This paper was converted on www.awesomepapers.org from LaTeX by an anonymous user.
Want to know more? Visit the Converter page.

institutetext: Department of Theoretical Physics,
Tata Institute of Fundamental Research,
Homi Bhabha Road, Mumbai 400005, India

TT¯T\overline{T} and JT¯J\overline{T} Deformations in Quantum Mechanics

Soumangsu Chakraborty, Amiya Mishra [email protected] [email protected]
Abstract

In this paper, we continue the study of TT¯T\overline{T} deformation in d=1d=1 quantum mechanical systems and propose possible analogues of JT¯J\overline{T} deformation and deformation by a general linear combination of TT¯T\overline{T} and JT¯J\overline{T} in quantum mechanics. We construct flow equations for the partition functions of the deformed theory, the solutions to which yields the deformed partition functions as integral of the undeformed partition function weighted by some kernels. The kernel formula turns out to be very useful in studying the deformed two-point functions and analyzing the thermodynamics of the deformed theory. Finally, we show that a non-perturbative UV completion of the deformed theory is given by minimally coupling the undeformed theory to worldline gravity and U(1)U(1) gauge theory.

preprint: TIFR/TH/20-28

1 Introduction

Over the past few years, there has been a growing interest in understanding irrelevant deformations of a CFT2CFT_{2}, the most well-studied among them are the TT¯T\overline{T} Smirnov:2016lqw ; Cavaglia:2016oda , JT¯J\overline{T} Guica:2017lia ; Chakraborty:2018vja and deformation by their general linear combination Chakraborty:2019mdf ; LeFloch:2019rut . Here, TT and T¯\overline{T} are the holomorphic and anti-holomorphic components of the stress tensor and JJ is the holomorphic component of a global U(1)U(1) current. Although such deformations are irrelevant, involving flowing up the renormalization group (RG) trajectory, the deformations turn out to be integrable Smirnov:2016lqw and one can compute the exact spectrum of the deformed theory in terms of the undeformed spectrum and the couplings. The resulting theories in the UV are non-local in the sense that their short distance behavior is not governed by fixed points.

It is well-known that for one sign of the TT¯T\overline{T} coupling, the deformed spectrum is non-unitary. Spectrum of states with energies above some critical value in the undeformed theory, upon deformation becomes complex. For this particular sign of the coupling, there has been a proposal for the holographic dual in the form of AdS3AdS_{3} with a hard radial cutoff McGough:2016lol .111For other holographic proposals and their generalizations, see Giveon:2017nie ; Chakraborty:2018vja ; Chakraborty:2019mdf ; Chakraborty:2020swe ; Guica:2019nzm . In an attempt to visualize holography with a hard Dirichlet wall in higher dimensions, the authors of Bonelli:2018kik ; Taylor:2018xcy ; Hartman:2018tkw ; Belin:2020oib constructed a dual effective field theory which in the deep IR appears to be a large NN CFT deformed by an irrelevant operator which is bilinear in the components of the stress tensor. Such a deformation is often regarded as an analogue of TT¯T\overline{T} deformation in higher dimensions.222In this paper, by abuse of language, we will keep referring to these deformations in d>2d>2 and in d=1d=1 as the TT¯T\overline{T} deformation, similarly for JT¯J\overline{T} and their general linear combinations. The TT¯T\overline{T} operator being a composite operator needs to be defined by point splitting. In d=2d=2 it has been proved that such an operator is indeed well-defined at the coincident limit Zamolodchikov:2004ce . In higher dimensions, however, the well-definedness at the coincident limit is yet to be proven.

Motivated from the holographic analysis with a hard UV wall in dimensions d2d\geq 2, the authors of Gross:2019ach ; Gross:2019uxi defined an analogue of TT¯T\overline{T} deformation in d=1d=1 from a thorough analysis of JT gravity in AdS2AdS_{2} with a hard radial cutoff (see also Iliesiu:2020zld ). In the absence of any spatial directions in d=1d=1 quantum mechanics, the TT¯T\overline{T} operator defined by the right hand side of the flow equation (14), is a composite operator and is well-defined by construction. Such an operator is irrelevant and deformation of a quantum mechanical system in the IR by this operator involves flowing up the RG trajectory much like its higher dimensional cousins. The thermodynamic stability of TT¯T\overline{T} deformation in d=1d=1 has been studied in Barbon:2020amo .

