1. BESIII March 2016, with: D. Weissman 1402.5603, 1403.0763, 1504.xxxx,

2. Introduction The stringy description of hadrons has been thoroughly investigated since the seventies. What are the reasons to go back to “square one" and revisit this question? (i) Holography, or gauge/string duality, provides a bridge between the underlying theory of QCD (in certain limits) and a bosonic string model of mesons and baryons. (ii) There is a wide range of heavy mesonic and baryonic resonances that have been discovered in recent years, and (iii) up to date we lack a full exact procedure of quantizing a rotating string with massive endpoints.

3. Introduction The holographic duality is an equivalence between certain bulk string theories and boundary field theories. Practically most of the applications of holography are based on relating bulk fields ( not strings) and operators on the dual boundary field theory. This is based on the usual limit of  ‘ 0 with which we go for instance from a closed string theory to a gravity theory. However, to describe hadrons in reality it seems that we need strings since after all in reality the string tension is not very large ( of order one)

4. Introduction The main theme of this talk is that certain hadronic physical observables like spectra and decays cannot be faithfully described by bulk fields but rather require dual stringy phenomena It is well known that this is the case for Wilson, ‘t Hooft and Polyakov lines and also Entanglement entropy We argue here that in fact also the spectra, decays and other properties of hadrons: mesons, baryons and glueballs can be recast only by holographic stringy hadrons

5. Introduction The major argument against describing the hadron spectra in terms of fluctuations of fields like bulk fields or modes on probe branes is that they generically do not admit properly the Regge behavior of the spectra. For as a function of J we get from flavor branes only J=0,J=1 mesons and there will be a big gap of order in comparison to high J mesons if we describe the latter in terms of strings. The attempts to get the linearity between and n basically face problems whereas for strings it is an obvious property.

6. Outline: A brief review of the HISH models Mesons as open strings with massive endpoints Baryons as open strings connecting a quark di-quark Glueballs as closed strings The phenomenology Fits to mesonic data Fits to baryonic data Decays of hadrons Predictions about excited mesons Predictions about excited baryons Identifying glueballs Exotic hadrons

7. From AdS/Cft Duality to Holography of Confining backgrounds

13. From Ads/CFT to general string(gravity)/gauge duality The basic duality relates the string theory on Ads5xS5 to N=4 SYM. Both sides are invariant under the maximal super-conformal symmetries. To get to N=0 YM theory we need to break all the supersymmetries. The need to introduce a scale in the gauge theory translates to deforming the bulk into a non-Ads one. What are bulk geometries that correspond to a confining gauge theory? Confining means admitting an area law Wilson line

14. Sufficient conditions for confinement We proved a theorem that sufficient conditions for a background to admit a confining Wilson line are if either (Y.kinar, E. Schreiber J.S) (i) (f(u))^2 = G 00 G xx (u) has a minimum at u min and f(u min )>0 (i) (g(u))^2= G 00 G uu (u) diverges at u div and f(udiv)>0

15. Confining backgorunds There are handful of backgrounds that admit confining Wilson lines. There are bottom-up scenarios like the hard and soft wall Here we mention top-down models Most of the analysis here is model independent. A prototype model of the pure gauge sector is: The compactified D4 brane background: (i) The critical ( 10d) model (Witten) (ii) The non-critical (6d) model. (S. Kuperstein J.S)

16. Compactified D4 model ( Witten’s model) D4

17. The gauge theory and sugra parametrs are related via 5d coupling 4d coupling glueball mass String tension The gravity picture is valid only provided that  >> R The theory is 5d. At energies E<< 1/R the theory is effectively 4d. However it is not really QCD since M gb ~ M KK In the opposite limit of   R we approach QCD

18. Adding flavor We would like to introduce flavor degrees of freedom We add N f flavor branes to a non-supersymmetric confining background. A natural candidate is therefore Witten’s model. For N f << N c the flavor brane do not back-react on the background thus they are probe branes What type of Dp branes should we add D4, D6 or D8 branes? How do we incorporate a full chiral flavor global symmetry of the form U(Nf)xU(Nf), with left and right handed chiral quarks?

