Resurgence, Trans-series and Non-perturbative Physics

Transcription

1 Resurgence, Trans-series and Non-perturbative Physics Gerald Dunne University of Connecticut UK Theory Meeting, December 16, 2014 GD & M. Ünsal, , , , GD, lectures at CERN 2014 Winter School also with: G. Başar, A. Cherman, D. Dorigoni, R. Dabrowski: , , , , 1412.xxxx

2 Physical Motivation infrared renormalon puzzle in asymptotically free QFT (i) IR renormalons perturbation theory ill-defined (ii) IĪ interactions instanton-gas ill-defined non-perturbative physics without instantons: physical meaning of non-bps saddle configurations Bigger Picture non-perturbative definition of nontrivial QFT in continuum analytic continuation of path integrals dynamical and non-equilibrium physics from path integrals exact asymptotics in QFT and string theory

3 Mathematical Motivation Resurgence: new idea in mathematics (Écalle, 1980; Stokes, 1850) explore implications for physics resurgence = unification of perturbation theory and non-perturbative physics perturbation theory generally divergent series series expansion trans-series expansion trans-series well-defined under analytic continuation perturbative and non-perturbative physics entwined applications: ODEs, PDEs, fluids, QM, Matrix Models, QFT, String Theory,... philosophical shift: view semiclassical expansions as potentially exact

4 Resurgent Trans-Series trans-series expansion in QM and QFT applications: f(g 2 ) = k 1 c k,l,p g 2p }{{} p=0 k=0 l=1 perturbative fluctuations J. Écalle (1980): set of functions closed under: ( [ exp c ]) k ( g }{{ 2 ln [± 1g ]) l }}{{ 2 } k instantons quasi-zero-modes (Borel transform) + (analytic continuation) + (Laplace transform) trans-monomial elements: g 2, e 1 g 2, ln(g 2 ), are familiar multi-instanton calculus in QFT new: trans-series coefficients c k,l,p highly correlated new: analytic continuation under control new: exponentially improved asymptotics

5 Trans-series No function has yet presented itself in analysis the laws of whose increase, in so far as they can be stated at all, cannot be stated, so to say, in logarithmico-exponential terms G. H. Hardy, Divergent Series, 1949

6 Resurgence resurgent functions display at each of their singular points a behaviour closely related to their behaviour at the origin. Loosely speaking, these functions resurrect, or surge up - in a slightly different guise, as it were - at their singularities J. Écalle, 1980 n m

7 recap: rough basics of Borel summation (i) divergent, alternating: n=0 ( 1)n n! g 2n = 0 dt e t 1 1+g 2 t (ii) divergent, non-alternating: n=0 n! g2n = 0 dt e t 1 1 g 2 t

8 recap: rough basics of Borel summation (i) divergent, alternating: n=0 ( 1)n n! g 2n = 0 dt e t 1 1+g 2 t (ii) divergent, non-alternating: n=0 n! g2n = 0 dt e t 1 1 g 2 t ambiguous imaginary non-pert. term: ± i π g 2 e 1/g2 avoid singularities on R + : lateral Borel sums: C+ C- θ = 0 ± non-perturbative ambiguity: ±Im[Bf(g 2 )] challenge: use physical input to resolve ambiguity

9 Borel summation in practice (physical applications) direct quantitative correspondence between: rate of growth Borel poles non-perturbative exponent non-alternating factorial growth: c n b n n! positive Borel singularity: t c = 1 b g 2 non-perturbative exponent: ±i [ ( )] π 1 b g 2 exp b g 2

10 Analogue of IR Renormalon Problem in QM degenerate vacua: double-well, Sine-Gordon,... splitting of levels: a real one-instanton effect: E e S g 2

11 Analogue of IR Renormalon Problem in QM degenerate vacua: double-well, Sine-Gordon,... splitting of levels: a real one-instanton effect: E e S g 2 surprise: pert. theory non-borel-summable: c n n! (2S) n stable systems ambiguous imaginary part ±i e 2S g 2, a 2-instanton effect

