of nodal domains: the area-to-perimeter ratio. J. Phys. A 40, 2689 (2007)

    MathSciNet  Article  Google Scholar 

  14. 14.

    Federer, H.: Geometric measure theory. Die Grundlehren der mathematischen Wissenschaften, Band 153. Springer, New York (1969)

    Google Scholar 

  15. 15.

    Filonov, N.: On an inequality for the eigenvalues of the Dirichlet and Neumann problems for the Laplace operator. Algebra i Analiz 16(2), 172–176 (2004)

    MathSciNet  Google Scholar 

  16. 16.

    Friedlander, L.: Some inequalities between Dirichlet and Neumann eigenvalues. Arch. Ration. Mech. Anal. 116(2), 153–160 (1991)

    MathSciNet  Article  Google Scholar 

  17. 17.

    Girouard, A., Nadirashvili, N., Polterovich, I.: Maximization of the second positive Neumann eigenvalue for planar domains. J. Differ. Geom. 83(3), 637–661 (2009)

    MathSciNet  Article  Google Scholar 

  18. 18.

    Grisvard, P.: Elliptic problems in nonsmooth domains. SIAM (2011)

  19. 19.

    Hecht, F.: New development in FreeFem++. J. Num. Math. 20(3–4), 251–265 (2012)

    MathSciNet  MATH  Google Scholar 

  20. 20.

    Kröger, P.: Upper bounds for the Neumann eigenvalues on a bounded domain in Euclidean space. J. Funct. Anal. 106(2), 353–357 (1992)

    MathSciNet  Article  Google Scholar 

  21. 21.

    Kuchment, P., Zeng, H.: Convergence of spectra of mesoscopic systems collapsing onto a graph. J. Math. Anal. Appl. 258(2), 671–700 (2001)

    MathSciNet  Article  Google Scholar 

  22. 22.

    Maggi, F.: Sets of Finite Perimeter and Geometric Variational Problems. Cambridge University Press, Cambridge (2012)

    Google Scholar 

  23. 23.

    Magnus, W., Oberhettinger, F., Soni, R.P.: Formulas and Theorems for the Special Functions of Mathematical Physics. Springer, New York (1966)

    Google Scholar 

  24. 24.

    Math, P., Zeng, H.: Asymptotics of spectra of Neumann Laplacians in thin domains. In: Karpeshina, Y., Stolz, G., Weikard, R., Zeng, Y. (eds.) Advances in Differential Equations and Mathematical Physics (Birmingham, AL, 2002). American Mathematical Society, Providence (2003)

    Google Scholar 

  25. 25.

    Mazya, V.G., Poborchi, S.V.: Differentiable Functions on Bad Domains. World Scientific Publishing Co, Geneva (1997)

    Google Scholar 

  26. 26.

    Mazzolo, A., de Mulatier, C., Zoia, A.: Cauchy’s formulas for random walks in bounded domains. J. Math. Phys. 55(8), 083308 (2014)

    MathSciNet  Article  Google Scholar 

  27. 27.

    McDonald, R.B., Fulling, S.A.: Neumann nodal domains. Philos. Trans. R. Soc. Lond. Ser. A 372, 25 (2007)

  28. 28.

    Parzanchevski, O., Band, R.: Linear representations and isospectrality with boundary conditions. J. Geom. Anal. 20(2), 439–471 (2010)

    MathSciNet  Article  Google Scholar 

  29. 29.

    Payne, L.E.: Isoperimetric inequalities and their applications. SIAM Rev. 9, 453–488 (1967)

    MathSciNet  Article  Google Scholar 

  30. 30.

    Pleijel, A.: Remarks on Courant’s nodal line theorem. Commun. Pure Appl. Math. 9(3), 543–550 (1956)

    MathSciNet  Article  Google Scholar 

  31. 31.

    Pólya, G.: Remarks on the foregoing paper. J. Math. Phys. 31, 55–57 (1952)

    MathSciNet  Article  Google Scholar 

  32. 32.

    Reed, M.: Methods of Modern Mathematical Physics. I–4. Functional Analysis. Academic Press, New York (1972)

    Google Scholar 

  33. 33.