19. Adding flavor We place the two endpoints of the probe branes on the compactified circle. If there are additional transverse directions to the probe branes then one can move them along those directions and by that the strings will acquire length and the corresponding fields mass. Thus this will contradict the chiral symmetry which prevents a mass term. Thus we are forced to use D8 branes that do not have additional transverse directions. The fact that the strings are indeed chiral follows also from analyzing the representation of the strings under the Lorentz group

20. Adding flavor U(N f ) x U(N f ) global flavor symmetry in the UV calls for two separate stacks of branes. To have a breakdown of this chiral symmetry to the diagonal U(N f ) D we need the two stacks of branes to merge one into the other. This requires a U shape profile of the probe branes. The opposite orientations of the probe brane at their two ends implies that in fact these are stacks of N f D8 branes and a stack of N f anti D8 branes. ( Thus there is no net D8 brane charge) This is the Sakai Sugimoto model.

21. Adding flavor: The Sakai Sugimoto model Adding N f D8 anti-D8 branes into Witten’s model Sakai Sugimoto modelGeneralized SS model f

22. Adding flavor: The Sakai Sugimoto model

23. Stringy holographic Mesons

24. Stringy meson in U shape flavor brane setup In the generalized Sakai Sugimoto model or its non-critical partner the meson looks like ufuf

25. Example: The B meson

26. Rotating Strings ending on flavor branes Consider a general background of the form G mm (u) is a function of the radial direction u We look for rotating solutions of the eom We assume that u f >u>u 

27. Strings ending on flavor branes Denote with and The NG action in the  =R gauge than reads The equation of motion for u(R) where

28. Strings ending on flavor branes We now separate the profile into two regions: Region (I) vertical Region (II) horizontal II II Flavor brane wall Stringy meson

29. String end-point mass We define the string end-point quark mass For  S=0 the system has to obey the condition This requires that

30. Condition for a stringy meson The conditions to have a solution read The conditions to have mesons with Regge behavior in the limit of small m sep are precisely the conditions to have a confining Wilson line

31. How close is the |_| string to the real holographic one This is a numerical calculation of the profile for a string with J=3 rotating in Witten’s model background For the static case

32. String spectrum: Energy and Angular momentum The Noether charges associated with the shift of t and  The contribution of the vertical segments

33. Energy and Angular momentum Recall the string end-point mass defined as The horizontal segment contributes Combining together all the segments we get

34. Small and large mass approximations We can get a relation between J and E for Small mass Large mass

35. From holographic string to string with massive endpoints It is now clear that we can approximate the holographic spinning string with a string in flat space time with massive endpoints. The masses are and (M. Kruczenski, L. Pando Zayas D. Vaman )

36. On the quantization of bosonic string with massive particles on its ends

37. The classical action There are two ways to write the bulk string action Polyakov There are two ways to write the endpoints action

38. Possible classical actions Thus there are 4 possible ways for the combined action In fact there is also another Weyl invariant action For (iv) we associate with    or to take it independent

39. The equations of motion The variation of the bulk of the NG action yields At the two boundaries we get In (ii) the boundary equations and equations are

40. The equations of motion In (iii) the bulk equation is The boundary equation is The variation of the metric

41. The solutions of the equations of motion A rotating classical solution in Correspondingly the boundary condition In particular

42. The quantum action Even without the massive endpoints the stringy actions in D space-time dimensions are not conformal invariant Q.M. Polyakov suggested to add the Liuville term For the NG case Polchinski and Strominger took The PS action reads

43. On the quantization ( preliminary) The quantization of the holographic string is a difficult problem Instead we consider the quantization of an open string with massive endpoints. The exact solution for that question in D dimensions is not known There are two obvious limits of : (i) The static case (v=0, ) (ii) The massless case ( v=1, m=0)

44. On the quantization The energy of the quantized static open string with no massive endpoints in the D dimension is (Arvis) A naïve generalization of the static to a rotating string with no massive endpoints Which translates to the Regge relation

45. On the quantization (preliminary) For strings with massive endpoints there are two major differences: (i) The relation between J T and E is more complicated as we have seen above (ii) The eigenfrequencies are not anymore In addition one has to incorporate the PS non-critical term

46. On the quantization (preliminary) In the Polyakov formulation the solutions of the EQN are The eigenfrequencies and phases are given by In the limit of massless and infinite mass we get

