kp theory planar bicolored networks in the disk and
play

KP theory, planar bicolored networks in the disk and rational - PowerPoint PPT Presentation

KP theory, planar bicolored networks in the disk and rational degenerations of M-curves Simonetta Abenda (UniBo) and Petr G. Grinevich (LITP,RAS) CIRM Luminy, April 9, 2019 Simonetta Abenda (UniBo) and Petr G. Grinevich (LITP,RAS) KP, networks


  1. Results Consider the case when k , β ∈ N . Then k β 2 ≥ 1 so we expect MoM N ( k , β ) ∼ ρ k ,β N k 2 β 2 − k +1 . Theorem [B.-Keating (2018)] Let k , β ∈ N . Then MoM N ( k , β ) is a polynomial in N . Theorem [B.-Keating (2018)] Let k , β ∈ N . Then with ρ k ,β an explicit function of k and β , MoM N ( k , β ) = ρ k ,β N k 2 β 2 − k +1 + O ( N k 2 β 2 − k ) . Emma Bailey Moments of Moments CIRM 2019 4 / 6

  2. Example MoM N (2 , 3) = ( N +1)( N +2)( N +3)( N +4)( N +5)( N +6)( N +7)( N +8)( N +9)( N +10)( N +11) 1722191327731024154944441889587200000000 � 12308743625763 N 24+1772459082109872 N 23+121902830804059138 N 22+ × +5328802119564663432 N 21+166214570195622478453 N 20+3937056259812505643352 N 19 +73583663800226157619008 N 18+1113109355823972261429312 N 17+13869840005250869763713293 N 16 +144126954435929329947378912 N 15+1259786144898207172443272698 N 14 +9315726913410827893883025672 N 13+58475127984013141340467825323 N 12 +311978271286536355427593012632 N 11+1413794106539529439589778645028 N 10 +5427439874579682729570383266992 N 9+17564370687865211818995713096848 N 8 +47561382824003032731805262975232 N 7+106610927256886475209611301000128 N 6 +194861499503272627170466392014592 N 5+284303877221735683573377603640320 N 4 +320989495108428049992898521600000 N 3+266974288159876385845370793984000 N 2 � +148918006780282798012340305920000 N +43144523802785397500411904000000 Emma Bailey Moments of Moments CIRM 2019 5 / 6

  3. Thank you Emma Bailey Moments of Moments CIRM 2019 6 / 6

  4. Speed of Convergence in the Gaussian Distribution for Laguerre Ensembles Under Double Scaling Sergey Berezin 1 and Alexander Bufetov 2 1 CNRS, PDMI RAS, 2 CNRS, Steklov IITP RAS E-mail: 1 servberezin@yandex.ru, 2 bufetov@mi-ras.ru

  5. Problem statement Consider Laguerre Unitary Ensemble : M = U ∗ diag { Λ 1 , . . . , Λ n } U, (1) where U is distributed uniformly on the unitary group U ( n ) The random variables Λ 1 , . . . , Λ n have the joint probability density 1 ∏︂ j e − 4 mλ j ∏︂ 1 [ λ j > 0] λ α ( λ k − λ j ) 2 , P n,m ( λ 1 , . . . , λ n ) = (2) Z n,m j j<k where α > − 1 , m ∈ N , and Z n,m is the partition function. Let f ( M ) be a real-valued function defined on the spectrum of M . Our goal is to study the characteristic function ∫︂ e ih ∑︁ f ( λ j ) P n,m ( λ 1 , . . . , λ n ) dλ 1 · . . . · dλ n e ih Tr f ( M ) ]︁ [︁ = (3) E n,m of the linear statistic Tr f ( M ) in a double-scaling limit as n = m → ∞ . Conference “Integrability and Randomness in Mathematical Physics and Geometry”, April 8–12, 2019 1

  6. Main results Let f : R + → R be locally H¨ older continuous such that it admits the analytic continuation to some neighborhood of [0 , 1] . Theorem (Convergence to the Gaussian law) d Tr f ( M ) − n κ [ f ] − → N ( µ [ f ] , K [ f ]) , n = m → ∞ . (4) The linear functionals κ [ f ] , µ [ f ] , and the quadratic functional K [ f ] are given with the explicit formulas. Theorem (Speed of convergence) Let f ( x ) also satisfy f ( x ) = O ( e Ax ) , A > 0 , as x → + ∞ . Define the cumulative distribution functions F n ( x ) and F ( x ) corresponding to Tr f ( M ) − n κ [ f ] − µ [ f ] and to N (0 , K [ f ]) , respectively. Then sup x | F n ( x ) − F ( x ) | = O (1 /n ) , n = m → ∞ . (5) Conference “Integrability and Randomness in Mathematical Physics and Geometry”, April 8–12, 2019 2

  7. The proof of Theorems is based on the Riemann–Hilbert analysis similar to Charlier&Gharakhloo (2019). However, unlike them, we are interested in complex exponents . In such a case the corresponding Hankel determinants and/or the weight of the corresponding orthogonal polynomials can be zero. Also we need the exponents that grow with n . To succeed we adopt the approach from Deift, Its&Krasovsky (2014) and use the deformation of w ( x ) = x α e − 4 nx e ihf ( x ) . (6) into w l,t ( x ) = x α e − 4 nx (︁ 1 − t + te ih 1 [ l<n γ +1] f ( x ) )︁ e ih ( l − 1) 1 [ l<n γ +1] f ( x ) , ˜ (7) We choose ε > 0 small enough so that 1 − t + te ih 1 [ l<n γ +1] f ( x ) ̸ = 0 , γ ∈ [0 , 1] , (8) for all t ∈ [0 , 1] , x in the neighborhood of [0 , 1] , h such that | h | < ε , and for all n, l . Conference “Integrability and Randomness in Mathematical Physics and Geometry”, April 8–12, 2019 3