However, RG flows in one dimension enjoy many properties that are absent in higher dimensions. In d=1d=1, there exists only a handful of irrelevant operators in contrast to a gigantic set of relevant operators. This behavior is quite opposite to what happens in d2d\geq 2. As a result, the space of theories in the IR is very rich compared to the landscape in the UV which is highly universal. In some sense, the ambiguities in flowing up the RG triggered by some irrelevant deformation in d2d\geq 2 doesn’t exist in d=1d=1. This itself is a good enough motivation to study irrelevant deformations like TT¯T\overline{T} and JT¯J\overline{T} deformation in d=1d=1 quantum mechanical systems.

In this paper, we continue the study of TT¯T\overline{T} deformation in d=1d=1 initiated in Gross:2019ach and introduce an analogue of JT¯J\overline{T} deformation in quantum mechanics and eventually study deformation by general linear combination of TT¯T\overline{T} and JT¯J\overline{T}. We have shown that the deformed spectrum has the same square root brunch cut singularity as in the case of their higher dimensional cousins. In the case of pure JT¯J\overline{T} deformation, we have shown that the deformed spectrum is non-unitary for either sign of the coupling as in d=2d=2 and upon switching on the TT¯T\overline{T} deformation we see that there is a region in the parameter space (i.e. the space of TT¯T\overline{T} and JT¯J\overline{T} coupling) where the theory is unitary. We introduce a kernel formula such that the integral of the undeformed partition function weighted by some kernel gives rise to the deformed partition function. As an application of the kernel formula, we compute the deformed two-point functions of operators333We have considered only those operators that do not develop explicit dependence on the coupling upon deformation. of the deformed theory. Then, we analyze the thermodynamics of a certain class of theories (systems with linear specific heat) deformed by a general linear combination of TT¯T\overline{T} and JT¯J\overline{T} and show that there is a region in the parameter space, where the theory is unitary and exhibits Hagedorn behavior, similar to the two dimensional case Chakraborty:2020xyz . Unlike d=2d=2, the Hagedorn behavior at high energies is typical to systems with linear specific heat (e.g. Schwarzian theories Stanford:2017thb ). Systems with non-linear specific heat are not likely to exhibit Hagedorn growth.

It has been shown in Gross:2019ach , that the non-perturbative UV completion of TT¯T\overline{T} deformed quantum mechanics is obtained by coupling the theory to one-dimensional worldline gravity. This can be realized as the one-dimensional version of the non-perturbative UV completion of TT¯T\overline{T} deformed quantum field theory in two dimensions where the deformed theory is obtained by minimally coupling the undeformed theory with JT gravity in flat space Dubovsky:2018bmo . In the same spirit, we show a possible non-perturbative UV completion of quantum mechanics deformed by a general linear combination of TT¯T\overline{T} and JT¯J\overline{T} is obtained by coupling the undeformed theory to worldline gravity and U(1)U(1) gauge theory.

Plan of the paper

The organization of this paper is as follows. In section 2, we give a brief review of TT¯T\overline{T}, JT¯J\overline{T} and TT¯+JT¯T\overline{T}+J\overline{T} deformation and their deformed spectrum in two dimensions. In section 3, we define an analogue of TT¯T\overline{T}, JT¯J\overline{T} and TT¯+JT¯T\overline{T}+J\overline{T} deformation in quantum mechanics through their respective flow equations and derive the kernel formulas to compute the deformed partition functions. In section 4, we compute the deformed two-point function and analyze the thermodynamics of the deformed theory. We also comment on the thermodynamic stability of the deformed theory. In section 5, we discuss a possible non-perturbative UV completion of the deformed theory. Finally, in section 6, we conclude our findings and discuss various outlooks and potential future directions.

2 Brief review of TT¯T\overline{T} and related deformations in d=2d=2

In this section, we will present a brief review of what it means by deformation of a d=2d=2 QFT by the operators TT¯T\overline{T}, JT¯J\overline{T} and a linear combination of TT¯T\overline{T}, JT¯J\overline{T}. Here, we will be brief and sketch only the expression of the deformed spectrum without going into much details.