12 Bogomolny/Zinn-Justin mechanism Bogomolny 1980; Zinn-Justin degenerate vacua: double-well, Sine-Gordon, perturbation theory non-borel-summable: ill-defined/incomplete 2. instanton gas picture ill-defined/incomplete: I and Ī attract regularize both by analytic continuation of coupling ambiguous, imaginary non-perturbative terms cancel! resurgence cancellation to all orders

13 Towards Resurgence in QFT resurgence analytic continuation of trans-series effective actions, partition functions,..., have natural integral representations with resurgent asymptotic expansions analytic continuation of external parameters: temperature, chemical potential, external fields,... e.g., magnetic electric; de Sitter anti de Sitter,... matrix models, large N, strings (Mariño, Schiappa, Aniceto,...) soluble QFT: Chern-Simons, ABJM, matrix integrals asymptotically free QFT?

14 QFT: Renormalons QM: divergence of perturbation theory due to factorial growth of number of Feynman diagrams QFT: new physical effects occur, due to running of couplings with momentum faster source of divergence: renormalons both positive and negative Borel poles

15 IR Renormalon Puzzle in Asymptotically Free QFT perturbation theory: ± i e 2S β 0 g 2 instantons on R 2 or R 4 : ± i e 2S g 2 instanton/anti-instanton poles UV renormalon poles IR renormalon poles appears that BZJ cancellation cannot occur asymptotically free theories remain inconsistent t Hooft, 1980; David, 1981

16 IR Renormalon Puzzle in Asymptotically Free QFT resolution: there is another problem with the non-perturbative instanton gas analysis (Argyres, Ünsal ; GD, Ünsal, ) scale modulus of instantons spatial compactification and principle of continuity 2 dim. CP N 1 model: instanton/anti-instanton poles UV renormalon poles IR renormalon poles neutral bion poles cancellation occurs! (GD, Ünsal, , ) semiclassical realization of IR renormalons

17 Topological Molecules in Spatially Compactified Theories CP N 1 : regulate scale modulus problem with (spatial) compactification R 2 SL 1 x R 1 x 2 x 2 x 1 x 1 Euclidean time instantons fractionalize Kraan/van Baal; Lee/Yi; Bruckmann; Brendel et al

18 Topological Molecules in Spatially Compactified Theories temporal compactification: information about deconfined phase R 1 x Sᵦ1 R 1 high T low T R 2 spatial compactification: semi-classical small L regime continuously connected to large L: principle of continuity R 1 1 SL x R 1 continuity R 2 SUSY: Seiberg-Witten 1996, Davies-Hollowood-Khoze-Mattis 1999,... non-susy: Kovtun-Ünsal-Yaffe, 2007; Ünsal, 2009,...

19 CP N 1 Model 2d sigma model analogue of 4d Yang-Mills asymptotically free: β 0 = N (independent of N f ) instantons, theta vacua, fermion zero modes,... divergent perturbation theory (non-borel summable) renormalons (both UV and IR) large-n analysis non-perturbative mass gap: m g = µ e 4π/(g2 N) couple to fermions, SUSY,... analogue of center symmetry

20 Fractionalized Instantons in CP N 1 on S 1 R 1 Z N twisted instantons fractionalize Bruckmann, 2007; Brendel et al, 2009 spatial compactification Z N twist: ( ) 1 v twisted = ( λ 1 + λ 2 e 2π z) L e 2π L µ 2 z (twist in x 2 )+(holomorphicity) fractionalization along x 1 S inst S inst N = S inst β 0

21 Fractionalized Bions in CP N 1 GD/Ünsal, bions: topological molecules of instantons/anti-instantons orientation dependence of IĪ interaction: charged bions: repulsive bosonic interaction B ij = [I i Ī j ] e S i(ϕ) S j (ϕ) e iσ(α i α j ) neutral bions: attractive bosonic interaction B ii = [I i Ī i ] e 2S i(ϕ) instanton/anti-instanton amplitude is ambiguous: [ Ii Ī i ] ± = (ln ( g 2 ) ) N 16 γ 8π g 2 N e 8π g 2 N ± iπ 16 g 2 N e 8π g 2 N