47. The Casimir energy The Casimir energy ( or the intercept) is given by For the special cases For finite mass and we cannot use the zeta function regularization

48. The Casimir energy How can we sum over the eigenfrequencies for the massive case? We use a contour integral to compute the sum using (Lambiase Nesterenko) we take zeros poles So the Casimir energy is

49. The Casimir energy Where C is a contour that includes the real semi- axis where all the roots of f(w) occur. Since f(w) does not have poles we deform the contour to a semi-circle of radius  and a segment along the imaginary axis The Casimir energy thus reads To regularize and renormalize the result we subtract

50. The Casimir energy The subtracted energy is The renormalized Casimir energy is thus For the massless and infinite mass cases

51. The Casimir energy Denoting a=m/Tl we definine the ratio

52. The Casimir energy For the rotating string we simply replace For the massless and small mass cases we have

53. The non criticality term: Liuville term The quantum string action is inconsistent for a non- critical D dimensions. In the Polyakov formulation for quantum conformal invariance one has to add a Liouville term. It can be built from a ``composite Liouville field” The action then reads The Liouville term is where

54. The non criticality term: The Polchinsky Strominger term In the Nambu-Goto formulation the anomaly is cancelled by adding a Polchinky Strominger term For a classical rotating string parameterized as The induced metric is For the range of (  ) The boundary condition is The PS term is

55. The non-criticality term for the massless case Inserting the rotating classical string to the Liuville field one finds that The Liouville term= The Polchinsky Strominger term For the massless case  and hence the non- critical term diverges. Hellerman et al suggested a procedure to regularize and renormalize this divergence for the massless case. They found an amazing result that the intercept dose not depend on D The generalization of this result to the massive case is under current investigation

56. Leading 1/m order quantum correction In the limit of large m/TL ( v<<1) the boundary eom The classical trajectory The quantum corrected trajectory involves

57. Leading 1/m order quantum correction Thus the corrected trajectory reads The contribution of Sps to the intercept for D=4 We can replace the dependence on TL with We can approximate the a Cas

58. Fits of Stringy mesons with massive endpoints to experimental data

59. Fitting analysis Now we leave the holographic world and go down to earth to fit the data using the analogous string with massive endpoints. We confront the theoretical massive modified Regge relations with experimental data. It is easier to analyze separately and For we use the following models: (i) The linear original Regge relation

60. Fitting analysis: (2) The modified massive Regge relation

61. The small and large mass limits In the small mass limit m/TL<<1 the trajectory reads The large mass limit

62. Fitting analysis The original linear Regge relation The WKB approximation

63. Extracted parameters The parameters we extract from the fits with the lowest are (i) measured in or the string tension (ii) the intercept ( dimensionless) (iii) (m 1, m 2 ) the string endpoint masses We define in a non-standard but convenient way

64. The meson trajectories fitted For we compared with the following Regge trajectories For the trajectories used

65. The botomonium trajectories To emphasize the deviation from the linearity we start with the botomonium trajectories

66. The charmonium trajectories For the charmonium trajectory we get Now the improvement over the linear is

67. The K* trajectories The K* mesons are constructed from and have S=1 The best fitted tension and intercept are The best fitted masses are

68. The  trajectories For the  mesons trajectories the difference between the original linear and modified trajectories is the smallest For the linear For the massive The ratio is

69. Optimization in the m  ’ plane for  and 

70. Optimization in the m  ’ plane : s quark

71. Toward a universal model The fit results for several trajectories simultaneously. The trajectories of We take the string endpoint masses Only the intercept was allowed to change. We got

72. Toward a universal model

73. Predictions

75. Holography versus massive endpoints toy model In the toy model of string with massive endpoints for vanishing orbital angular momentum J=0 the length of the string vanishes and hence only the quarks at the endpoints constitute the meson mass In holography we get non trivial contribution of the string even with no angular momentum Thus the comparison with data favors holography over the massive endpoints toy model.