2.1 TT¯T\overline{T} deformation

Let us consider a d=2d=2 Lorentz invariant QFTQFT on a cylinder ×S1\mathbb{R}\times S^{1} of radius RR with a stress tensor TμνT_{\mu\nu} which is conserved. Conservation implies

μTμν=0.\displaystyle\partial_{\mu}T^{\mu\nu}=0. (1)

Then, the TT¯T\overline{T} operator of such a system is defined as

TT¯=det(Tμν).\displaystyle T\overline{T}=-{\rm{det}}(T_{\mu\nu}). (2)

Let the Lagrangian density of such a theory (before deformation) be denoted by 0\mathcal{L}_{0}. Note that the TT¯T\overline{T} operator is a composite operator and must be defined by point splitting. That the coincident limit is finite up to total derivative terms has been proved in Zamolodchikov:2004ce .

Next, let us consider an RG flow in the space of theories parametrized by the affine parameter λ\lambda such that at each point on the trajectory in the space of theories, the flow is triggered by the operator TT¯(λ)T\overline{T}(\lambda) at that point on the trajectory. Setting λ=0\lambda=0 gives back the seed theory one started with.

Let the Lagrangian of the deformed theory at an arbitrary point on the RG trajectory be given by λ\mathcal{L}_{\lambda}. Then the deformation of a d=2d=2 Lorentz invariant QFTQFT by the operator TT¯T\overline{T} is given by the flow equation

λλ=12TT¯(λ).\displaystyle{\partial\mathcal{L}_{\lambda}\over\partial\lambda}=-{1\over 2}T\overline{T}(\lambda)~{}. (3)

Although this deformation is irrelevant, it turns out to be solvable. The deformed spectrum, E(λ;E0,P)E(\lambda;E_{0},P), can be expressed as a function of the spectrum E0E_{0} and momentum PP of the undeformed theory as

E(λ;E0,P)=Rλ(11+2λE0R+λ2P2R2).E(\lambda;E_{0},P)=-{R\over\lambda}\left(1-\sqrt{1+{2\lambda E_{0}\over R}+{\lambda^{2}P^{2}\over R^{2}}}\right). (4)

Note that for λ>0\lambda>0, the deformed spectrum is unitary as opposed to for λ<0\lambda<0, in which case the spectrum becomes complex above some threshold.

2.2 JT¯J\overline{T} deformation

To define a JT¯J\overline{T} deformation of a CFT, let us start with a d=2d=2 CFT on a cylinder of radius RR that contains a conserved left moving U(1)U(1) current JJ. Then the flow equation of JT¯J\overline{T} deformation is given by

αα=2J(α)T¯(α),\displaystyle{\partial\mathcal{L}_{\alpha}\over\partial\alpha}=2J(\alpha)\overline{T}(\alpha), (5)

where α\alpha is the coupling, J(α)J(\alpha) is the deformed left-moving U(1)U(1) current and T¯(α)\overline{T}(\alpha) is the deformed right-moving component of the stress tensor. Similar to the TT¯T\overline{T} deformation, the JT¯J\overline{T} deformation is irrelevant, but unlike the previous case, JT¯J\overline{T} deformation breaks Lorentz invariance.

It has been argued in Chakraborty:2018vja that all along the RG flow,

  1. (i)

    the left-moving component of the stress tensor, T(α)T(\alpha), remains holomorphic i.e.  ¯T(α)=0\overline{\partial}T(\alpha)=0,

  2. (ii)

    the left moving component of the global U(1)U(1) current remains holomorphic i.e. ¯J(α)=0\overline{\partial}J(\alpha)=0,

  3. (iii)

    the quantity T(α)12J(α)2T(\alpha)-{1\over 2}J(\alpha)^{2} is independent of the coupling α\alpha.

From the above three conditions one can construct the deformed spectrum that takes the following form Guica:2017lia ; Chakraborty:2018vja :

E(α;E0,P,Q)=P+2Rα2((1αQR)(1αQR)2α2R(E0P)),E(\alpha;E_{0},P,Q)=P+{2R\over\alpha^{2}}\left(\left(1-\alpha{Q\over R}\right)-\sqrt{\left(1-\alpha{Q\over R}\right)^{2}-{\alpha^{2}\over R}\left(E_{0}-P\right)}\right), (6)

where, QQ is the undeformed left-moving U(1)U(1) charge. Note that in this case the spectrum above some threshold, becomes complex for both signs of the coupling.

