variation of the nazarov sodin
play

Variation of the Nazarov-Sodin constant for random plane waves and - PowerPoint PPT Presentation

Variation of the Nazarov-Sodin constant for random plane waves and arithmetic random waves Par Kurlberg (KTH Stockholm) Igor Wigman, KCL Random Waves in Oxford Oxford, June 19, 2018 1. Motivation & Background General Setup (, )


  1. Variation of the Nazarov-Sodin constant for random plane waves and arithmetic random waves Par Kurlberg (KTH Stockholm) Igor Wigman, KCL Random Waves in Oxford Oxford, June 19, 2018

  2. 1. Motivation & Background

  3. General Setup (𝑁, 𝑕) – Compact smooth surface (can ο‚— generalize higher dimensions) Ξ” Laplace-Beltrami on M ο‚— Eigenfunctions: (boundary condition) ο‚— πœ‡ π‘˜ β‰₯ 0 Ξ”πœ’ π‘˜ + πœ‡ π‘˜ πœ’ π‘˜ = 0 ο‚— Orthonormal basis of L 2 (M,dVol), πœ‡ π‘˜ β†’ ∞

  4. Nodal components & domains βˆ’1 (0) ο‚— Nodal set: π‘Ž πœ’ π‘˜ = πœ’ π‘˜ ο‚— Nodal components: Connected components βˆ’1 (0) . of πœ’ π‘˜ ο‚— Nodal domains: Connected components of βˆ’1 (0) smooth 𝑁 βˆ– πœ’ π‘˜ ο‚— Nodal count: How many components (domains)?

  5. Interesting Questions – non-local Nodal count (Nazarov-Sodin `09,`12, `15, ο‚— Kurlberg-W `14, `17) ο‚— T opology, nesting,geometry(Sarnak-W, Beliaev-W) ο‚— Local: π‘€π‘“π‘œ 𝐡βˆͺ𝐢 𝑔 = π‘€π‘“π‘œ 𝐡 𝑔 + π‘€π‘“π‘œ 𝐢 𝑔 𝐡 ∩ 𝐢 = βˆ… ο‚— Semilocality: β€œMost” of the nodal domains of 𝑆 diameter πœ‡ , R>>0. ο‚— Approximate locally

  6. Nodal count (deterministic) ο‚— Nodal Count. Courant: 𝑂 πœ’ π‘˜ ≀ π‘˜ 𝑂 πœ’ π‘˜ ο‚— Pleijel: limsup ≀ 0.691 … π‘˜ π‘˜β†’βˆž ο‚— Constant improved by 3 βˆ™ 10 βˆ’9 (Bourgain) ο‚— No lower bound 𝑂 πœ’ π‘˜ β‰₯ 2 Nodal picture for the square, arbitrarily high energy. A. Stern’s thesis, Gottingen, 1925. Courtesy of P. Sarnak.

  7. Berry’s Random Wave Model ο‚— M chaotic. As πœ‡ β†’ ∞, πœ’ π‘˜ β€œbehave randomly” wavenumber πœ‡ monochromatic wave ℝ 2 1 𝐾 𝑓 𝑗( πœ‡ 𝑦,𝜘 π‘˜ +πœ” π‘˜ ) 𝐾 𝔒𝑓 π‘˜=1 𝑣 πœ‡ (𝑦) = ο‚— Scale invariant, assume 𝑣 = 𝑣 1 ο‚— Centered Gaussian, covariance E[𝑣(𝑦) βˆ™ 𝑣(𝑧)] = 𝐾 0 (|𝑦 βˆ’ 𝑧|) ο‚— Spectral measure – arc length on unit circle

  8. 2. Random Band Limited Functions

  9. Random Band-Limited Functions ο‚— Fix M – smooth n-manifold, 0 ≀ 𝛽 ≀ 1 πœ‡ x = π›½πœ‡β‰€πœ‡ π‘˜ β‰€πœ‡ 𝑏 π‘˜ πœ’ π‘˜ (𝑦) , 𝑏 π‘˜ - N(0,1) i.i.d. 𝑔 ( 𝛽 = 1 summation over πœ‡ βˆ’ 𝑝 πœ‡ ≀ πœ‡ π‘˜ ≀ πœ‡ ) ο‚— Covariance function 𝔽 𝑔 πœ‡ 𝑦 𝑔 πœ‡ 𝑧 = πœ’ π‘˜ (𝑦)πœ’ π‘˜ (𝑧) i.e. the spectral projector.