22 Perturbation Theory in Spatially Compactified CP N 1 small radius limit effective QM Hamiltonian H zero = g2 2 P θ 2 + ξ2 sin 2 θ + g2 2g 2 2 sin 2 θ P φ 2, ξ = 2π N Born-Oppenheimer approximation: drop high φ-sector modes effective Mathieu equation: 1 2 ψ + ξ2 2g 2 sin2 (gθ)ψ = E ψ Stone-Reeve (Bender-Wu methods): E(g 2 ) = a n (g 2 ) n, n=0 non-borel-summable! a n 2 π ( ) N n ( n! 1 5 ) 8π 2n +...

23 BZJ cancellation in Spatially Compactified CP N 1 ( ) perturbative sector: lateral Borel summation B ± E(g 2 ) = 1 g 2 dt BE(t) e t/g2 = Re BE(g 2 ) iπ 16 C ± g 2 N e non-perturbative sector: bion-bion amplitudes ( [ ] g Ii Ī i (ln 2 ) ) ± = N 16 8π γ 8π g 2 N e g 2 N ± iπ 16 g 2 N e exact cancellation! explicit application of resurgence to nontrivial QFT 8π g 2 N 8π g 2 N

24 Graded Resurgence Triangle saddle points labelled by instanton/anti-instanton number: [0] [I] [Ī] [I 2 ] [I Ī] [Ī2 ] [I 3 ] [I 2 Ī] [I Ī2 ] [Ī3 ] [I 4 ] [I 3 Ī] [I 2 Ī 2 ] [I Ī3 ] [Ī4 ] cancellations can only occur within columns

25 Graded Resurgence Triangle and Extended SUSY extended SUSY: no superpotential; no bions, no condensates [0] [I] [Ī] [I 2 ] 0 [Ī2 ] [I 3 ] 0 0 [Ī3 ] [I 4 ] [Ī4 ] perturbative expansions must be Borel summable! GD, Shifman, Ünsal,...

26 The Bigger Picture Q: should we expect resurgent behavior in QM and QFT? QM uniform WKB (i) trans-series structure is generic (ii) all multi-instanton effects encoded in perturbation theory (GD, Ünsal, , ) Q: what is behind this resurgent structure? basic property of all-orders steepest descents integrals Q: could this extend to (path) functional integrals?

27 Uniform WKB and Resurgent Trans-Series for Eigenvalues (GD, Ünsal, , ) d2 dx g 2 2 ψ + V (g x) ψ = E ψ g 4 d2 dy 2 ψ(y) + V (y)ψ(y) = g2 E ψ(y) weak coupling: degenerate harmonic classical vacua ( ) non-perturbative effects: g 2 exp c g 2 approximately harmonic uniform WKB with parabolic cylinder functions

28 Uniform WKB and Resurgent Trans-Series for Eigenvalues ( ) 1 uniform WKB ansatz (parameter ν): ψ(y) = Dν g u(y) u (y) perturbative expansion for E and u(y): E = E(ν, g 2 ) = g 2k E k (ν) k=0 ν = N: usual perturbation theory (not Borel summable) global analysis boundary conditions: y Π 2 Π Π 2 Π 2 Π 3 Π 2 midpoint 1 g ; non-borel summability g2 e ±i ɛ g 2

29 Uniform WKB and Resurgent Trans-Series for Eigenvalues 2π D ν (z) z ν e z2 /4 ( ) + e ±iπν Γ( ν) z 1 ν e z2 /4 ( ) exact quantization condition 1 Γ( ν) ( e ±iπ 2 g 2 ) ν = e S/g2 π g 2 F(ν, g2 ) ν is only exponentially close to N (here ξ e S/g2 ): π g 2 ν = N + ( ) 2N 2 g 2 (N!) 2 ( ) N 2 g F(N, g 2 ) 2 N! [ F F ( N + ln ξ ( e ±iπ 2 g 2 ) ) ] ψ(n + 1) F 2 ξ 2 + O(ξ 3 ) insert: E = E(ν, g 2 ) = k=0 g2k E k (ν) trans-series

30 Connecting Perturbative and Non-Perturbative Sector trans-series form for energy eigenvalues arises from the (resurgent) analytic continuation properties of the parabolic cylinder functions generic and universal Zinn-Justin/Jentschura: generate entire trans-series from (i) perturbative expansion E = E(N, g 2 ) (ii) single-instanton fluctuation function F(N, g 2 ) (iii) rule connecting neighbouring vacua (parity, Bloch,...)

