Nodal lengths of eigenfunctions in the disc

# Nodal lengths of eigenfunctions in the disc

Xiaolong Han Michael Murray  and  Chuong Tran Department of Mathematics, California State University, Northridge, CA 91325, USA
###### Abstract.

In this paper, we derive the sharp lower and upper bounds of nodal lengths of Laplacian eigenfunctions in the disc. Furthermore, we observe a geometric property of the eigenfunctions whose nodal curves maximize the nodal length.

###### Key words and phrases:
Laplacian eigenfunctions, nodal sets, geodesics
###### 2010 Mathematics Subject Classification:
58J50, 35J05, 35P15
\@testdef

undefined

## 1. Introduction

In an - smooth and compact Riemannian manifold , let be the Laplacian and be an eigenfunction with eigenvalue , i.e. . If has smooth boundary, we impose Dirchlet or Neumann boundary condition. Yau [Y] conjectured that

 c1√λ≤Hn−1(N(u))≤c2√λ (1.1)

for some constants depending on and independent of the eigenvalues . Here, denotes the - Hausdorff measure and denotes the nodal set of the function .

If the metric is analytic, then (1.1) was proved by Donnelly-Fefferman [DF1] (see also Lin [Li] for the upper bound); if is smooth, then Logunov [Lo1, Lo2] recently showed that

 c1√λ≤Hn−1(N(u))≤c2λα,

in which depends only on . There are partial results in this direction [Bru, ChMu, CoMi, DF2, DF3, HL, HaSi, HeSo, HW, SZ1, SZ2] etc, c.f. the survey by Zelditch [Z2].

In this paper, we are concerned with the precise and sharp dependence of the constants and in (1.1) on the geometry . That is, write

 H1(M)=liminfλ→∞Hn−1(N(u))√λandH2(M)=limsupλ→∞Hn−1(N(u))√λ. (1.2)

Thanks to [DF1, Lo2], we know that on analytic manifolds and on smooth manifolds.

Regarding the two limits in (1.2), we also pursue the categorization of the eigenfunctions (in relation to the geometry of the manifold) which saturate and of as . That is, what geometric properties do the nodal sets of these eigenfunctions achieving the limits in (1.2) have? Our primary interest is to categorize the sequence of eigenfunctions whose nodal curves are geodesics in the manifold, if these eigenfunctions exist. See Corollary 3 and Problem 4.

We begin from the easiest case, an interval . The -th Dirichlet eigenfunction is

 uj(x)=sin(jπLx) with eigenvalue λj=(jπL)2, j=1,2,3,...

The nodal set (in the interior of the domain) of is a collection of nodal points

 N(uj)={Lj⋅l:l=1,...,j−1}.

Hence, the size of the nodal set of the -th Dirichlet eigenfunction on , i.e. the number of nodal points, is

 H0(N(uj))=j−1=L√λjπ−1,j=1,2,3,...

Therefore, in (1.2) we have that

 H1(IL)=H2(IL)=limλ→∞H0(N(u))√λ=1πL.

One can similarly show the same results for Neumann eigenfunctions in .

If , then very little is known about (1.2). To the authors’ knowledge, the first result in this direction is due to Brüning-Gromes [BG, Equation (8)]: They remarked that in an irrational rectangle with side-lengths and for which is irrational,

 H1(R)=1πArea(R)andH2(R)=√2πArea(R). (1.3)

However, and are not known in more general rectangles. In §2.1, we discuss finding and in the rectangles and the tori.

Gichev [G, Theorem 3] proved that on the - unit sphere ,

 H2(Sn)=limsupλ→∞Hn−1(N(u))√λ=Vol(Sn−1)=Γ((n+1)/2)Γ(n/2)√πVol(Sn), (1.4)

in which is the volume of . For example, . In fact, the eigenfunctions on the sphere are spherical harmonics (i.e. homogeneous harmonic polynomials restricted to the sphere) and Gichev proved a stronger result that

 Hn−1(N(u))≤kVol(Sn−1),

in which is the homogeneous degree of . Moreover, the equation in the above inequality is obtained by the Gaussian beams (i.e. highest weight spherical harmonics). One then deduces (1.4) by observing that . In the same paper [G, Page 563], Gichev conjectured that on ,

and the limit is achieved by the zonal harmonics.

Our main result is to provide the case in the disc for which both of the sharp constants and in (1.2) are explicitly proved, see Theorem 1; moreover, we observe the geometric properties of the eigenfunctions which achieve the as , see Corollary 3.