  10. Example 1. Random Spherical harmonics ο‚— 𝛽 = 1, 𝑁 = 𝒯 2 , 2d sphere. ο‚— 𝔽 π‘ˆ π‘š x βˆ™ π‘ˆ π‘š y = 𝑄 π‘š cos(𝑒 𝑦, 𝑧 ) . ο‚— 𝑄 π‘š cos(𝑒) β‰ˆ 𝐾 0 π‘šπ‘’ Legendre fast uniform ο‚— Scales Berry’s RWM RWM Random spherical harmonics A. Barnett

  11. 𝛽 =1 vs 𝛽 =0 (Alex Barnett) 𝛽 =1 𝛽 =0 Random spherical harmonics β€œReal Fubini- Study”

  12. Example 2. T oral eigenfunctions. ℝ 2 ο‚— π•Œ = 𝑁 = β„€ 2 π‘œ 𝑦 = 𝜈 2 =π‘œ 𝑏 𝜈 βˆ™ 𝑓 ο‚— 𝑔 𝑦, 𝜈 𝑏 𝜈 standard Gaussian i.i.d. (save to 𝑏 βˆ’πœˆ = 𝑏 𝜈 ) β€œ arithmetic random waves ” 𝜈 2 = π‘œ ο‚— Summation over 𝜈 ∈ β„€ 2 : lattice points on radius π‘œ circle

  13. More general: limiting ensembles ο‚— Natural scaling around any point of M. ο‚— Scaling for covariance (values & derivatives) (classical Hormander, Lax) 𝔽 𝑔 πœ‡ x βˆ™ 𝑔 πœ‡ y β‰ˆ 𝐿 𝛽 πœ‡ βˆ™ 𝑒 𝑦, 𝑧 ο‚— 𝐿 𝛽 π‘₯ = 𝐿 𝛽 π‘₯ = 𝛽≀ π‘₯ ≀1 𝑓 π‘₯, 𝜊 π‘’πœŠ ο‚— Canzani-Hanin `16 𝛽 = 1 thin window. ο‚— Define 𝑕 ∞ on ℝ 2 , β€œclean” covariance 𝔽[𝑕 ∞ 𝑨 βˆ™ 𝑕 ∞ 𝑨′ ] = 𝐿 𝛽 𝑨 βˆ’ 𝑨′

  14. Limiting ensembles (cont.) ο‚— 𝔽[𝑕 ∞ 𝑨 βˆ™ 𝑕 ∞ 𝑨′ ] = 𝐿 𝛽 𝑨 βˆ’ 𝑨′ ο‚— 𝑕 ∞ scaling limit 𝑔 πœ‡ (everywhere) ο‚— 𝑕 ∞ depends on 𝛽 , not on M, x (universality) ο‚— Spectral measure 𝛽 1 ο‚— Relevant: nodal structures of 𝑕 ∞ restricted on ball 𝐢 𝑆 , 𝑆 β†’ ∞. ο‚— E.g. nodal count of domains lying in 𝐢 𝑆 .

  15. 3. Nazarov-Sodin Constant

  16. Scale invariant (Euclidean) case ο‚— 𝐺: ℝ 2 β†’ ℝ stationary Gaussian field ο‚— 𝜍 spectral measure of 𝐺 ο‚— 𝑂(𝐺; 𝑆) is the number of connected components (domains) of 𝐺 inside 𝐢(𝑆) ο‚— Assuming: 1. 𝐺 ergodic ( 𝜍 has no atoms) 2. 𝐺 smooth. 3. Non-degeneracy ο‚— Nazarov-Sodin (`12,`15): 𝑑 = 𝑑 𝑂𝑇 (𝜍) β‰₯ 0 𝐹[𝑂 𝐺; 𝑆 ] = 𝑑 βˆ™ 𝑆 2 + 𝑝 π‘†β†’βˆž (𝑆 2 ) ο‚— β€œUsually” 𝑑 > 0 (support of 𝜍 )