76. Decay width of Stringy holographic Mesons

77. The structure of a rotating holographic string

78. A cartoon of possible decays of a (h,m) meson

79. Holographic decay- qualitative picture Quantum mechanically the stringy meson is unstable. Fluctuations of endpoints splitting of the string The string has to split in such a way that the new endpoints are on a flavor brane. The decay probability= (to split at a given point ) X (that the split point is on a flavor brane ) The probability to split of an open string in flat space time was computed by Dai and Polchinski and by Turok et al.

80. The split of an open sting in flat space time Intuitively the string can split at any point and hence we expect width~ L The idea is to use the optical theorem and compute the total rate by computing the imaginary part of the self energy diagram An exercise in one loop string calculation Consider a string stretched around a long compact spatial direction. A winding state splits and joins. In terms of vertex operators it translates to a disk with two closed string vertex operators

81. The corresponding amplitude takes the form where k is the gravitational coupling, g the coefficient of the open string tachyon, the factor L comes from the zero mode. Using the ope’s we get where Performing the integral, taking the imaginary part

82. String bit approximation Using a string bit model the integration over the right subset of configurations becomes easier. K.Peeters M. Zamaklar J.S The discretized string consists of a number of horizontal rigid rods connected by vertical springs.

83. The mass of each bead is M, the length is L=(N+1)a and the action is The normal modes and their frequencies are In the relativistic limit and large N The action now is of N decoupled normal modes The wave function is a product of the wave functions of the normal modes

84. Note that the width of the Gaussian depends on T eff and not on L The integration interval is when the bead is “at the brane” defined by By computing the decay width for various values of N and extrapolating to large N we find that the decay rate is approximated by Casher Neuberger Nussinov

85. The basic CNN model predicts In fact and hence incorporating the corrections due to the massive endpoints we find the following blue curve which fits the data points of the K* mesons a function of M  /M- Correction to the decay width due to msep

86. Stringy holographic Baryons

87. Stringy Baryons in hologrphy How do we identify a baryon in holography ? Since a quark corresponds to a string, the baryon has to be a structure with N c strings connected to it. Witten proposed a baryonic vertex in AdS 5 xS 5 in the form of a wrapped D5 brane over the S 5. On the world volume of the wrapped D5 brane there is a CS term of the form Scs=

88. Baryonic vertex The flux of the five form is c This implies that there is a charge N c for the abelian gauge field. Since in a compact space one cannot have non-balanced charges there must be N c strings attached to it.

89. External baryon External baryon – Nc strings connecting the baryonic vertex and the boundary boundary Wrapped D brane

90. Dynamical baryon Dynamical baryon – Nc strings connecting the baryonic vertex and flavor branes boundary Flavor brane dynami Wrapped D brane

91. Dynamical baryon in the n-c Ads6 model In this model the baryonic vertex is a D0 brane of the non-critical compact D4 brane background. boundary Flavor brane dynami unwrapped D0 brane

92. A possible baryon layout A possible dynamical baryon is with Nc strings connected to the flavor brane and to the BV which is also on the flavor brane. boundary Flavor brane Baryonic vertex

93. Nc-1 quarks around the Baryonic vertex Another possible layout is that of one quark connected with a string to the BV to which the rest of the Nc-1 quarks are attached.

94. From large Nc to three colors Naturally the analog at Nc=3 of the symmetric configuration with a central baryonic vertex is the old Y shape baryon The analog of the asymmetric setup with one quarks on one end and Nc-1 on the other is a straight string with quark and a di-quark on its ends.

95. Stability of an excited baryon Sharov and ‘t Hooft showed that the classical Y shape three string configuration is unstable. An arm that is slightly shortened will eventually shrink to zero size. We have examined Y shape strings with massive endpoints and with a massive baryonic vertex in the middle. G. Harpaz J.S The analysis included numerical simulations of the motions of mesons and Y shape baryons under the influence of symmetric and asymmetric disturbance. We indeed detected the instability We also performed a perturbative analysis where the instability does not show up.

96. Baryonic instability The conclusion from both the simulations and the qualitative analysis is that indeed the Y shape string configuration is unstable to asymmetric deformations. Thus an excited baryon is an unbalanced single string with a quark on one side and a di-quark and the baryonic vertex on the other side.