Let be the unit disc. Consider the eigenfunctions with Dirichlet boundary condition. In polar coordinates , the real-valued Dirichlet eigenfunctions are

 uk,s(r,θ)=Jk(√λk,s⋅r)sin(kθ+θ0),where k=0,1,2,3,... and s=1,2,3,...

Here, is the -th Bessel function and , where is the -th nonnegative zero of , and . So all the Dirichlet eigenvalues are the squares of zeros of Bessel functions. In particular, the eigenvalues are distinct for different values of and , (c.f. [W, Section 15.23]) and the multiplicity of is two with eigenspace spanned by

 Jk(√λk,s⋅r)sin(kθ)andJk(√λk,s⋅r)cos(kθ).

The nodal set (in the interior of the disc) of the eigenfunction with eigenvalue is a collection of radials (i.e. diameters) and concentric circles with radii , . Our main theorem states that

###### Theorem 1.

In the disc , let be a Dirichlet eigenfunction with eigenvalue . Then

in which the limit is achieved by the eigenfunctions as ; and

in which the limit is achieved by the eigenfunctions as .

###### Remark.

The same results as in Theorem 1 hold for Neumann eigenfunctions in the disc. See the discussion in §2.3.

A simple dilation gives

###### Corollary 2.

In the disc , let be a Dirichlet eigenfunction with eigenvalue . Then

and

###### Remark.

Some bounds of in (1.1) on Riemannian surfaces are previously known, with which one can have some non-sharp estimates of . In a Euclidean domain , Brüning-Gromes [BG, Equation (4)] proved that

 H1(Ω)≥12j0,1Area(Ω),

where, as before, is the first zero of the Bessel function . On a smooth Riemannian surface , Savo [S, Theorem 13] proved that

 H1(M)≥111Area(M).

So our calculation of in Corollary 2 can be regarded as the sharp improvement of these results applied to the disc.

The other problem in question is to characterize the (possible) geometric properties of the eigenfunctions that achieve or of as . Here, we make the observation that the nodal set of

 uk,1=Jk(√λk,1⋅r)sin(kθ+θ0),

is a collection of diameters that pass through the origin. Hence,

###### Corollary 3.

In the disc ,

is saturated by a sequence of eigenfunctions whose nodal curves in the interior are geodesics, i.e. pieces of straight lines.

Recall that on proved by Gichev [G], is saturated by the Gaussian beams, whose nodal sets are totally geodesic. Based on these evidence, we propose the following problem.

###### Problem 4.

In what manifold , one has that

 H2(M)=limsupλ→∞Hn−1(N(u))√λ=limk→∞Hn−1(N(uk))√λk

for a sequence of eigenfunctions with eigenvalues such that the nodal sets of are totally geodesic in the interior of ?

###### Remark.

The answer to Problem 4 is positive on the spheres by Gichev [G] and in the disc by Corollary 3. In the irrational rectangles (see §2.1), the nodal curves of all eigenfunctions are geodesics so the answer to Problem 4 is trivially positive. It would be interesting to see if Problem 4 holds in other rectangles (or tori). On a general manifold, the answer to Problem 4 is not known and in fact it is not even known whether there exists a sequence of eigenfunctions whose nodal sets are totally geodesic.

In all the manifolds that we consider in this paper (irrational rectangles and tori, spheres, and discs), we have that . So a natural question follows as

###### Problem 5.

In what manifold , ?

###### Remark.

In the case when , there is a unique limit of as for all the eigenfunctions. This is not known to be positive on any manifold with dimension higher than one.

On an analytic manifold, one can extend the Laplacian eigenfunctions to a complex neighborhood of the manifold. In [Z1, Corollary 1.2], Zelditch showed that on an analytic manifold with ergodic geodesic flow, there is a full density subsequence of eigenfunctions for which has a unique limit as . Here, denotes the complex extension of the eigenfunctions and denotes the complex hypersurface measure of the nodal set of . Even though this result is for the complex extensions of a full density subsequence of eigenfunctions, it suggests that on manifolds with ergodic geodesic flow (e.g. negatively curved manifolds), the answer to Problem 5 might be positive.

## 2. Proof of Theorem 1

### 2.1. The irrational rectangles and tori

Before proving Theorem 1, we discuss the proof of (1.3) in an irrational rectangle , where is irrational. (This is observed in [BG].) We then make some remarks about finding and in more general rectangles and tori.

The Dirichlet eigenfunctions in have the form

 uk,j(x,y)=sin(πkax)sin(πjby) (2.1)

with the eigenvalues

 λk,j=(πka)2+(πjb)2,where k,j=1,2,3...