31 Connecting Perturbative and Non-Perturbative Sector trans-series form for energy eigenvalues arises from the (resurgent) analytic continuation properties of the parabolic cylinder functions generic and universal Zinn-Justin/Jentschura: generate entire trans-series from (i) perturbative expansion E = E(N, g 2 ) (ii) single-instanton fluctuation function F(N, g 2 ) (iii) rule connecting neighbouring vacua (parity, Bloch,...) in fact... (GD, Ünsal, , ) [ g 2 F(N, g 2 dg 2 ) = exp S g 4 0 ( E(N, g 2 ) 1 + N ( N S ) )] g 2 implication: perturbation theory encodes everything!

32 Connecting Perturbative and Non-Perturbative Sector e.g. double-well potential: B N ( E(N, g 2 ) = B g 2 3B ) ( g 4 4 ( 375 g 6 2 B B B non-perturbative function (F (...) exp[ A/2]): A(N, g 2 ) = 1 ( 3g 2 + g2 17B simple relation: )... ) 4 B ) ( + g 4 125B B 4 ( g 6 12 B B ( E B = 3g2 2B g 2 A ) g 2 ) + )

33 Connecting Perturbative and Non-Perturbative Sector all orders of multi-instanton trans-series are encoded in perturbation theory of fluctuations about perturbative vacuum n m why? turn to path integrals... (also: QFT)

34 All-Orders Steepest Descents: Darboux Theorem all-orders steepest descents for contour integrals: hyperasymptotics (Berry/Howls 1991, Howls 1992) I (n) (g 2 ) = dz e C n 1 g 2 f(z) = 1 1/g 2 e 1 g 2 fn T (n) (g 2 ) T (n) (g 2 ): beyond the usual Gaussian approximation asymptotic expansion of fluctuations about the saddle n: T (n) (g 2 ) r=0 T r (n) g 2r

35 All-Orders Steepest Descents: Darboux Theorem universal resurgent relation between different saddles: T (n) (g 2 ) = 1 2π i ( 1) γnm m 0 dv v e v 1 g 2 v/(f nm ) T (m) ( Fnm v ) exact resurgent relation between fluctuations about n th saddle and about neighboring saddles m [ T r (n) (r 1)! ( 1) γnm = 2π i (F m nm ) r T (m) 0 + F nm (r 1) T (m) (F nm ) (r 1)(r 2) T (m) universal factorial divergence of fluctuations (Darboux) fluctuations about different saddles explicitly related!

36 All-Orders Steepest Descents: Darboux Theorem d = 0 partition function for periodic potential V (z) = sin 2 (z) I(g 2 ) = π 0 dz e 1 g 2 two saddle points: z 0 = 0 and z 1 = π 2. sin2 (z) vacuum min. saddle IĪ min. vacuum

37 All-Orders Steepest Descents: Darboux Theorem T (0) large order behavior about saddle z 0 : ) 2 r = Γ ( r π Γ(r + 1) (r 1)! π ( (r 1) + 32 (r 1)(r 2) 128 (r 1)(r 2)(r 3) +

38 All-Orders Steepest Descents: Darboux Theorem T (0) large order behavior about saddle z 0 : ) 2 r = Γ ( r π Γ(r + 1) (r 1)! π ( (r 1) + 32 (r 1)(r 2) 128 (r 1)(r 2)(r 3) + low order coefficients about saddle z 1 : T (1) (g 2 ) i ( π g g4 75 ) 128 g fluctuations about the two saddles are explicitly related

39 Resurgence in Path Integrals: Functional Darboux Theorem could something like this work for path integrals? functional Darboux theorem? multi-dimensional case is already non-trivial and interesting Pham (1965); Delabaere/Howls (2002) Picard-Lefschetz theory do a computation to see what happens...