    Reed, M., Simon, B.: Methods of Modern Mathematical Physics, vol. IV. Academic Press, New York, N.Y. (1978)

    Google Scholar 

  34. 34.

    Rubinstein, J., Schatzman, M.: Arch. Ration. Mech. Anal. Variational problems on multiply connected thin strips. I. Basic estimates and convergence of the Laplacian spectrum. 160(4), 271–308 (2001)

    Google Scholar 

  35. 35.

    Szegö, G.: Inequalities for certain eigenvalues of a membrane of given area. J. Rational Mech. Anal. 3, 343–356 (1954)

    MathSciNet  MATH  Google Scholar 

  36. 36.

    Uhlenbeck, K.: Generic properties of eigenfunctions. Am. J. Math. 98(4), 1059–1078 (1976)

    MathSciNet  Article  Google Scholar 

  37. 37.

    Weinberger, H.F.: An isoperimetric inequality for the \(N\)-dimensional free membrane problem. J. Rational Mech. Anal. 5, 633–636 (1956)

    MathSciNet  MATH  Google Scholar 

  38. 38.

    Zelditch, S.: Eigenfunctions and nodal sets. Surv. Differ. Geom. 18, 237–308 (2013)

    MathSciNet  Article  Google Scholar 

Download references


We are grateful to Michael Levitin for his encouraging comments and useful ideas for further investigations. We would like to thank Emanuel Milman for stimulating discussions and for pointing out helpful references. We thank Gregory Berkolaiko and Mark Dennis for interesting discussions in various stages of this ongoing work. We thank Luc Hillairet and Graham Cox for pointing out to us the second proof of Theorem 1.4,(1). Band and Egger were supported by ISF (Grant No. 494/14). Taylor was funded by the Leverhulme Trust Research Programme Grant No. RP2013-K-009.

Author information



Corresponding author

Correspondence to Sebastian K. Egger.

Additional information

Publisher's Note

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

A. The Boundary of \(\Omega _{a,b}^{\mathrm {star}}\) and its Area

A. The Boundary of \(\Omega _{a,b}^{\mathrm {star}}\) and its Area

We consider a separable eigenfunction \(f_{a,b}\) on the torus, (8), and its star-like Neumann domain, \(\Omega _{a,b}^{\mathrm {star}}\). In this appendix, we derive the explicit expression for the boundary of \(\Omega _{a,b}^{\mathrm {star}}\) (Lemma 7.1) and show that it is of class C (Lemma 7.3). This boundary characterization is needed to justify the application of some Sobolev space analysis (done in Proposition 3.1). Furthermore, we perform here an asymptotic calculation of the \(\Omega _{a,b}^{\mathrm {star}}\) area (Lemma 7.4) which is used in the proofs of Theorem 1.4, (1) (first proof) and Lemma 5.2.

Lemma 7.1

We have

$$ \Omega _{a,b}^{\mathrm {star}}=\left\{ \left( x,y\right) ~:~\left| x\right|<a,~~\left| y\right| <\gamma _{a,b}(x)\right\} , $$


$$\begin{aligned} \gamma _{a,b}(x):=\frac{2b}{\pi }\arcsin \left( \left[ \cos \left( \frac{\pi }{2a}x\right) \right] ^{\left( \frac{a}{b}\right) ^{2}}\right) . \end{aligned}$$


To prove the lemma, we parameterize the Neumann line which connects the extremal point, (a, 0) to the saddle point, (0, b) (see Fig. 2,(ii)) and show that it is given by (33). The other four Neumann lines which form the boundary of \(\Omega _{a,b}^{\mathrm {star}}\) are obtained by noting that \(\Omega _{a,b}^{\mathrm {star}}\) is symmetric with respect to horizontal and vertical reflections (see Fig. 3). Plugging the expression of the eigenfunction (8) in the flow equations (2), we get

$$ \begin{pmatrix}\dot{x}\\ \dot{y} \end{pmatrix}=-\frac{\pi }{2}\begin{pmatrix}a^{-1}\cos \left( \frac{\pi }{2a}x\right) \cos \left( \frac{\pi }{2b}y\right) \\ b^{-1}\sin \left( \frac{\pi }{2a}x\right) \sin \left( \frac{\pi }{2b}y\right) \end{pmatrix}. $$