  17. Nodal count band-limited functions = 𝑑 βˆ™ 𝑆 2 + 𝑝 π‘†β†’βˆž (𝑆 2 ) ο‚— 𝑑 = 𝑑 𝑂𝑇 (𝜍) , E 𝑂 𝐺; 𝑆 ο‚— Stronger convergence in mean (ergodicity) 𝑂 𝐺;𝑆 βˆ’π‘‘βˆ™π‘† 2 E β†’ 0 𝑆 2 ο‚— Band-limited functions 𝑑 = 𝑑 𝑂𝑇 𝜍 𝛽 > 0 𝜍 𝛽 area measure annulus πœ‡ )]~𝑑 βˆ™ πœ‡ (NS `12,`15) ο‚— E[𝑂(𝑔 𝛽 1 ο‚— Convergence in mean

  18. Random spherical harmonics ο‚— 𝑣 = Plane monochromatic waves (RWM) ο‚— 𝐹[𝑂 𝑣; 𝑆 ]~𝑑 𝑆𝑋𝑁 βˆ™ 𝑆 2 , universal NS constant 2𝜌 > 0 percolation? π‘’πœ„ ο‚— 𝑑 𝑆𝑋𝑁 = 𝑑 𝑂𝑇 ο‚— 𝐹 𝑂(π‘ˆ π‘š ) ~𝑑 𝑆𝑋𝑁 βˆ™ π‘š 2 (NS `09) ο‚— Convergence in mean 1 ο‚— Exponential probability concentration

  19. 4. Variation on Nazarov-Sodin consant

  20. Generalise NS constant ο‚— Restrict to 𝜍 supported on the unit ball 𝒬 (spectral moments), includes band limited case ο‚— Proposition 1(Kurlberg-W): c = 𝑑 𝑂𝑇 (𝜍) , E[𝑂 𝐺; 𝑆 ] = 𝑑 βˆ™ 𝑆 2 + 𝑃(𝑆) , 𝜍 ∈ 𝒬 arbitrary, absolute constant (uniform) ο‚— 𝑑 𝑂𝑇 (𝜍) bounded (e.g. critical points Kac-Rice) ο‚— No convergence in mean. Can construct 𝑂 𝐺;𝑆 βˆ’π‘‘βˆ™π‘† 2 examples E doesn’t vanish 𝑆 2 ο‚— Example: 𝜍 atomic supported at 0. 𝐺 ≑ π‘‘π‘π‘œπ‘‘π‘’ , Gaussian 𝑂 𝐺; 𝑆 ≑ 0 , β‡’ 𝑑 𝑂𝑇 𝜍 = 0.

  21. Variation of NS constant ο‚— Proposition 2 (Kurlberg-W): 𝑂 𝐺;𝑆 βˆ’π‘‘βˆ™π‘† 2 exists 𝑒 𝑂𝑇 𝜍 ≔ lim π‘†β†’βˆž E 𝑆 2 (β€œNS discrepancy functional”) non -uniform, discontinous ο‚— Theorem 1 (Kurlberg-W): 𝑑 𝑂𝑇 𝜍 : 𝒬 β†’ ℝ β‰₯0 is continuous (weak* topology on 𝒬 ). ο‚— Corollary: 𝑑 𝑂𝑇 𝜍 attains an interval [0, 𝑑 0 ] ( 𝒬 is essential) ο‚— Q: Is it true that 𝑑 = 𝑑 𝑆𝑋𝑁 , uniquely ?

  22. 5. Toral Eigenfunctions

  23. Example 2. T oral eigenfunctions. ℝ 2 ο‚— π•Œ = 𝑁 = β„€ 2 π‘œ 𝑦 = 𝜈 2 =π‘œ 𝑏 𝜈 βˆ™ 𝑓 ο‚— 𝑔 𝑦, 𝜈 𝑏 𝜈 standard Gaussian i.i.d. (save to 𝑏 βˆ’πœˆ = 𝑏 𝜈 ) β€œ arithmetic random waves ” 𝜈 2 = π‘œ ο‚— Summation over 𝜈 ∈ β„€ 2 : lattice points on radius π‘œ circle π‘œ 𝑦 ) nodal count ο‚— 𝑂(𝑔