  8. From Gumbel to Tracy–Widom II, via integer partitions Dan Betea University of Bonn based on joint work with J. Bouttier (Math. Phys. Anal. Geom. 2019, arXiv:1807.09022 [math-ph] ) CIRM 1X.IV.MM19

  9. Partitions • • ◦ • • ◦ • • • ◦ ◦ ◦ ◦ ◦ Figure: Partition (Young diagram) λ = (2 , 2 , 2 , 1 , 1) (Frobenius coordinates (1 , 0 | 4 , 1)) in English, French and Russian notation, with associated Maya diagram (particle-hole representation). Size | λ | = 8, length ℓ ( λ ) = 5. Figure: Skew partitions (Young diagrams) (4 , 3 , 2 , 1) / (2 , 1) (but also (5 , 4 , 3 , 2 , 1) / (5 , 2 , 1) , . . . ) and (4 , 4 , 2 , 1) / (2 , 2) (but also (6 , 4 , 4 , 2 , 1) / (6 , 2 , 2) , . . . )

  10. Counting tableaux A standard Young tableau (SYT) is a filling of a (possibly skew) Young diagram with numbers 1 , 2 , . . . strictly increasing down columns and rows. 1 3 5 6 1 7 2 4 9 3 4 7 2 5 8 6 dim λ := number of SYTs of shape λ and similarly for dim λ/µ .

  11. Measures on partitions There are two natural measures on all partitions: poissonized Plancherel vs. (grand canonical) uniform Prob ( λ ) = e − ǫ 2 ǫ 2 | λ | (dim λ ) 2 Prob ( λ ) = u | λ | � (1 − u i ) vs. ( | λ | !) 2 i ≥ 1 with ǫ ≥ 0 and 1 > u ≥ 0 parameters.

  12. Ulam’s problem and Hammersley last passage percolation Quantity of interest: L = longest up-right path from (0 , 0) to (1 , 1) (= 4 here). Schensted’s theorem yields that, in distribution, L = λ 1 with λ coming from the poissonized Plancherel measure.

  13. The Baik–Deift–Johansson theorem and Tracy–Widom Theorem (BaiDeiJoh 1999) If λ is distributed as poissonized Plancherel, we have: � λ 1 − 2 ǫ � ǫ →∞ Prob lim ≤ s = F TW ( x ) := det(1 − Ai 2 ) ( s , ∞ ) ǫ 1 / 3 with � ∞ Ai 2 ( x , y ) := Ai ( x + s ) Ai ( y + s ) ds . 0 and Ai the Airy function (solution of y ′′ = xy decaying at ∞ ). F TW is the Tracy–Widom GUE distribution. It is by (original) construction the extreme distribution of the largest eigenvalue of a random hermitian matrix with iid standard Gaussian entries as the size of the matrix goes to infinity.

  14. The Erd˝ os–Lehner theorem and Gumbel Theorem (ErdLeh 1941) For the uniform measure Prob ( λ ) ∝ u | λ | we have: � λ 1 < − log(1 − u ) ξ � = e − e − ξ . u → 1 − Prob lim + log u | log u |

  15. The finite temperature Plancherel measure On pairs of partitions µ ⊂ λ ⊃ µ consider the measure Prob ( µ, λ ) ∝ u | µ | · ǫ 2( | λ |−| µ | ) dim 2 ( λ/µ ) ( | λ/µ | !) 2 with u = e − β , β = inverse temperature. ◮ u = 0 yields the poissonized Plancherel measure ◮ ǫ = 0 yields the (grand canonical) uniform measure

  16. The finite temperature Plancherel measure II Theorem (B/Bouttier 2019) ǫ 1 − u → ∞ and u = exp( − α M − 1 / 3 ) → 1 . Then Let M = � λ 1 − 2 M � = F α ( x ) := det(1 − Ai α ) ( s , ∞ ) M →∞ Prob lim ≤ s M 1 / 3 with � ∞ e α s Ai α ( x , y ) := 1 + e α s · Ai ( x + s ) Ai ( y + s ) ds . −∞ the finite temperature Airy kernel.

  17. What is in a part? PPP ( u 4 ǫ 2 ) PPP ( u 3 ǫ 2 ) PPP ( u 2 ǫ 2 ) PPP ( uǫ 2 ) PPP ( ǫ 2 ) With L the longest up-right path in this cylindric geometry, in distribution, Schensted’s theorem states that λ 1 = L + κ 1 where κ is a uniform partition Prob ( κ ) ∝ u | κ | independent of everything else.