2.3 Deformation by a general linear combination of TT¯T\overline{T}, JT¯J\overline{T} and TJ¯T\overline{J}

Next, one can consider a deformation by a general linear combination of TT¯T\overline{T}, JT¯J\overline{T} and TJ¯T\overline{J}. The flow equations are given by

λ,α±λ\displaystyle{\partial\mathcal{L}_{\lambda,\alpha_{\pm}}\over\partial{\lambda}} =\displaystyle= 12TT¯,\displaystyle-{1\over 2}T\overline{T}, (7)
λ,α±α+\displaystyle{\partial\mathcal{L}_{\lambda,\alpha_{\pm}}\over\partial{\alpha_{+}}} =\displaystyle= 2JT¯,\displaystyle 2J\overline{T}, (8)
λ,α±α\displaystyle{\partial\mathcal{L}_{\lambda,\alpha_{\pm}}\over\partial{\alpha_{-}}} =\displaystyle= 2TJ¯,\displaystyle 2T\overline{J}, (9)

where, α±\alpha_{\pm} are respectively the JT¯J\overline{T} and TJ¯T\overline{J} coupling. Defining the theory is a bit involved, so we will refer to the references LeFloch:2019rut ; Chakraborty:2019mdf for more details. The deformed spectrum takes the following form

E(λ,α±,E0,P,QL,R)=P12AR(BB2+4AC),\displaystyle E(\lambda,\alpha_{\pm},E_{0},P,Q_{L,R})=P-{1\over 2AR}\left(-B-\sqrt{B^{2}+4AC}\right), (10)

with,

A\displaystyle A =\displaystyle= 14R2(2λ(α++α)2),\displaystyle{1\over 4R^{2}}\left(2\lambda-(\alpha_{+}+\alpha_{-})^{2}\right),
B\displaystyle B =\displaystyle= 1+1R(α+QL+αQR)+PRα(α++α)λPR,\displaystyle-1+{1\over R}(\alpha_{+}Q_{L}+\alpha_{-}Q_{R})+{P\over R}\alpha_{-}(\alpha_{+}+\alpha_{-})-{\lambda P\over R}, (11)
C\displaystyle C =\displaystyle= (E0P)R+2αQRP+α2P2,\displaystyle(E_{0}-P)R+2\alpha_{-}Q_{R}P+\alpha_{-}^{2}P^{2},

where, QL,RQ_{L,R} are the left and right-moving U(1)U(1) charges. For A0A\geq 0 the deformed spectrum is unitary, else there is a threshold above which the spectrum becomes complex.

In the following section, we are going to draw motivations from the expressions of the deformed spectrum in d=2d=2 to define an analogue of TT¯T\overline{T} and JT¯J\overline{T} deformations in one-dimensional quantum mechanical system.

3 TT¯T\overline{T} and related deformation in quantum mechanics

The authors of Gross:2019ach ; Gross:2019uxi discussed an infinite class of one parameter family of solvable deformations of quantum mechanical systems where the undeformed Hamiltonian, H0H_{0}, gets mapped to some function of itself i.e. H=f(λ,H0)H=f(\lambda,H_{0}) where, HH denotes the deformed Hamiltonian and H(λ=0)=H0H(\lambda=0)=H_{0}. Since [H0,H]=0[H_{0},H]=0, the energy eigenstates remain undeformed under the deformation. The eigenvalues of the deformed Hamiltonian, on the other hand, change as E=f(λ,E0)E=f(\lambda,E_{0}), where E0E_{0} is the spectrum of the undeformed system. Such a deformation is defined by a flow equation

Hλ=F(λ,H),{\partial H\over\partial\lambda}=F(\lambda,H), (12)

where, F(λ,H)F(\lambda,H) is the deforming operator and λ\lambda is the deformation parameter. Note that (12) is a first order differential equation implying that its solution is unique given an initial condition.

These solvable deformations have an obvious generalization to systems that have additional commuting global symmetries. In that case the deformed Hamiltonian takes the form H=f(λ,H0,Q)H=f(\lambda,H_{0},Q) where QQ is the charge associated with the additional global symmetry.444Here we are assuming that QQ remains undeformed upon deformation. An example of such a deformation would be the analogue of JT¯J\overline{T} deformation in quantum mechanics that we will discuss in details in the subsection 3.2.