40 Resurgence in Path Integrals periodic potential: V (x) = 1 g 2 sin 2 (g x) vacuum saddle point ( c n n! n 13 ) 8 1 n(n 1)... instanton/anti-instanton saddle point: Im E π e 2 1 2g ( g2 13 ) 8 g4...

41 Resurgence in Path Integrals periodic potential: V (x) = 1 g 2 sin 2 (g x) vacuum saddle point ( c n n! n 13 ) 8 1 n(n 1)... instanton/anti-instanton saddle point: Im E π e 2 1 2g ( g2 13 ) 8 g4... double-well potential: V (x) = x 2 (1 gx) 2 vacuum saddle point ( c n 3 n n! n ) n(n 1)... instanton/anti-instanton saddle point: ( Im E π e 2 1 6g g ) 72 g4...

42 Resurgence in Path Integrals fluctuations about the instanton/anti-instanton saddle are determined by those about the vacuum saddle

43 Analytic Continuation of Path Integrals: Lefschetz Thimbles DA e 1 g 2 S[A] = N k e i g 2 S imag[a k ] DA e 1 g 2 S real[a] Γ k thimbles k Lefschetz thimble = functional steepest descents contour remaining path integral has real measure: (i) Monte Carlo (ii) semiclassical expansion (iii) exact resurgent analysis

44 Analytic Continuation of Path Integrals: Lefschetz Thimbles DA e 1 g 2 S[A] = N k e i g 2 S imag[a k ] DA e 1 g 2 S real[a] Γ k thimbles k Lefschetz thimble = functional steepest descents contour remaining path integral has real measure: (i) Monte Carlo (ii) semiclassical expansion (iii) exact resurgent analysis resurgence: asymptotic expansions about different saddles are closely related requires a deeper understanding of complex configurations and analytic continuation of path integrals... Stokes phenomenon: intersection numbers N k can change with phase of parameters

45 Analytic Continuation of Path Integrals brute-force approach: x 1 x 2 x 3 xn 2 xn 1 x N x 0 t 0 t 1 t 2 t 3 tn 2 tn 1 t N

46 Analytic Continuation of Path Integrals brute-force approach: x 1 x 2 x 3 xn 2 xn 1 x N x 0 t 0 t 1 t 2 t 3 tn 2 tn 1 t N

47 Analytic Continuation of Path Integrals brute-force approach: x 1 x 2 x 3 xn 2 xn 1 x N x 0 t 0 t 1 t 2 t 3 tn 2 tn 1 t N

48 Analytic Continuation of Path Integrals brute-force approach: x 1 x 2 x 3 xn 2 xn 1 x N x 0 t 0 t 1 t 2 t 3 tn 2 tn 1 t N

49 Thimbles from Gradient Flow gradient flow to generate steepest descent thimble: τ A(x; τ) = δs δa(x; τ) keeps Im[S] constant, and Re[S] is monotonic τ ( ) S S = 1 2i 2i τ ( δs A δa τ δs δa ( ) S + S δs = 2 δa Chern-Simons theory (Witten 2001) comparison with complex Langevin (Aarts 2013,...) 2 ) A = 0 τ lattice (Tokyo/RIKEN, Aurora, 2013): Bose-gas

50 Thimbles, Gradient Flow and Resurgence uartic model [ Z = dx exp bles GA 13 ( σ 2 x2 + x4 4 )] (Aarts, 2013; GD, Unsal,...) 2 1 σ = 1+i, λ = stable thimble unstable thimble y 0 stable thimble unstable thimble not contributing y x 1 2 λ contributing thimbles change with phase of σ ble contributes need all three thimbles for Re[σ] < 0 mble gives correct result, with Jacobian integrals along thimbles are related (resurgence) SIGN 2014 p. 6 resurgence: preferred unique field choice x

51 Ghost Instantons: Analytic Continuation of Path Integrals (Başar, GD, Ünsal, arxiv: ) periodic elliptic potential V (x) = 1 g 2 sd2 (g x m) large order growth of perturbation theory (Bender/Wu, Bogomolny, Zinn-Justin, Lipatov,...) a n (m) 16 π n! 1 (S I Ī (m))n n naive ratio d 1

52 Ghost Instantons: Analytic Continuation of Path Integrals (Başar, GD, Ünsal, arxiv: ) periodic elliptic potential V (x) = 1 g 2 sd2 (g x m) large order growth of perturbation theory (Bender/Wu, Bogomolny, Zinn-Justin, Lipatov,...) a n (m) 16 π n! 1 (S I Ī (m))n n naive ratio d 1 fails miserably!