  18. A word on the finite temperature Airy kernel Ai α ◮ introduced by Johansson (Joh07) ◮ also appearing as the KPZ crossover kernel: SasSpo10 and AmiCorQua11; in random directed polymers BorCorFer11; cylindric OU processes LeDMajSch15 ◮ interpolates between the Airy kernel and a diagonal exponential kernel: α →∞ Ai α ( x , y ) = Ai 2 ( x , y ) , lim � x 1 α − 1 2 α log(4 πα 3 ) , y α − 1 � α Ai α 2 α log(4 πα 3 ) = e − x δ x , y lim α → 0+ ◮ with F α ( s ) , F TW ( s ), and G ( s ) the Fredholm determinants on ( s , ∞ ) of Ai α , Ai 2 and e − x δ x , y , (Joh07) α →∞ F α ( s ) = F TW ( s ) , lim � s α − 1 � = G ( s ) = e − e − s α → 0+ F α 2 α log(4 πα 3 ) lim

  19. Direct limit to Tracy–Widom Theorem (B/Bouttier 2019) Let u → 1 and ǫ → ∞ in such a way that ǫ (1 − u ) 2 → ∞ . Then we have � λ 1 − 2 M � ǫ Prob ≤ s → F TW ( s ) , M := 1 − u . M 1 / 3

  20. Direct limit to Gumbel Theorem (B/Bouttier 2019) Set u = e − r and assume that r → 0+ and ǫ r 2 → 0+ (with ǫ possibly remaining finite). Then: � r λ 1 − ln I 0 (2 ǫ + ǫ r ) � → e − e − s Prob ≤ s r � π 1 − π e x cos φ d φ is the modified Bessel function of the first kind and order where I 0 ( x ) := 2 π zero.

  21. Thank you!

  22. Last passage percolation Character identities and LPP Duality between determinants and Pfaffians Transition between characters of classical groups, decomposition of Gelfand-Tsetlin paterns, and last passage percolation (joint work with Nikos Zygouras) Elia Bisi University College Dublin Integrability and Randomness in Mathematical Physics and Geometry Marseille, 9 April 2019 1 / 4

  23. Last passage percolation Character identities and LPP Duality between determinants and Pfaffians Last passage percolation (LPP) � L ( 2 n , 2 n ) : = max W i , j π ∈ Π 2 n , 2 n ( i , j ) ∈ π Π 2 n , 2 n is the set of directed paths in { 1 ,..., 2 n } 2 starting from ( 1 , 1 ) and ending at ( 2 n , 2 n ) ; { W i , j } 1 ≤ i , j ≤ 2 n is a field of independent geometric random variables with various symmetries (1 , 1) (1 , 1) (1 , 1) (2 n, 2 n ) (2 n, 2 n ) (2 n, 2 n ) Antidiagonal symmetry Diagonal symmetry Double symmetry 2 / 4

  24. Last passage percolation Character identities and LPP Duality between determinants and Pfaffians Character identities and LPP � 2 n � i = 1 ( µ i mod 2 ) · s ( 2 n ) � � L β ( 2 n , 2 n ) ≤ 2 u β ( p 1 ,..., p 2 n ) P ∝ µ µ ⊆ ( 2 u ) ( 2 n )   u 2 n    �  s CB p i u ( 2 n ) ( p 1 ,..., p 2 n ; β )   =     i = 1   u � 2 n     � s CB λ ( p 1 ,..., p n ; β ) · s CB p i λ ( p n + 1 ,..., p 2 n ; β )   =     i = 1 λ ⊆ u ( n ) (1 , 1) s ( 2 n ) is a classical Schur polynomial; µ s CB λ is a Schur polynomial that interpolates between symplectic and odd orthogonal characters. (2 n, 2 n ) 3 / 4

  25. Last passage percolation Character identities and LPP Duality between determinants and Pfaffians Duality between determinants and Pfaffians Baik-Rains’ formulas and ours show a duality between Pfaffians and determinants, for finite N . Fredholm Pfaffian and Fredholm determinantal expressions of the limiting distribution functions, as N → ∞ . E.g., we obtain: Sasamoto’s Fredholm determinant for the GOE Tracy-Widom distribution in the case of antidiagonal symmetry : F 1 ( s ) = det ( I − B s ) Ferrari-Spohn’s Fredholm determinant for the GSE Tracy-Widom distribution in the case of diagonal symmetry : � � F 4 ( s ) = 1 det ( I − B √ 2 s ) + det ( I + B √ 2 s ) 2 with the kernel being B s ( x , y ) : = Ai ( x + y + s ) on L 2 ( [0 , ∞ )) . 4 / 4

  26. Painlev´ e II τ -function as a Fredholm determinant Harini Desiraju SISSA, Trieste

  27. Introduction • Question: Can the τ -function of Painlev´ e II be expressed as a Fredholm determinant? 1

  28. Introduction • Question: Can the τ -function of Painlev´ e II be expressed as a Fredholm determinant? • Painlev´ e II q ss = sq − 2 q 3 (1) 1

  29. Introduction • Question: Can the τ -function of Painlev´ e II be expressed as a Fredholm determinant? • Painlev´ e II q ss = sq − 2 q 3 (1) • The τ -function of Painlev´ e II is related to its transcendent d 2 ds 2 ln τ [ s ] = − q 2 ( s ) (2) 1

  30. Introduction • Question: Can the τ -function of Painlev´ e II be expressed as a Fredholm determinant? • Painlev´ e II q ss = sq − 2 q 3 (1) • The τ -function of Painlev´ e II is related to its transcendent d 2 ds 2 ln τ [ s ] = − q 2 ( s ) (2) • What is known? 1

  31. Introduction • Question: Can the τ -function of Painlev´ e II be expressed as a Fredholm determinant? • Painlev´ e II q ss = sq − 2 q 3 (1) • The τ -function of Painlev´ e II is related to its transcendent d 2 ds 2 ln τ [ s ] = − q 2 ( s ) (2) • What is known? • Ablowitz-Segur family is a special solution of PII q ( s ) ≈ κ Ai ( s ); κ ∈ C ; s → ∞ (3) 1