These solvable deformations have a further generalization to two or more parameter family of flow equations namely

Hλi=Fi({λi},H), with i=1,2,.\displaystyle{\partial H\over\partial\lambda_{i}}=F_{i}(\{\lambda_{i}\},H),\ \text{ with }\ i=1,2,\cdots. (13)

An example of such a deformation would be deformation by a general linear combination of TT¯T\overline{T} and JT¯J\overline{T} in quantum mechanical systems with a global U(1)U(1) symmetry, which we are going to introduce in subsection 3.3.

3.1 TT¯T\overline{T} deformation in quantum mechanics

The motivation for studying TT¯T\overline{T} deformation in quantum mechanics is two-fold. Firstly, to understand JTJT gravity with a finite radial cutoff in the bulk AdS2AdS_{2} Gross:2019ach ; Gross:2019uxi ; Iliesiu:2020zld and secondly to study a solvable deformation of quantum mechanics that would give rise to a deformed spectrum with a square root branch cut singularity like the one that appears in two dimensions (4). The TT¯T\overline{T} flow equation as defined in Gross:2019ach ; Gross:2019uxi takes the following form:

SEλ=𝑑τT22+2λT,\displaystyle{\partial S_{E}\over\partial\lambda}=-\int d\tau~{}{T^{2}\over 2+2\lambda T}, (14)

where, SES_{E} is the euclidean action of the deformed theory, τ\tau is the Euclidean time, and T=Tττ=HT=T^{\tau}_{\tau}=H is the stress tensor of the system. This leads to the differential equation for the deformed spectrum

2dEdλ+2λEdEdλ+E2=0,2{dE\over d\lambda}+2\lambda E{dE\over d\lambda}+E^{2}=0, (15)

which solves as

E(λ)=1λ(11+2λE0).E(\lambda)=-{1\over\lambda}\left(1-\sqrt{1+2\lambda E_{0}}\right). (16)

The spectrum agrees precisely with the energy of a two dimensional black hole in AdS2AdS_{2} with a finite radial cutoff Gross:2019ach for λ<0\lambda<0. Note that, the limit λ0\lambda\to 0 leads to the undeformed spectrum. As in the case of two dimensions, the deformed spectrum is unitary for λ>0\lambda>0 whereas for λ<0\lambda<0, there is a threshold above which the spectrum becomes complex. One can thus read off the deformed Hamiltonian operator as

H(λ)=1λ(11+2λH0).\displaystyle H(\lambda)=-{1\over\lambda}\left(1-\sqrt{1+2\lambda H_{0}}\right). (17)

3.1.1 Thermal partition function and its flow equation

In this subsubsection, we are going to present an alternative way to realize the flow equations in terms of the thermal partition function eventually leading to a kernel formula that generates the deformed thermal partition function. This method turns out to be more useful in analyzing the deformed thermal partition function in the case of JT¯J\overline{T} deformation and eventually to a class of more general deformations.

The thermal partition function of the deformed theory is given by the celebrated trace formula

Z(λ,β)=tr(eβH(λ))=eβE(λ),\displaystyle Z(\lambda,\beta)={\rm tr}\left(e^{-\beta H(\lambda)}\right)=\sum e^{-\beta E(\lambda)}, (18)

where, β\beta is the inverse temperature. Note that β\beta can also be realized as the periodicity of the Euclidean time τ\tau namely ττ+β\tau\sim\tau+\beta.

Equation (15) implies

tr[(2dHdλ+2λHdHdλ+H2)eβH]=0.{\rm tr}\left[\left(2{dH\over d\lambda}+2\lambda H{dH\over d\lambda}+H^{2}\right)e^{-\beta H}\right]=0. (19)

The expectations of the individual terms on the l.h.s. of (19) take the following forms:

ZdHdλ=1βZλ,ZHdHdλ=1β(β1β)Zλ,ZH2=Z2β2.Z\left\langle{dH\over d\lambda}\right\rangle=-{1\over\beta}{\partial Z\over\partial\lambda},\ \ \ \ \ Z\left\langle H{dH\over d\lambda}\right\rangle={1\over\beta}\bigg{(}{\partial\over\partial\beta}-{1\over\beta}\bigg{)}{\partial Z\over\partial\lambda},\ \ \ \ \ Z\left\langle H^{2}\right\rangle={\partial Z^{2}\over\partial\beta^{2}}. (20)

Substituting (20) in (19) gives the flow equation in terms of the thermal partition function:

2βZλ+2λβ(β1β)Zλ+Z2β2=0,-{2\over\beta}{\partial Z\over\partial\lambda}+{2\lambda\over\beta}\bigg{(}{\partial\over\partial\beta}-{1\over\beta}\bigg{)}{\partial Z\over\partial\lambda}+{\partial Z^{2}\over\partial\beta^{2}}=0, (21)

which matches exactly with the flow equation constructed from the bulk JT gravity in cutoff AdS2AdS_{2} with Dirichlet boundary condition on the cutoff surface Iliesiu:2020zld . This can also be visualized as the defining equation for TT¯T\overline{T} deformation in quantum mechanics. Equation (21) solves as

Z(λ,β)=𝒞β𝑑βK(λ,β,β)Z0(β),Z(\lambda,\beta)=\int_{\mathcal{C}_{\beta}}d\beta^{\prime}\ K(\lambda,\beta,\beta^{\prime})~{}Z_{0}(\beta^{\prime}), (22)

where Z0(β)Z_{0}(\beta^{\prime}) is the partition function of the undeformed theory at temperature 1/β1/\beta^{\prime}, the contour 𝒞β\mathcal{C}_{\beta} runs from 0 to \infty along the real β\beta axis and the kernel K(λ,β,β)K(\lambda,\beta,\beta^{\prime}) is given by 555If the quantum mechanical system has a finite chemical potential then kernel (23) can be trivially generalized by multiplying it with δ(μμ)\delta(\mu^{\prime}-\mu) (where μ\mu^{\prime} and μ\mu are respectively the chemical potentials of the undeformed and deformed systems.) and integrating over μ\mu^{\prime} in (22) from -\infty to \infty.

K(λ,β,β)=ββ3/22πλexp((ββ)22βλ).K(\lambda,\beta,\beta^{\prime})={\beta\over\beta^{\prime 3/2}\sqrt{2\pi\lambda}}~{}{\rm exp}\bigg{(}-{(\beta-\beta^{\prime})^{2}\over 2\beta^{\prime}\lambda}\bigg{)}. (23)

The kernel agrees with the one presented in Gross:2019uxi . It is obvious that the kernel (23) also satisfies the differential equation (21). From (22) one can read off the spectrum of the deformed theory which turns out to be (16).

Note that the kernel formula (22) with (23) is strictly well-defined for λ>0\lambda>0. For λ<0\lambda<0 the integral in (22) is divergent. This is related to the fact that for λ<0\lambda<0, the spectrum is non-unitary. We will hence take the attitude of defining the integral for λ>0\lambda>0 and analytically continue the final result for λ<0\lambda<0.666Alternatively, for λ<0\lambda<0 one can choose a separate contour such that the β\beta^{\prime} integral is convergent.

3.2 JT¯J\overline{T} deformation in quantum mechanics

An obvious question that one may raise at this point is: what would be an analogue of JT¯J\overline{T} deformation in quantum mechanics? In the absence of any nice holographic interpretation, we would like to draw motivation from the spectrum (6) of a CFT2CFT_{2} deformed by the JT¯J\overline{T} operator. The deformed spectrum in two dimensions has the properties that it has a square root branch cut much like the case of TT¯T\overline{T} and that for both signs of the JT¯J\overline{T} coupling, the spectrum is non-unitary. Inspired from (6), let us define the deformed spectrum of a one-dimensional quantum mechanical system with a global U(1)U(1) charge QQ (which in quantum mechanics is also the conserved current JJ) as follows:777Note that in quantum mechanics, there are no space directions. Thus in defining the deformed spectrum in one dimension, one sets the momentum P=0P=0.

E(α)=2α2(1αQ(1αQ)2α2E0),E(\alpha)={2\over\alpha^{2}}\left(1-{\alpha Q\over\sqrt{\ell}}-\sqrt{\left(1-{\alpha Q\over\sqrt{\ell}}\right)^{2}-\alpha^{2}E_{0}}~{}\right), (24)

where α\alpha is the coupling of the deforming operator, and \ell is some length scale associated with the field space of undeformed theory (e.g. for a free compact scalar yy+2πy\sim y+2\pi\sqrt{\ell}, \sqrt{\ell} is the radius of the compact boson in the configuration space.).888In the canonical formalism, the charge QQ is dimensionless in all dimensions. The coupling α\alpha has length dimension 1/21/2. Thus to make the quantity α/\alpha/\sqrt{\ell} dimensionless, \ell must have dimension of length.