Notice that if is irrational, then all the eigenvalues are simple. Indeed, if for another pair , , then

 k2−~k2a2=~j2−j2b2,

which forces since is irrational. The nodal set (in the interior of ) consists of line segments of length and line segments of length . So the nodal length of is

 H1(N(uk,j))=(k−1)b+(j−1)a.

Hence,

 H1(N(uk,j))√λk,j=Area(R)π×(k−1)b+(j−1)a√(kb)2+(ja)2.

One then sees from for that

 H1(R)=liminfλ→∞H1(N(uk,j))√λk,j=1πArea(R), (2.2)

in which the limit is achieved by as and by as , and

 H2(R)=limsupλ→∞H1(N(uk,j))√λk,j=√2πArea(R), (2.3)

in which the limit is achieved by such that as .

###### Remark.
• The above proof holds with little modification for Neumann eigenfunctions in these irrational rectangles.

• In a more general rectangles, the eigenvalues may not be simple and can have high multiplicity, e.g. in the rectangle , there are eigenvalues with multiplicity of the order as . In the case of high multiplicity, one has to estimate the precise nodal lengths of linear combinations of eigenfunctions of the form (2.1). These linear combinations have complex nodal portraits and the problem of finding and becomes challenging.

• In the disc, the eigenvalues have multiplicity two. Therefore, in the following subsection, we can use explicit formulae to deduce and .

If we identify the two opposing sides of the rectangle and define the torus (i.e. without boundary), then the real-valued eigenfunctions are spanned by

 sin(2πkax±2πjby)andcos(2πkax±2πjby)

with eigenvalue

 (2πka)2+(2πjb)2,%wherek,j=0,1,2,3...

Given that is irrational, the eigenvalues are not simple (except when ) but their multiplicity is uniformly bounded by . A similar argument as in the corresponding irrational rectangle shows that the same results of and in (2.2) and (2.3) hold on the torus. However, and remain unknown on other tori, for the same reason as described in the above remark.

### 2.2. Proof of Theorem 1

Now we prove Theorem 1. Recall that the Dirichlet eigenfunction with eigenvalue has the form

 uk,s=Jk(√λk,s⋅r)sin(kθ+θ0),where k=0,1,2,3,... and s=1,2,3,...

Since the nodal length is independent of here, we assume without loss of generality. The nodal set of is a collection of diameters and concentric circles with radii , . In particular, consists of circles only and consists of diameters only.

Here, we provide the graphs of the nodal curves of some eigenfunctions with different nodal portrait.

From left to right: , , and . Their eigenvalues are approximately and one can check that , which is a reflection of Theorem 1.

To estimate

we pick any subsequence of such that

 limλ→∞H1(N(u))√λexists,

and divide into three cases.

• Case 1: tends to infinity and is bounded;

• Case 2: tends to infinity and is bounded;

• Case 3: and both tend to infinity.

Case 1. As , is bounded so . First set . Then the nodal set of is the union of radials, that is,

 H1(N(uk,1))=2k,

and

 limk→∞H1(N(uk,1))√λk,1=limk→∞2kjk,1=2.

Here, we use the fact that from [AS, Equation 9.5.14, pp. 371],

 jk,1=k+O(k13)as k→∞. (2.4)

This argument works for all the subsequences of for which is bounded and . Indeed, If is bounded by , then

 2k≤H1(N(uk,s))=2πs−1∑l=1jk,ljk,s+2k≤2πM+2k.

So by squeezing,

 limk→∞, s boundedH1(N(uk,s))√λk,s=2.

Here, we need to use a formula [AS, Equation 9.5.22, pp. 371] that if is bounded, then

 jk,s=k+o(k)as k→∞.

Case 2. As , is bounded so . First set . Then the nodal set of is the union of concentric circles with radii , , that is,

 H1(N(u0,s))=2πs−1∑l=1j0,lj0,s,

and

 lims→∞H1(N(u0,s))√λ0,s=2πlims→∞s−1∑l=1j0,lj20,s=2πlims→∞∑sl=1(l−14)π[(s−14)π]2=1.

Here, we use the fact that from [AS, Equation 9.5.12, pp. 371]: If , then

 jk,s=(s+k2−14)π+O(s−1)as s→∞. (2.5)

This argument works for all the subsequences of for which is bounded and . Indeed, If is bounded by , then contains circles and at most diameters. Hence,

 2πs−1∑l=1jk,ljk,s≤H1(N(uk,s))=2πs−1∑l=1jk,ljk,s+2k≤2πs∑l=1jk,ljk,s+2M.