  24. On the 2 squares problem 𝑏, 𝑐 πœ—β„€ 2 : 𝑏 2 + 𝑐 2 = π‘œ ο‚— 𝑠 2 π‘œ = # ο‚— On average 𝑠 2 π‘œ ~c βˆ™ log(π‘œ) (E. Landau) Equidistributed Exceptional β€œ Cilleruelo ” generic 𝑠 2 (n) β†’ ∞ 1 1 ο‚— Partial classification (P . Kurlberg-IW `15)

  25. Some pics π‘œ = 1105 32 directions π‘œ = 9676418088513347624474653 256 directions Fragment, domains long and narrow

  26. On the 2 squares problem 1 𝑠 2 (π‘œ) 𝜈 2 =π‘œ πœ€ 𝜈/ π‘œ probability 𝒯 1 ο‚— 𝜐 π‘œ = (spectral measure) π‘’πœ„ ο‚— Equidistributed ( 𝜐 π‘œ π‘˜ β‡’ 𝜐 ) 𝜐 π‘œ π‘˜ β‡’ 2𝜌 ο‚— Angular distribution β†­ Nodal structure Local: Krishnapur-Kurlberg-W `13, Rudnick-W `14, Rossi-W `17 ο‚— Nonlocal: N(𝑔 π‘œ 𝑦 ) – total nodal count Nazarov-Sodin `12,`15+Kurlberg-W `14,17

  27. T oral eigenfunctions arithmetic π‘œ 𝑦 = 𝜈 2 =π‘œ 𝑏 𝜈 βˆ™ 𝑓 𝑦, 𝜈 ο‚— 𝑔 random waves 1 𝑠 2 (π‘œ) 𝜈 2 =π‘œ πœ€ 𝜈/ π‘œ on 𝑇 1 ο‚— 𝜐 π‘œ = ο‚— Apply N-S: if 𝜐 π‘œ β‡’ 𝜐 then 𝑑 = 𝑑 𝑂𝑇 𝜐 (generalised) 𝐹 𝑂(𝑔 π‘œ 𝑦 ) = 𝑑 βˆ™ π‘œ + 𝑝 π‘œ 𝑂 𝑔 π‘œ 𝑦 βˆ’π‘‘ 𝑆𝑋𝑁 ο‚— Generic 𝐹 β†’ 0 π‘œ ο‚— Exponential concentration (Y. Rozenshein `15)

  28. T oral eigenfunctions (cont.) arithmetic π‘œ 𝑦 = 𝜈 2 =π‘œ 𝑏 𝜈 βˆ™ 𝑓 𝑦, 𝜈 ο‚— 𝑔 random waves 1 𝑠 2 (π‘œ) 𝜈 2 =π‘œ πœ€ 𝜈/ π‘œ on 𝑇 1 ο‚— 𝜐 π‘œ = ο‚— Theorem 2 (Kurlberg-W): 1. Uniformly 𝐹 𝑂(𝑔 π‘œ 𝑦 ) = 𝑑 𝑂𝑇 (𝜐 π‘œ ) βˆ™ π‘œ + 𝑃 π‘œ 2. 𝑑 𝑂𝑇 𝜐 = 0 iff 𝜐 is Cilleruelo or its tilt ( 𝜐 restricted, in particular by symmetries)

  29. T oral eigenfunctions (cont.) ο‚— 𝑑 𝑂𝑇 𝜐 = 0 iff 𝜐 is Cilleruelo or its tilt (restricted) ο‚— 𝑑 𝑂𝑇 𝜐 attains an interval 0, 𝑑 1 . ο‚— Q.: Is it true that 𝑑 1 = 𝑑 0 = 𝑑 𝑆𝑋𝑁 uniquely? ο‚— Q.: For Cilleruelo: 𝐹 𝑂(𝑔 π‘œ 𝑦 ) - ? π‘œ 𝑦 ) β†’ ∞ - ? 𝐹 𝑂(𝑔 π‘œ - ? 𝐹 𝑂(𝑔 π‘œ 𝑦 ) ≫ ο‚— Meanwhile full classification 𝑑 𝑂𝑇 𝜍 = 0 (Beliaev-McAuley-Muirhead). Same for 𝑒 𝑂𝑇 𝜍 ?

Recommend


More recommend