The flow equation that would give rise to such a deformed spectrum is given by

SEα=𝑑ταT2+2JT2(1αJ)α2T.{\partial S_{E}\over\partial\alpha}=\int d\tau~{}{\alpha T^{2}+{2JT\over\sqrt{\ell}}\over 2\left(1-{\alpha J\over\sqrt{\ell}}\right)-\alpha^{2}T}. (25)

This leads to the following differential equation of the deformed spectrum

2dEdα2αQdEdαα2EdEdα2QEαE2=0,2{dE\over d\alpha}-{2\alpha Q\over\sqrt{\ell}}{dE\over d\alpha}-\alpha^{2}E{dE\over d\alpha}-{2Q\over\sqrt{\ell}}E-\alpha E^{2}=0, (26)

which solves as (24).999Note that, Q=0Q=0 and α22λ\alpha^{2}\to-2\lambda give rise to the flow equation and deformed spectrum for the TT¯T\overline{T} case.101010Recall that we assumed QQ remains undeformed all along the flow.

3.2.1 Thermal partition function and its flow equation

Next, we are going to define a flow equation in terms of the thermal partition function. Here we follow the same line of arguments as used in the case of TT¯T\overline{T} deformation. The thermal partition function of the deformed system coupled to a finite chemical potential μ\mu is given by

Z(α,β,μ)=tr(eβH+μQ/).\displaystyle Z(\alpha,\beta,\mu)={\rm tr}\left(e^{-\beta H+\mu Q/\sqrt{\ell}}\right). (27)

The flow equation of the deformed spectrum (26) implies

tr[(2dHdα2αJdHdαα2HdHdα2JHαH2)eβH+μQ/]=0.{\rm tr}\left[\left(2{dH\over d\alpha}-{2\alpha J\over\sqrt{\ell}}{dH\over d\alpha}-\alpha^{2}H{dH\over d\alpha}-{2J\over\sqrt{\ell}}H-\alpha H^{2}\right)e^{-\beta H+\mu Q/\sqrt{\ell}}\right]=0. (28)

The expectation of the individual terms on the l.h.s. of (28) take the following forms:

ZdHdα=1βZα,ZJdHdα=β2Zμα,ZHdHdα=1β(β1β)Zα,ZJH=2Zβμ,ZH2=Z2β2.\begin{split}Z\langle{dH\over d\alpha}\rangle=-{1\over\beta}{\partial Z\over\partial\alpha},\ \ &\ \ Z\langle J{dH\over d\alpha}\rangle=-{\sqrt{\ell}\over\beta}{\partial^{2}Z\over\partial\mu\partial\alpha},\\ Z\langle H{dH\over d\alpha}\rangle={1\over\beta}\bigg{(}{\partial\over\partial\beta}-{1\over\beta}\bigg{)}{\partial Z\over\partial\alpha},\ \ &\ \ Z\langle JH\rangle=-\sqrt{\ell}{\partial^{2}Z\over\partial\beta\partial\mu},\ \ \ \ Z\langle H^{2}\rangle={\partial Z^{2}\over\partial\beta^{2}}.\end{split} (29)

Substituting (29) in (28) yields

2βZα+2αβ2Zμαα2β(β1β)Zα+22Zβμα2Zβ2=0.\begin{split}-{2\over\beta}{\partial Z\over\partial\alpha}+{2\alpha\over\beta}{\partial^{2}Z\over\partial\mu\partial\alpha}-{\alpha^{2}\over\beta}\bigg{(}{\partial\over\partial\beta}-{1\over\beta}\bigg{)}{\partial Z\over\partial\alpha}+2{\partial^{2}Z\over\partial\beta\partial\mu}-\alpha{\partial^{2}Z\over\partial\beta^{2}}=0.\end{split} (30)

As in the case of TT¯T\overline{T} deformation, this can be realized as the defining equation of an analogue of JT¯J\overline{T} deformation in quantum mechanics. Equation (30) solves as