  32. Introduction • Question: Can the τ -function of Painlev´ e II be expressed as a Fredholm determinant? • Painlev´ e II q ss = sq − 2 q 3 (1) • The τ -function of Painlev´ e II is related to its transcendent d 2 ds 2 ln τ [ s ] = − q 2 ( s ) (2) • What is known? • Ablowitz-Segur family is a special solution of PII q ( s ) ≈ κ Ai ( s ); κ ∈ C ; s → ∞ (3) • It is a known result that the τ -function in this case is the determinant of the Airy Kernel. τ [ s ] = det[ I − κ 2 K Ai ] | [ s , ∞ )] (4) 1

  33. General Painlev´ e II using IIKS construction • The Riemann Hilbert problem of Painlev´ e II, after some transformations, can be reduced to the following RHP on i R Γ( z ) = 1 + O ( z − 1 ) Γ + ( z ) = Γ − ( z ) J ( z ); z → ∞ (5) as χ 1 χ 2 i R χ 3 χ 4 2

  34. General Painlev´ e II using IIKS construction • The Riemann Hilbert problem of Painlev´ e II, after some transformations, can be reduced to the following RHP on i R Γ( z ) = 1 + O ( z − 1 ) Γ + ( z ) = Γ − ( z ) J ( z ); z → ∞ (5) as χ 1 χ 2 i R χ 3 χ 4 � a ( z ) b ( z ) � = 1 − 2 π if ( z ) g T ( z ) • Using χ i , the jump function is J ( z ) = c ( z ) d ( z ) 2

  35. General Painlev´ e II using IIKS construction • The Riemann Hilbert problem of Painlev´ e II, after some transformations, can be reduced to the following RHP on i R Γ( z ) = 1 + O ( z − 1 ) Γ + ( z ) = Γ − ( z ) J ( z ); z → ∞ (5) as χ 1 χ 2 i R χ 3 χ 4 � a ( z ) b ( z ) � = 1 − 2 π if ( z ) g T ( z ) • Using χ i , the jump function is J ( z ) = c ( z ) d ( z ) • with χ 2 ( z ) + ( b ( z ) − 1) � � � � χ 4 ( z ) 1 χ 1 ( z ) + χ 3 ( z ) a ( z ) f ( z ) = ; g ( z ) = (1+ c ( z ) − a ( z )) 2 π i χ 2 ( z ) + χ 4 ( z ) χ 1 ( z ) + ( a ( z ) − 1) χ 3 ( z ) a ( z ) a ( z ) , b ( z ) , c ( z ) , d ( z ) are given in terms of parabolic cylinder functions. 2

  36. Results • The integrable kernel on L 2 ( i R ) is given by K ( z , w ) = f T ( z ) g ( w ) (6) 2 π i ( z − w ) 3

  37. Results • The integrable kernel on L 2 ( i R ) is given by K ( z , w ) = f T ( z ) g ( w ) (6) 2 π i ( z − w ) • τ -function: τ [ s ] = det( 1 − K ) (7) 3

  38. Results • The integrable kernel on L 2 ( i R ) is given by K ( z , w ) = f T ( z ) g ( w ) (6) 2 π i ( z − w ) • τ -function: τ [ s ] = det( 1 − K ) (7) • τ [ s ] is related to the JMU τ -function as 3 + ν 2 � 2 i ν � ∂ s ln τ [ s ] = ∂ s ln τ JMU + + A ( ν ) (8) s where ν = − 1 2 π i ln(1 − s 1 s 3 ) and s 1 , s 3 are Stokes’ parameters and s is the PII parameter and A ( ν ) is a non-vanishing depending only on ν . 3

  39. References Fokas, A.S., Its, A.R., Kapaev, A.A., Kapaev, A.I., Novokshenov, V.Y. and Novokshenov, V.I., 2006. Painlev´ e transcendents: the Riemann-Hilbert approach (No. 128). American Mathematical Soc. Bertola, M., 2017. The Malgrange form and Fredholm determinants. arXiv preprint arXiv:1703.00046. Its, A.R., Izergin, A.G., Korepin, V.E. and Slavnov, N.A., 1990. Differential equations for quantum correlation functions. International Journal of Modern Physics B, 4(05), pp.1003-1037. Cafasso, M., Gavrylenko, P. and Lisovyy, O., 2017. Tau functions as Widom constants . arXiv preprint arXiv:1712.08546. Bothner, T. and Its, A., 2012. Asymptotics of a Fredholm determinant involving the second Painlev´ e transcendent. arXiv preprint arXiv:1209.5415. 4

  40. Extreme gap problems in random matrix theory Renjie Feng BIMCR, Peking University Renjie Feng (BICMR) 1 / 10

  41. Previous results I: smallest gaps for CUE Let e i θ 1 , · · · , e i θ n be n eigenvalues of CUE, consider n � χ n = δ ( n 4 / 3 ( θ i +1 − θ i ) ,θ i ) . i =1 Theorem (Vinson, Soshnikov, Ben Arous-Bourgade) χ n tends to a Poisson process χ with intensity � 1 � � �� du � u 2 du E χ ( A × I ) = . 24 π 2 π A I The k th smallest gap has limiting density 3 ( k − 1)! x 3 k − 1 e − x 3 . Renjie Feng (BICMR) 2 / 10