So by squeezing, using (2.5) again, we have that

 lims→∞, k boundedH1(N(uk,s))√λk,s=1.

Case 3. As , . Suppose that in such a subsequence

 limλ→∞H1(N(uk,s))√λk,s=p.

Our goal is then to prove that .

We need the uniform bounds of the zeros of Bessel functions as . By [Bre, Equation (1) in Theorem 1], we have that

 jk,s>k+23|as−1|32for k=0,1,2,3,...%ands=1,2,3,...

Here, is the -th negative zero of the Airy function. By [AS, Equations 10.4.94 and 10.4.105], we have that as ,

 as=−[3π(4s−1)8]23[1+O(s−2)].

Hence,

 jk,s>k+23|as−1|32=k+πs+O(s−1). (2.6)

We now proceed to prove the upper bound that . Using (2.6),

 H1(N(uk,s))√λk,s=2π∑s−1l=1jk,ljk,s+2kjk,s<2πs+2kk+πs+O(s−1)≤2as k,s→∞.

We then prove the lower bound that . By [Bre, Equation (2) in Theorem 1], we have that

 jk,s<π2k+23|as|32=π2k+πs+O(s−1)for k=0,1,2,3,... and s=1,2,3,... (2.7)

Using (2.6) and (2.7), we compute that

 s−1∑l=1jk,ljk,s ≥ 1π2k+πs+O(s−1)s−1∑l=1[k+πl+O(l−1)] ≥ π2(s−1)s+(s−1)k+O(s)π2k+πs+O(s−1) ≥ s2,if s and k are large enough.
###### Remark.

Notice that , , are fractions which are distributed in the interval . If , then by the asymptotic formula (2.5), these fractions are rather equidistributed. So the above inequality is natural in this case. If , then by (2.4) and (2.7), one sees that and therefore for all . So the above inequality is natural in this case as well. The above inequality in fact shows that it is true for all sufficiently large and .

Now by (2.7) again,

 H1(N(uk,s))√λk,s=2π∑s−1l=1jk,ljk,s+2kjk,s≥πs+2kπ2k+πs+O(s−1)≥1as k,s→∞.

Hence, the lower bound is proved.

### 2.3. Neumann eigenfunctions

The Neumann eigenfunctions in can be written as

 vk,s(r,θ)=Jk(√μk,s⋅r)sin(kθ+θ0),where k=0,1,2,3,... and s=1,2,3,...

Here, is the -th Bessel function and , where is the -th nonnegative zero of . So all the Neumann eigenvalues are the squares of zeros of the derivatives of Bessel functions.

Here for Neumann eigenfunctions, we could repeat the argument in the previous subsection, using instead the estimates of . However, notice that Neumann eigenfunctions in extends to and in fact defines a Dirichlet eigenfunction in a slightly larger disc. So we can estimate the nodal set of Neumann eigenfunctions by Corollary 2 for Dirichlet eigenfunctions.

Indeed, the zeros and interlace according to

 k≤j′k,s

See [AS, Equation 9.5.2]. Using this relation, we see that extends from to as a Dirichlet eigenfunction with

 R=jk,sj′k,s→1as k or s→∞. (2.9)

Now the nodal set of in and the nodal set of in differ by the radials in . That is,

 H1(N(vk,s))=H1(N(uk,s))−2k(R−1).

Hence,

 H1(N(vk,s))j′k,s=H1(N(uk,s))−2k(R−1)jk,s⋅R.

By (2.8) and (2.9), we see that

 2k(R−1)jk,s→0as jk,s→∞.

Then applying Corollary 2 and again (2.9), we have that

 liminfλ→∞H1(N(v))√μ=1andlimsupλ→∞H1(N(v))√μ=2.

## Acknowledgments

XH wants to thank Stephen Breen, Andrew Hassell, Hamid Hezari, and Steve Zelditch for all the discussions that are related to this article, in particular, Problem 4; XH also wants to thank Zeév Rudnick for informing him the results in Gichev [G] and Werner Horn for his translation of Brüning-Gromes [BG].