97. Stringy holographic Baryons versus experimental data

98. Baryons are straigh strings! It is straightforward to realize that the Y shape structure has  Y  ‘  l ‘ A quick glance on the baryon trajectories shows that they admit roughly ( 5%) the same  ‘ as that of the mesons. Thus we conclude that baryons are straight strings and not Y shape strings

99. Excited baryon as a single string Thus we are led to a picture where the baryon is a single string with a quark on one end and a di- quark (+ a baryonic vertex) at the other end. This is in accordance with stability analysis which shows that a small instability in one arm will cause it to shrink so that the final state is a single string

100. Fit to Regge trajectories of Nucleons Fit of the Regge trajectories of the Nucleons

101. Fitting the Nucleon trajectories Notice that there are separate trajectories for even L and for odd L. Assuming that m1=115 Mev the best fit for m2= 57Mev with The fit with m2=240 Mev is much worth

102. The trajectories of  c  c

103. The structure of the stringy nucleon We conclude that the setup is So in the right hand side we have m q and not 2m q There does not seem to be a contribution to the mass from the Baryonic vertex Nucleon string Baryonic vertex quark Di-quark

104. Central baryonic vertex is excluded The fit analysis definitely prefers the previous setup over a one with a central baryonic vertex A fit to such a scenario yields zero mass to the bayonic vertex and fails to see a 2m sep on the rhs

105. From holographic to flat space-time configurations

106.  ’ and 2m for the nucleon trajectory

107. Summary of the baryonic fits The range associates with of 10%

108. Summary of the baryonic fits Fits for the optimal fixed It is harder to construct a unified stringy model for the baryons than for the mesons. The model of a quark and a di-quark is best supported The mesonic and baryonic results are similar the  is an exception it prefers massless s quark

109. Glueballs as closed strings

110. Mesons are open strings with a massive quark and an anti-quark on its ends. Baryons are open strings with a massive quark on one end and a baryonic vertex and a di-quark on the other end. What are glue balls? Since they do not incorporate quarks it is natural to assume that they are rotating closed strings Angular momentum associates with rotation of folded closed strings

111. Closed strings versus open strings The spectrum of states of a closed string admits The spectrum of an open string The slope of the closed string is ½ of the open one The closed string ground states has The intercept is 2

112. Closed strings versus open ones In the terminology of QCD the tension of the string associate with the Quadratic Casimir and hence the ratio This is in accordance with the ratio Slope closed = ½ Slope open

113. Holographic mesons and glueballs and their map

114. Phenomenology A rotating and exciting folded closed string admits in flat space-time a linear Regge trajectory = The basic candidates of glueballs are flavorless hadrons f 0 of 0++ and f 2 of 2++. There are 9 (+3) f0 and 12 (+5) f2. The question is whether one can fit all of them into meson and separately some glueball trajectories. We found various different possibilities of fits.

115. Glueball 0++ fits of experimental data Assignment with f 0 (1380) as the glueball ground-state

116. Glueball 0++ fits of experimental data The meson and glueball trajectories based on f 0 (1380) as a glueball lowest state.

117. On the identification of glueball trajectory Unfortunately there exists no unambiguous way to assign the known flavorless hadrons into trajectories of mesons and glueballs, But it is clear that one cannot sort all the known resonances into meson trajectories alone. One of the main problems in identifying glueball trajectories is simply the lack of experimental data, particularly in the mass region between 2.4 GeV and the cc threshold, where we expect the first excited states of the glueballs to be found. It is because of this that we cannot find a glueball trajectory in the angular momentum plane.

118. Decays of glueballs Recall that the width of the decay of a meson into two mesons is In a similar way the width for the decay of a glueball into two mesons is Thus we get the following hierarchy for the decay of glueballs

119. Decays of glueballs versus mesons

120. Decays of glueballs into two mesons A closed strings decay into two open strings An interesting observation we made there is that the two very narrow resonances f2(1430) and fJ(2220) (the latter being a popular candidate for the tensor glueball) can be connected by a line with a glueball-like slope.