  42. Previous results II: smallest gaps for GUE For GUE n � χ n = δ ( n 3 ( λ i +1 − λ i ) ,λ i ) 1 | λ i | < 2 − η 4 i =1 Theorem (Ben Arous-Bourgade, AOP 2013) χ n tends to a Poisson process χ with intensity 1 � � u 2 du )( (4 − x 2 ) 2 dx ) , E χ ( A × I ) = ( 48 π 2 A I where A ⊂ R + and I ⊂ ( − 2 + η, 2 − η ). ( k − 1)! x 3 k − 1 e − x 3 , same as 3 The k th smallest gap has the limiting density CUE. Renjie Feng (BICMR) 3 / 10

  43. New results I: smallest gaps for C β E When β is an positive integer, consider n � χ n = δ β +2 β +1 ( θ i +1 − θ i ) ,θ i ) ( n i =1 Theorem [F.-Wei] χ n tends to a Poisson point process χ with intensity E χ ( A × I ) = A β | I | � u β du , 2 π A where A β = (2 π ) − 1 ( β/ 2) β (Γ( β/ 2+1)) 3 Γ(3 β/ 2+1)Γ( β +1) . For COE, CUE and CSE, A 1 = 1 1 1 24 , A 2 = 24 π, A 4 = 270 π. Renjie Feng (BICMR) 4 / 10

  44. New results II: smallest gaps for GOE For GOE n − 1 χ ( n ) = � δ n 3 / 2 ( λ ( i +1) − λ ( i ) ) i =1 Theorem [F.-Tian-Wei] χ ( n ) converges to a Poisson point process χ with intensity E χ ( A ) = 1 � udu . 4 A the limiting density of the k th smallest gap is 2 ( k − 1)! x 2 k − 1 e − x 2 , same as COE. Conjecture: C β E and G β E share the same smallest gaps. Renjie Feng (BICMR) 5 / 10

  45. Previous III: order of largest gaps For CUE and interior of GUE, m k is the k th largest gap, Theorem (Ben Arous-Bourgade, AOP 2013) For any p > 0 and l n = n o (1) , one has n L p m l n × √ → 1 . 32 ln n Renjie Feng (BICMR) 6 / 10

  46. New results III: fluctuation of largest gaps Theorem (F.-Wei) Let’s denote m k as the k -th largest gap of CUE, and 1 1 τ n 2 ( nm k − (32 ln n ) 2 ) / 4 − (3 / 8) ln(2 ln n ) , k = (2 ln n ) then { τ n k } tends to a Poisson process and τ n k has the limit of the Gumbel distribution, e k ( c 1 − x ) ( k − 1)! e − e c 1 − x . Here, c 1 = 1 12 ln 2 + 3 ζ ′ ( − 1) + ln π 2 . Renjie Feng (BICMR) 7 / 10

  47. New results III: fluctuation of largest gaps Theorem (F.-Wei) √ 4 − x 2 and Let’s denote m ∗ k as the k -th largest gap of GUE, S ( I ) = inf I 1 1 τ ∗ 2 ( nS ( I ) m ∗ 2 ) / 4 + (5 / 8) ln(2 ln n ) , k = (2 ln n ) k − (32 ln n ) { τ ∗ k } tends to a Poisson process and has the limit of the Gumbel distribution, e k ( c 2 − x ) ( k − 1)! e − e c 2 − x . Here, c 2 = 1 12 ln 2 + 3 ζ ′ ( − 1) + M 0 ( I ) depending on I , where M 0 ( I ) = (3 / 2) ln(4 − a 2 ) − ln(4 | a | ) if a + b < 0 , M 0 ( I ) = (3 / 2) ln(4 − b 2 ) − ln(4 | b | ) if a + b > 0, M 0 ( I ) = (3 / 2) ln(4 − a 2 ) − ln(2 | a | ) if a + b = 0 . Renjie Feng (BICMR) 8 / 10

  48. Extreme gaps IV: universality of extreme gaps Recently, our results are generalized for Hermitian/symmetric Wigner matrices with mild assumptions. P. Bourgade, Extreme gaps between eigenvalues of Wigner matrices, arXiv:1812.10376. B. Landon, P. Lopatto, J. Marcinek, Comparison theorem for some extremal eigenvalue statistics, arXiv:1812.10022. Renjie Feng (BICMR) 9 / 10

  49. References Large gaps of CUE and GUE, arXiv:1807.02149. Small gaps of circular beta-ensemble, arXiv:1806.01555 Small gaps of GOE, arXiv:1901.01567. Renjie Feng (BICMR) 10 / 10

  50. MATRIX MODELS FOR CLASSICAL GROUPS AND TOEPLITZ+HANKEL MINORS WITH APPLICATIONS TO CHERN-SIMONS THEORY AND FERMIONIC MODELS David García-García Joint work with Miguel Tierz (arXiv:1901.08922)

  51. MINORS OF TOEPLITZ+HANKEL MATRICES d 1 d 2 d 7 d 3 d 6 d 4 d 10 d 2 d 9 d 8 d 1 d 0 d 7 d 1 d 6 d 2 d 5 d 3 d 9 d 3 d 8 d 9 d 2 . . . . . . . . . . . d 0 . . . . . . d 11 d 1 d 10 d 8 d 7 U N d 1 d 6 d 1 d 5 d 3 d 4 d 2 d 3 d 2 d 7 d 0 N N det N 1 f M dM d 5 d 4 d 1 d 4 d 6 d 0 d 5 d 1 d 3 d 2 d 8 d 4 d 7 d 3 d 6 d 2 d 5 d 1 d 4 d 0 . ∫ f ( e i θ k ) d θ k [ 0 , 2 π ] N | ∆( e i θ ) | 2 ∏ N ! 2 π k = 1