## References

• [AS] M. Abramowitz and I. Stegun, Handbook of mathematical functions with formulas, graphs, and mathematical tables. National Bureau of Standards. U.S. Government Printing Office, Washington, DC, 1964.
• [Bre] S. Breen, Uniform upper and lower bounds on the zeros of Bessel functions of the first kind. J. Math. Anal. Appl. 196 (1995), no. 1, 1–17.
• [Bru] J. Brüning, Über Knoten von Eigenfunktionen des Laplace-Beltrami-Operators. Math. Z. 158 (1978), no. 1, 15–21.
• [BG] J. Brüning and D. Gromes, Ïber die Länge der Knotenlinien schwingender Membranen. Math. Z. 124 (1972), 79–82.
• [ChMu] S. Chanillo and B. Muckenhoupt, Nodal geometry on Riemannian manifolds. J. Differential Geom. 34 (1991), no. 1, 85–91.
• [CoMi] T. Colding and W. Minicozzi, Lower bounds for nodal sets of eigenfunctions. Comm. Math. Phys. 306 (2011), no. 3, 777–784.
• [D] R.-T. Dong, Nodal sets of eigenfunctions on Riemann surfaces. J. Differential Geom. 36 (1992), no. 2, 493–506.
• [DF1] H. Donnelly and C. Fefferman, Nodal sets of eigenfunctions on Riemannian manifolds. Invent. Math. 93 (1988), no. 1, 161–183.
• [DF2] H. Donnelly and C. Fefferman, Nodal sets for eigenfunctions of the Laplacian on surfaces. J. Amer. Math. Soc. 3 (1990), no. 2, 333–353.
• [DF3] H. Donnelly and C. Fefferman, Growth and geometry of eigenfunctions of the Laplacian. Analysis and partial differential equations, 635–655, Lecture Notes in Pure and Appl. Math., 122, Dekker, New York, 1990.
• [HL] X. Han and G. Lu, A geometric covering lemma and nodal sets of eigenfunctions. Math. Res. Lett. 18 (2011), no. 2, 337–352.
• [HaSi] R. Hardt and L. Simon, Nodal sets for solutions of elliptic equations. J. Differential Geom. 30 (1989), no. 2, 505–522.
• [HeSo] H. Hezari and C. Sogge, A natural lower bound for the size of nodal sets. Anal. PDE 5 (2012), no. 5, 1133–1137.
• [HW] H. Hezari and Z. Wang, Lower bounds for volumes of nodal sets: an improvement of a result of Sogge-Zelditch. Spectral geometry, 229–235, Proc. Sympos. Pure Math., 84, Amer. Math. Soc., Providence, RI, 2012.
• [G] V. M. Gichev, Some remarks on spherical harmonics. St. Petersburg Math. J. 20 (2009), no. 4, 553â-567.
• [Li] F.-H. Lin, Nodal sets of solutions of elliptic and parabolic equations. Comm. Pure Appl. Math. 44 (1991), no. 3, 287–308.
• [Lo1] A. Logunov, Nodal sets of Laplace eigenfunctions: polynomial upper estimates of the Hausdorff measure. arXiv:1605.02587.
• [Lo2] A. Logunov, Nodal sets of Laplace eigenfunctions: proof of Nadirashvili’s conjecture and of the lower bound in Yau’s conjecture. arXiv:1605.02589.
• [S] A. Savo, Lower bounds for the nodal length of eigenfunctions of the Laplacian. Ann. Global Anal. Geom. 19 (2001), no. 2, 133–151.
• [SZ1] C. Sogge and S. Zelditch, Lower bounds on the Hausdorff measure of nodal sets. Math. Res. Lett. 18 (2011), no. 1, 25–37.
• [SZ2] C. Sogge and S. Zelditch, Lower bounds on the Hausdorff measure of nodal sets II. Math. Res. Lett. 19 (2012), no. 6, 1361–1364.
• [W] G. N. Watson, A treatise on the theory of Bessel functions. Cambridge University Press, Cambridge, England; The Macmillan Company, New York, 1944.
• [Y] S.-T. Yau, Open problems in geometry. Proc. Sympos. Pure Math. Vol. 54, Part 1, Providence RI: Amer. Math. Soc. 1993, pp. 1–28.
• [Z1] S. Zelditch, Complex zeros of real ergodic eigenfunctions. Invent. Math. 167 (2007), no. 2, 419–443.
• [Z2] S. Zelditch, Eigenfunctions and nodal sets. Surveys in differential geometry. Geometry and topology, 237–308, Surv. Differ. Geom., 18, Int. Press, Somerville, MA, 2013.
You are adding the first comment!
How to quickly get a good reply:
• Give credit where it’s due by listing out the positive aspects of a paper before getting into which changes should be made.
• Be specific in your critique, and provide supporting evidence with appropriate references to substantiate general statements.
• Your comment should inspire ideas to flow and help the author improves the paper.

The better we are at sharing our knowledge with each other, the faster we move forward.
The feedback must be of minimum 40 characters and the title a minimum of 5 characters