Hence, the tangent to any gradient flow line is

$$ \frac{\mathrm {d}y}{\mathrm {d}x}=\frac{a}{b}\tan \left( \frac{\pi }{2a}x\right) \tan \left( \frac{\pi }{2b}y\right) . $$

Integrating this, we obtain the gradient flow lines

$$\begin{aligned} y(x)=\frac{2b}{\pi }\arcsin \left( \sin \left( \frac{\pi }{2b}y_{0}\right) \left[ \cos \left( \frac{\pi }{2a}x\right) \right] ^{\left( \frac{a}{b}\right) {}^{2}}\right) , \end{aligned}$$

where \((0,y_{0})\) is a point through which the gradient flow line passes. Note that for \(-b{<}y_{0}{<}b\), each of the gradient flow lines in (34) is connected to the extremal point (a, 0), but only the one with \(y_{0}=b\) is connected to the saddle point (0, b) and hence it is the desired Neumann lineFootnote 5. \(\square \)

Remark 7.2

From the proof of Lemma 7.1, one may also obtain that there is no gradient flow line which connects two saddle points of the eigenfunction \(f_{a,b}\). From this, we conclude that \(f_{a,b}\) is a Morse–Smale function [5, Proposition A.7].

The next lemma shows that the boundary of \(\Omega _{a,b}^{\mathrm {star}}\) is regular enough for applying an appropriate Sobolev space analysis. The classification of the boundary in the lemma is based on [12, Definition 4.1].

Lemma 7.3

The boundary of the star-like domain, \(\partial \Omega _{a,b}^{\mathrm {star}}\), is of class C.

Namely, for any \(\varvec{p}\in \partial \Omega _{a,b}^{\mathrm {star}}\), there exists an open neighborhood \(U(\varvec{p})\subset \mathbb {R}^{2}\) and a continuous function \(h\in C(I)\) on an interval \(I\subset \mathbb {R}\) such that for suitable local Cartesian coordinates

$$\begin{aligned} \partial \Omega _{a,b}^{\mathrm {star}}\cap U(\varvec{p})=\left\{ (s,t):\ t=h(s),\ s\in I\right\} \end{aligned}$$



Since the boundary consists of gradient flow lines, the claim is obvious for every point \(\varvec{p}\) not being an end point of such a flow line (i.e., for every \(\varvec{p}\) which is not a critical point). At a saddle point, any two adjacent Neumann lines meet with an angle of \(\tfrac{\pi }{2}\), [27, Theorem 3.2.]. Hence, the boundary at a neighborhood of a saddle point is also a continuous function. At the extremal points \((\pm a,0)\), adjacent Neumann lines meet with an angle of 0 and form a cusp. We derive the asymptotics of \(\gamma _{a,b}(a-x)\), \(x\rightarrow 0^{+}\). Using

$$\begin{aligned} \cos \left( \frac{\pi }{2a}(a-x)\right)&=\sin \left( \tfrac{\pi }{2a}x\right) =\tfrac{\pi }{2a}x+\mathrm {O}\left( \left( \tfrac{\pi }{2a}x\right) {}^{3}\right) \\ (1+x)^{\beta }&=1+\mathrm {O}(x)\quad \quad \text {for }\beta >0\\ \arcsin (x)&=x+\mathrm {O}\left( x^{3}\right) , \end{aligned}$$

we get that for \(x\rightarrow 0^{+}\)

$$\begin{aligned} \gamma _{a,b}(a-x)=\frac{2b}{\pi }\arcsin \left( \left[ \cos \left( \frac{\pi }{2a}(a-x)\right) \right] ^{\left( \frac{a}{b}\right) ^{2}}\right) =\frac{2b}{\pi }\left( \frac{\pi }{2a}x\right) {}^{\left( \frac{a}{b}\right) {}^{2}}+\mathrm {O}\left( x^{3\left( \frac{a}{b}\right) {}^{2}}\right) . \end{aligned}$$