  52. MINORS OF TOEPLITZ+HANKEL MATRICES d 1 d 2 d 7 d 3 d 6 d 4 d 10 d 2 d 9 d 8 d 1 d 0 d 7 d 1 d 6 d 2 d 5 d 3 d 9 d 3 d 8 d 9 d 2 . . . . . . . . . . . d 0 . . . . . . d 11 d 1 d 10 d 8 d 7 . d 3 d 6 d 1 d 5 d 3 d 4 d 2 d 1 d 7 d 2 d 0 N N det N 1 d 5 d 4 d 1 d 4 d 6 d 0 d 5 d 3 d 4 d 2 d 8 d 1 d 7 d 3 d 6 d 2 d 5 d 1 d 0 d 4 ∫ f ( M ) dM = U ( N ) ∫ f ( e i θ k ) d θ k [ 0 , 2 π ] N | ∆( e i θ ) | 2 ∏ 2 π = N ! k = 1

  53. MINORS OF TOEPLITZ+HANKEL MATRICES d 5 d 3 d 7 d 4 d 8 d 2 d 4 d 1 d 0 d 2 d 6 d 1 d 7 d 2 d 8 d 3 d 9 d 6 d 5 d 5 d 3 N d 0 d 2 d 1 d 3 d 2 d 4 d 5 d 1 d 4 d 6 d 5 d 7 d 1 d 3 d 0 d 4 d 3 d 2 det . . . . . . . . . d 11 . . . . . . . . d 1 d 6 d 10 d 1 d 7 d 0 d 8 d 1 d 9 d 2 d 4 d 10 d 6 d 3 d 7 d 2 d 8 d 1 d 9 d 0 N . d 3 d 0 . . . . d 0 d 1 d 2 d 4 d 1 . d 2 d 3 det d 0 d 0 d 1 d 2 d 0 d 1 . N . . . . . . 1 . . . . . . ∫ f ( M ) dM = U ( N ) ∫ f ( e i θ k ) d θ k [ 0 , 2 π ] N | ∆( e i θ ) | 2 ∏ 2 π = N ! k = 1   · · · d − 1 d − 2 d − 3 d − 4 d − 5 · · ·   d − 1 d − 2 d − 3 d − 4     · · · d − 1 d − 2 d − 3     · · ·  d − 1 d − 2  N × N   · · ·   d − 1    

  54. MINORS OF TOEPLITZ+HANKEL MATRICES d 4 d 2 d 6 d 3 d 7 d 4 d 8 d 2 d 1 d 1 d 5 d 0 d 6 d 1 d 7 d 2 d 8 d 3 d 5 d 4 d 3 d 2 det N N d 0 d 2 d 1 d 3 d 4 d 0 d 3 d 5 d 4 d 6 d 5 d 7 d 1 d 3 d 9 d 5 . . . . . . . . . . d 1 . . . . . . . . d 11 d 10 d 2 d 2 d 6 d 1 d 7 d 0 d 8 d 1 d 9 d 10 d 0 d 4 d 6 d 3 d 7 d 2 d 8 d 1 d 9 . . . d 0 d 2 d 3 3 d 2 d 1 d . d 1 d 2 2 d d 1 d 0 5 d 4 d 3 d 2 d 1 d d 0 det N 1 d 1 d 0 d 4 . . . . . . . . . . . . d 0 . . d 3 d 2 d 1 ∫ f ( M ) dM = U ( N ) ∫ f ( e i θ k ) d θ k [ 0 , 2 π ] N | ∆( e i θ ) | 2 ∏ 2 π = N ! k = 1   · · ·   d − 1 d − 3 d − 4         · · ·   d − 1 d − 2 N × N   · · ·   d − 1    

  55. MINORS OF TOEPLITZ+HANKEL MATRICES d 4 d 2 d 6 d 3 d 7 d 4 d 8 d 2 d 1 d 1 d 5 d 0 d 6 d 1 d 7 d 2 d 8 d 3 d 5 d 4 d 3 d 2 det N N d 0 d 2 d 1 d 3 d 4 d 0 d 3 d 5 d 4 d 6 d 5 d 7 d 1 d 3 d 9 d 5 . . . . . . . . . . d 1 . . . . . . . . d 11 d 10 d 2 d 2 d 6 d 1 d 7 d 0 d 8 d 1 d 9 d 10 d 0 d 4 d 6 d 3 d 7 d 2 d 8 d 1 d 9 . . . d 0 d 2 d 3 3 d 2 d 1 d . d 1 d 2 2 d d 1 d 0 5 d 4 d 3 d 2 d 1 d d 0 det N 1 d 1 d 0 d 4 . . . . . . . . . . . . d 0 . . d 3 d 2 d 1 ∫ s λ ( M − 1 ) s µ ( M ) f ( M ) dM = U ( N ) ∫ f ( e i θ k ) d θ k [ 0 , 2 π ] N s λ ( e − i θ ) s µ ( e i θ ) | ∆( e i θ ) | 2 ∏ 2 π = N ! k = 1   · · ·   d − 1 d − 3 d − 4         · · ·   d − 1 d − 2 N × N   · · ·   d − 1    