121. Predictions on excited hadrons

122. Predictions on excited baryons with higher J

123. Predictions on excited baryons with higher n

124. Predictions on excited mesons with higher J

125. Predictions on excited mesons with higher n

126. Predictions about Glueballs Unfortunately there are only few confirmed flavorless hadrons with higher J and higher n. When we use the glueball slope we can fit at most 2 points. Higher points are already in a mass range where not much states have been confirmed. We can predict the locations of the higher glueballs and their width based on

127. for glueball with f0(980) ground state

128. for glueball with f0(1370) ground state

129. for glueball with f0(1500) ground state

130. Predictions for glueball with f0(1500) ground state

131. Predictions of tetra quarks based on the Y(4630) Based on the Y(4630) that was observed to decay predominantly to. If we assume that it is on a Regge-like trajectory and we borrow the slop and the endpoint masses from the trajectory we get

132. Summary and Outlook

133. Summary and outlook Hadron spectra fit much better strings in holographic backgrounds rather than the spectra of bulk fields like fluctuations of flavor branes. Holographic Regge trajectories can be mapped into trajectories of strings with massive endpoints. Heavy quark mesons are described in a much better way by the holographic trajectories ( or massive) than the original linear trajectories. Even for the u and d quark there is a non vanishing string endpoint mass of ~60 Mev.

134. Summary and outlook Baryons are also straight strings with tension which is the same as the one of mesons. The baryonic vertex is still mysterious since data prefers it to be massless. Glueballs can be described as rotating folded closed strings Open questions: Quantizing a string with massive endpoints Accounting for the spin and for the intercept Scattering amplitudes of mesons and baryons like (proton-proton scattering) Nuclear interaction and nuclear matter Incorporating leptons….

138. Stringy loops

139. Confining Wilson Loop In SU(N) gauge theories one defines the following gauge invariant operator where C is some contour The quark – antiquark potential can be extracted from a strip Wilson line The signal for confinement is E ~ T st L

140. Stringy Wilson loop The natural stringy dual of the Wilson line ( which obeys the loop equation) is where is the renormalized Nambu Goto action, namely the renormalized world sheet area

141. Computing the stringy WL in general background The basic setup is a d dimensional space time where x || are p space coordinates, and s and x T are radial coordinate and transverse directions. The corresponding NG action is

142. Upon using the gauge the NG action reads where

144. 

145. Variational analysis Since we have decomposed the string profile to two regions, we have to re-investigate the variational analysis The two regions are

146. Variational analysis For region II Thus, the variation of the action is To have  S=0 we must have

147. Necessary conditions for a solution To solve the e.o.m in region II we expand in Hence to solve the e.o.m in II we need Almost the same conditions as for the Wilson line.

148. Decay width of Stringy holographic Mesons

149. A cartoon of possible decays of a (h,m) meson

150. Holographic decay- qualitative picture Quantum mechanically the stringy meson is unstable. Fluctuations of endpoints splitting of the string The string has to split in such a way that the new endpoints are on a flavor brane. The decay probability= (to split at a given point ) X (that the split point is on a flavor brane ) The probability to split of an open string in flat space time was computed by Dai and Polchinski and by Turok et al.

151. The split of an open sting in flat space time Intuitively the string can split at any point and hence we expect width~ L The idea is to use the optical theorem and compute the total rate by computing the imaginary part of the self energy diagram An exercise in one loop string calculation Consider a string stretched around a long compact spatial direction. A winding state splits and joins. In terms of vertex operators it translates to a disk with two closed string vertex operators

152. The corresponding amplitude takes the form where k is the gravitational coupling, g the coefficient of the open string tachyon, the factor L comes from the zero mode. Using the ope’s we get where Performing the integral, taking the imaginary part

153. String bit approximation Using a string bit model the integration over the right subset of configurations becomes easier. K.Peeters M. Zamaklar J.S The discretized string consists of a number of horizontal rigid rods connected by vertical springs.

154. The mass of each bead is M, the length is L=(N+1)a and the action is The normal modes and their frequencies are In the relativistic limit and large N The action now is of N decoupled normal modes The wave function is a product of the wave functions of the normal modes

155. Note that the width of the Gaussian depends on T eff and not on L The integration interval is when the bead is “at the brane” defined by By computing the decay width for various values of N and extrapolating to large N we find that the decay rate is approximated by

156. The basic CNN model predicts In fact and hence incorporating the corrections due to the massive endpoints we find the following blue curve which fits the data points of the K* mesons a function of M  /M- Correction to the decay width due to msep