53 Ghost Instantons: Analytic Continuation of Path Integrals Z(g 2 m) = Z Dx e S[x] = Z Dx e R dτ 1 2 x g sd2 (g x m) doubly periodic potential: real & complex instantons 10 instanton actions: arcsin( m) p SI (m) = m(1 m) 2 arcsin( 1 m) p SG (m) = m(1 m)

54 Ghost Instantons: Analytic Continuation of Path Integrals large order growth of perturbation theory: a n (m) 16 ( ) π n! 1 ( 1)n+1 (S I Ī (m))n+1 S G Ḡ (m) n+1 naive ratio d n without ghost instantons ratio d n with ghost instantons complex instantons directly affect perturbation theory, even though they are not in the original path integral measure

55 The Stokes Phenomenon and the Airy Function supernumerary rainbows 1 e i( 1 3 t3 +x t) dt 2π 2, x + π x 1/4 e 2 3 x3/2 sin( 2 3 ( x)3/2 + π 4 ) π ( x) 1/4, x

56 Non-perturbative Physics Without Instantons Dabrowski, GD, arxiv: , Cherman, Dorigoni, GD, Ünsal, Yang-Mills, CP N 1, O(N), PCM,... all have non-bps solutions with finite action (Din & Zakrzewski, 1980; Uhlenbeck 1985; Sibner, Sibner, Uhlenbeck, 1989) unstable : negative modes of fluctuation operator what do these mean physically? resurgence: ambiguous imaginary non-perturbative terms should cancel ambiguous imaginary terms coming from lateral Borel sums of perturbation theory DA e 1 g 2 S[A] = all saddles e 1 g 2 S[A saddle] (fluctuations) (qzm)

57 Non-perturbative Physics Without Instantons: Principal Chiral Model (Cherman, Dorigoni, GD, Ünsal, ) S[U] = 1 2g 2 d 2 x tr µ U µ U, U SU(N) non-borel-summable perturbation theory due to IR renomalons but, the theory has no instantons! resolution: non-bps saddle point solutions to 2nd-order classical Euclidean equations of motion: unitons ) µ (U µ U = 0 (Uhlenbeck 1985) have negative fluctuation modes: saddles, not minima fractionalize on cylinder BZJ cancellation

58 Resurgence and Localization (Mariño, ; Kallen, Mariño, ; Aniceto, Russo, Schiappa, ) certain protected quantities in especially symmetric QFTs can be reduced to matrix models resurgent asymptotics 3d Chern-Simons on S 3 matrix model 1 Z CS (N, g) = dm exp [ 1g ( )] 12 vol(u(n)) tr (ln M)2 ABJM: N = 6 SUSY CS, G = U(N) k U(N) k Z ABJM (N, k) = ( 1)ɛ(σ) N dx i 1 N! 2πk N σ S N i=1 i=1 2ch ( x i ) ( ) xi x 2 ch σ(i) 2k N = 4 SUSY Yang-Mills on S 4 Z SY M (N, g 2 ) = 1 vol(u(n)) dm exp [ 1g ] 2 trm 2

59 Conclusions Resurgence systematically unifies perturbative and non-perturbative analysis trans-series encode all information; expansions about different saddles are intimately related there is extra magic in perturbation theory matrix models, large N, strings, SUSY QFT IR renormalon puzzle in asymptotically free QFT multi-instanton physics from perturbation theory N = 2 and N = 2 SUSY gauge theory (Basar, GD, 1412.xxxx) fundamental property of steepest descents moral: go complex and consider all saddles, not just minima