These asymptotics show that \(\gamma _{a,b}\) is strictly monotonically decreasing in a left neighborhood of (a, 0) and its inverse exists there. Hence, the condition (35) is satisfied in a neighborhood of (a, 0) by choosing

$$ h(s)={\left\{ \begin{array}{ll} \gamma _{a,b}^{-1}(s) &{} s>0\\ \gamma _{a,b}^{-1}(-s) &{} s<0 \end{array}\right. }. $$

\(\square \)

Finally, we use the expression of \(\gamma _{a,b}\) to bound the area of \(\Omega _{a,b}^{\mathrm {star}}\) which is needed in the proofs of Theorem 1.4,(1) (first proof) and Lemma 5.2 (see (28) in that proof).

Lemma 7.4

There exists \(c>1\) such if then

$$\begin{aligned} \frac{1}{ab}\left( \frac{b}{a}+\frac{a}{b}\right) \left| \Omega _{a,b}^{\mathrm {star}}\right| <\frac{2}{\pi }(j_{0,1})^{2} \end{aligned}$$

where \(j_{0,1}\approx 2.4048\) is the first zero of \(J_{0}\), the zeroth Bessel function.


Using Lemma 7.1 we have


We may use the Taylor expansion of \(\mathrm {ln}\left[ \cos \left( z\right) \right] \), which converges for \(\left| z\right| <\frac{\pi }{2}\) (see e.g., [1, 4.3.72] and [23, p. 27]) to obtain the bound

$$\begin{aligned} \forall z\in (0,\frac{\pi }{2}),\quad \left[ \cos \left( z\right) \right] ^{\left( \frac{a}{b}\right) ^{2}}<\exp \left[ -\frac{1}{2}\left( \frac{a}{b}\right) ^{2}z^{2}\right] . \end{aligned}$$
Fig. 9

The left-hand side of (37) plotted as a function of . The right-hand side of (37) is indicated together with the corresponding value

Fig. 10

The first eigenvalue of \(-\Delta ^{h}_{a,b}\), the first eigenvalue of \(-\Delta ^{v}_{a,b}\), and \(\lambda _{a,b}\) plotted as a function of (i.e., we chose \(a=1\))

Another bound which we use is


To validate (40), we may observe that both functions at the RHS and LHS coincide for \(w=0\) and \(w=1\) and further check that the difference does not vanish anywhere else in \(\left( 0,1\right) \) (for example, since the difference has only a single critical point in this interval).

Plugging the bounds (39), (40) in (38) and using also the monotonicity of \(\arcsin (w)\) for \(w\in (0,1)\) we get


where moving to the last line we used integration over (half) Gaussian.

From the above, we get \(\frac{1}{ab}\left( \frac{a}{b}+\frac{b}{a}\right) \left| \Omega _{a,b}^{\mathrm {star}}\right| \lessapprox 2.7014\cdot \left( 1+\left( \frac{b}{a}\right) ^{2}\right) \). Now, since \(\frac{2}{\pi }(j_{0,1})^{2}\approx 3.68\) we get that (37) holds if \(\frac{b}{a}\) is small enough. \(\square \)

Remark 7.5

From the proof, one easily gets that (37) holds for . Numerically, it seems that choosing \(c\approx 1.1407\) already guarantees this bound. This can be seen in Fig. 9 and shows that the methods in the proof of Theorem 1.4, (2) cannot reduce the constant in the theorem below \(c\approx 1.1407\). Yet, a numerical experiment (see Figure 10) shows that the statement of the theorem should be valid also for \(c=1\) (which is the optimal result).

Rights and permissions

Reprints and Permissions

About this article

Verify currency and authenticity via CrossMark

Cite this article

Band, R., Egger, S.K. & Taylor, A.J. The Spectral Position of Neumann Domains on the Torus. J Geom Anal (2020).

Download citation


  • Neumann domains
  • Neumann lines
  • nodal domains
  • Laplacian eigenfunctions
  • Morse–Smale complexes

Mathematics Subject Classification

  • 58C40
  • 58J50
  • 35P05