  56. MINORS OF TOEPLITZ+HANKEL MATRICES d 8 d 2 d 7 d 3 d 6 d 4 d 10 d 2 d 9 d 1 d 0 d 1 d 7 d 1 d 6 d 2 d 5 d 3 d 9 d 3 d 8 d 8 d 9 d 7 . . . . . . . . . . . d 0 . . . . . . d 11 d 1 d 10 d 2 d 1 . d 3 d 6 d 4 d 6 d 3 d 4 d 2 d 1 d 7 d 2 d 0 N N det N 1 d 5 d 5 d 1 d 4 d 0 d 5 d 3 d 4 d 2 d 8 d 1 d 7 d 3 d 6 d 2 d 5 d 1 d 0 d 4 ∫ f ( M ) dM = ( G ( N ) = Sp ( 2 N ) , O ( 2 N ) , O ( 2 N + 1 )) G ( N ) ∫ f ( e i θ k ) d θ k [ 0 , 2 π ] N | ∆ G ( N ) ( e i θ ) | 2 ∏ 2 π = N ! k = 1

  57. MINORS OF TOEPLITZ+HANKEL MATRICES . . . . . . . . . . . . . . . . . . N 1 det ∫ f ( M ) dM = ( G ( N ) = Sp ( 2 N ) , O ( 2 N ) , O ( 2 N + 1 )) G ( N ) ∫ f ( e i θ k ) d θ k [ 0 , 2 π ] N | ∆ G ( N ) ( e i θ ) | 2 ∏ 2 π = N ! k = 1   d 0 − d 2 d 1 − d 3 d 2 − d 4 d 3 − d 5 d 4 − d 6 d 5 − d 7 · · · d 1 − d 3 d 0 − d 4 d 1 − d 5 d 2 − d 6 d 3 − d 7 d 4 − d 8 · · ·       d 2 − d 4 d 1 − d 5 d 0 − d 6 d 1 − d 7 d 2 − d 8 d 3 − d 9 · · ·     d 3 − d 5 d 2 − d 6 d 1 − d 7 d 0 − d 8 d 1 − d 9 d 2 − d 10 · · ·   N × N   d 4 − d 6 d 3 − d 7 d 2 − d 8 d 1 − d 9 d 0 − d 10 d 1 − d 11 · · ·      

  58. MINORS OF TOEPLITZ+HANKEL MATRICES d 2 d 8 d 0 d 6 d 2 d 9 d 3 d 8 d 7 d 7 d 1 d 6 d 0 d 1 d 4 d 2 d 6 d 2 d 3 d 1 d 0 . . . . . . . . . . d 9 . . . . . . . . d 4 d 5 N d 3 1 det d 0 d 2 d 3 d 2 d 4 d 1 d 5 d 4 d 5 d 7 d 6 ∫ χ λ G ( N ) ( M − 1 ) χ µ G ( N ) ( M ) f ( M ) dM = ( G ( N ) = Sp ( 2 N ) , O ( 2 N ) , O ( 2 N + 1 )) G ( N ) ∫ f ( e i θ k ) d θ k [ 0 , 2 π ] N χ λ G ( N ) ( e − i θ ) χ µ G ( N ) ( e i θ ) | ∆ G ( N ) ( e i θ ) | 2 ∏ 2 π = N ! k = 1   d 1 − d 3 d 1 − d 5 d 3 − d 7 d 4 − d 8 · · ·           d 3 − d 5 d 1 − d 7 d 1 − d 9 d 2 − d 10 · · ·   N × N   d 4 − d 6 d 2 − d 8 d 0 − d 10 d 1 − d 11 · · ·      

  59. SOME RESULTS AND APPLICATIONS · Factorizations · Chern-Simons theory N 2 N · Expansions in terms of Toeplitz minors Partition function ∫ ∫ ∫ f ( U ) dU = f ( U ) dU f ( − U ) dU , U ( 2 N ) O ( 2 N + 1 ) O ( 2 N + 1 ) ∫ ∫ ∫ f ( U ) dU = f ( U ) dU f ( U ) dU . U ( 2 N + 1 ) Sp ( 2 N ) O ( 2 N + 2 ) ∑ ( − 1 ) ( | λ | + | µ | ) / 2 D λ,µ det ( d j − k − d j + k ) N j , k = 1 = 1 ( f ) . λ,µ ∈ R ( N ) ∫ Θ( U ) dU G ( N )

  60. SOME RESULTS AND APPLICATIONS · Factorizations · Chern-Simons theory N 2 N · Expansions in terms of Toeplitz minors Wilson loop ∫ ∫ ∫ f ( U ) dU = f ( U ) dU f ( − U ) dU , U ( 2 N ) O ( 2 N + 1 ) O ( 2 N + 1 ) ∫ ∫ ∫ f ( U ) dU = f ( U ) dU f ( U ) dU . U ( 2 N + 1 ) Sp ( 2 N ) O ( 2 N + 2 ) ∑ ( − 1 ) ( | λ | + | µ | ) / 2 D λ,µ det ( d j − k − d j + k ) N j , k = 1 = 1 ( f ) . λ,µ ∈ R ( N ) ∫ χ µ G ( N ) ( U )Θ( U ) dU G ( N )

  61. SOME RESULTS AND APPLICATIONS · Factorizations · Chern-Simons theory N 2 N · Expansions in terms of Toeplitz minors Hopf link ∫ ∫ ∫ f ( U ) dU = f ( U ) dU f ( − U ) dU , U ( 2 N ) O ( 2 N + 1 ) O ( 2 N + 1 ) ∫ ∫ ∫ f ( U ) dU = f ( U ) dU f ( U ) dU . U ( 2 N + 1 ) Sp ( 2 N ) O ( 2 N + 2 ) ∑ ( − 1 ) ( | λ | + | µ | ) / 2 D λ,µ det ( d j − k − d j + k ) N j , k = 1 = 1 ( f ) . λ,µ ∈ R ( N ) ∫ χ λ G ( N ) ( U − 1 ) χ µ G ( N ) ( U )Θ( U ) dU G ( N )

  62. Thank you!

  63. Semigroups for One-Dimensional Schr¨ odinger Operators with Multiplicative White Noise 1 Pierre Yves Gaudreau Lamarre Princeton University 1 Based on a paper of the same name; arXiv:1902.05047. Pierre Yves Gaudreau Lamarre Semigroups for 1D Operators with Noise

  64. Let ξ be a Gaussian white noise on R d , and V : R d → R be a deterministic function. Pierre Yves Gaudreau Lamarre Semigroups for 1D Operators with Noise

  65. Let ξ be a Gaussian white noise on R d , and V : R d → R be a deterministic function. Consider the random Schr¨ odinger operator ˆ � − 1 � H := 2 ∆ + V + ξ. Pierre Yves Gaudreau Lamarre Semigroups for 1D Operators with Noise

  66. Let ξ be a Gaussian white noise on R d , and V : R d → R be a deterministic function. Consider the random Schr¨ odinger operator ˆ � − 1 � H := 2 ∆ + V + ξ. Problem. Develop a semigroup theory for ˆ H , i.e., � H : t > 0 � e − t ˆ Pierre Yves Gaudreau Lamarre Semigroups for 1D Operators with Noise

  67. 1 SPDEs: u ( t, x ) := e − t ˆ H u 0 ( x ) solves � 1 � ∂ t u = 2 ∆ − V u + ξu, u (0 , x ) = u 0 ( x ) . Pierre Yves Gaudreau Lamarre Semigroups for 1D Operators with Noise

  68. 1 SPDEs: u ( t, x ) := e − t ˆ H u 0 ( x ) solves � 1 � ∂ t u = 2 ∆ − V u + ξu, u (0 , x ) = u 0 ( x ) . 2 Spectral Analysis of SPDEs: ∞ e − t ˆ e − tλ k ( ˆ H ) � ψ k ( ˆ H ) , u 0 � ψ k ( ˆ � H u 0 = H ) . k =1 Pierre Yves Gaudreau Lamarre Semigroups for 1D Operators with Noise

  69. 1 SPDEs: u ( t, x ) := e − t ˆ H u 0 ( x ) solves � 1 � ∂ t u = 2 ∆ − V u + ξu, u (0 , x ) = u 0 ( x ) . 2 Spectral Analysis of SPDEs: ∞ e − t ˆ e − tλ k ( ˆ H ) � ψ k ( ˆ H ) , u 0 � ψ k ( ˆ � H u 0 = H ) . k =1 3 Feynman-Kac formula: e − t ˆ H f ( x ) � t � � � �� = E x � � � � � exp − V B ( s ) + ξ B ( s ) d s f B ( t ) . 0 Pierre Yves Gaudreau Lamarre Semigroups for 1D Operators with Noise

  70. ξ cannot be defined pointwise. Pierre Yves Gaudreau Lamarre Semigroups for 1D Operators with Noise

  71. ξ cannot be defined pointwise. It is thus nontrivial to define ˆ − 1 � � Hf = 2 ∆ + V f + ξf ; Pierre Yves Gaudreau Lamarre Semigroups for 1D Operators with Noise

  72. ξ cannot be defined pointwise. It is thus nontrivial to define ˆ − 1 � � Hf = 2 ∆ + V f + ξf ; � t � � � �� E x � � � � � exp − V B ( s ) + ξ B ( s ) d s f B ( t ) 0 Pierre Yves Gaudreau Lamarre Semigroups for 1D Operators with Noise

  73. At my poster: Pierre Yves Gaudreau Lamarre Semigroups for 1D Operators with Noise

  74. At my poster: 1 Show how these technical obstacles can be overcome in one dimension. Pierre Yves Gaudreau Lamarre Semigroups for 1D Operators with Noise

  75. At my poster: 1 Show how these technical obstacles can be overcome in one dimension. 2 Discuss applications in random matrix theory and SPDEs with multiplicative white noise. Pierre Yves Gaudreau Lamarre Semigroups for 1D Operators with Noise

  76. At my poster: 1 Show how these technical obstacles can be overcome in one dimension. 2 Discuss applications in random matrix theory and SPDEs with multiplicative white noise. 3 Discuss partial results in higher dimensions, and connection to regularity structures/paracontrolled calculus/renormalization of SPDEs. Pierre Yves Gaudreau Lamarre Semigroups for 1D Operators with Noise

  77. The longest increasing subsequence problem for correlated random variables J. Ricardo G. Mendonça LPTMS, CNRS, Université Paris-Sud, Université Paris Saclay, 91405 Orsay, France Escola de Artes, Ciências e Humanidades, Universidade de São Paulo, SP, Brasil L’Intégrabilité et l’Aléatoire en Physique Mathématique et en Géométrie CIRM, Marseille Luminy, France, 8–12 April 2019

Recommend


More recommend