Z(α,β,μ)=𝒞β𝑑β𝒞μ𝑑μK(α,β,β,μ,μ)Z0(β,μ),Z(\alpha,\beta,\mu)=\int_{\mathcal{C}_{\beta}}d\beta^{\prime}\int_{\mathcal{C_{\mu}}}d\mu^{\prime}\ K(\alpha,\beta,\beta^{\prime},\mu,\mu^{\prime})\ Z_{0}(\beta^{\prime},\mu^{\prime}), (31)

where, as before Z0(β,μ)Z_{0}(\beta^{\prime},\mu^{\prime}) is the partition function of the undeformed theory at temperature 1/β1/\beta^{\prime} and chemical potential μ\mu^{\prime}. The contour 𝒞μ\mathcal{C}_{\mu} runs all along the real μ\mu^{\prime} axis from -\infty to \infty and the contour 𝒞β\mathcal{C}_{\beta} runs all along the positive imaginary β\beta^{\prime} axis from 0 to ii\infty. The kernel K(α,β,β,μ,μ)K(\alpha,\beta,\beta^{\prime},\mu,\mu^{\prime}) takes the form

K(α,β,β,μ,μ)=β2πiβ2αexp((μμ)24β(ββ)αβ(μμ)).K(\alpha,\beta,\beta^{\prime},\mu,\mu^{\prime})={\beta\over 2\pi i\beta^{\prime 2}\alpha}~{}{\rm exp}\bigg{(}-{(\mu-\mu^{\prime})^{2}\over 4\beta^{\prime}}-{(\beta-\beta^{\prime})\over\alpha\beta^{\prime}}(\mu-\mu^{\prime})\bigg{)}. (32)

From (31) one can read off the deformed spectrum (24).

3.3 TT¯+JT¯T\overline{T}+J\overline{T} deformation in quantum mechanics

In this subsection, we are going to turn on a deformation that is a general linear combination of TT¯T\overline{T} and JT¯J\overline{T}. As stated in subsection 3.2, the JT¯J\overline{T} deformation alone makes the theory non-unitary. In the discussion that follows, we are going to argue that in the presence of both these deformations, there exist a region in the parameter space (i.e. the space spanned by λ\lambda and α\alpha) where the deformed theory is unitary. To define such a deformation, we are, as before, going to draw motivation from the deformed spectrum (10), (11). Thus let us define the spectrum of a one-dimensional quantum mechanical system with a global U(1)U(1) charge QQ to be of the form111111Note that, one can further generalize the deformed spectrum by keeping both α±\alpha_{\pm} and QL/RQ_{L/R} from (10) and defining resulting spectrum as the spectrum obtained by deforming a quantum mechanical system by a general linear combination of three operators namely TT¯T\overline{T}, JT¯J\overline{T} and TJ¯T\overline{J}. The charges QL/RQ_{L/R} can be thought of as charges coming from two global U(1)U(1)’s.

E(α,λ)=2h(1αQ(1αQ)2+E0h),E(\alpha,\lambda)=-2h\left(1-{\alpha Q\over\sqrt{\ell}}-\sqrt{\left(1-{\alpha Q\over\sqrt{\ell}}\right)^{2}+{E_{0}\over h}}~{}\right), (33)

where

h=12λα2\displaystyle h={1\over 2\lambda-\alpha^{2}} (34)

and as before λ\lambda and α\alpha are respectively the TT¯T\overline{T} and JT¯J\overline{T} coupling. It’s obvious from the deformed spectrum that the theory is unitary in the regime in the parameter space where h>0h>0. This corresponds to the region above the red curve (i.e. the locus λ=α2/2\lambda=\alpha^{2}/2) in figure 1. For h<0h<0, which corresponds to the region below the red curve in figure 1, the deformed spectrum above some threshold becomes complex. Approaching the red curve from the above (i.e. h+h\to+\infty) gives an interesting intermediate regime where the theory is unitary and exhibits many exotic behaviors, that we will discuss later.

α\alphaλ\lambdah>0h>0h<0h<0λ=α2/2\lambda=\alpha^{2}/2
Figure 1: The region on and above the red curve in the parameter space corresponds to deformed theories with unitary spectrum while its complement corresponds to those with non-unitary spectrum.

Note that in the limit α0\alpha\to 0, (33) gives rise to the TT¯T\overline{T} deformed spectrum (16) and in the limit λ0\lambda\to 0, it gives the JT¯J\overline{T} deformed spectrum (24). As specified earlier there is an intermediate regime hh\to\infty where the deformed spectrum takes the form

E=E01αQ.E={E_{0}\over 1-{\alpha Q\over\sqrt{\ell}}}~{}. (35)

The flow equations that would give rise to such a deformed spectrum are given by

SEλ=𝑑τT22(1αJ)+Th,SEα=𝑑ταT2+2JT2(1αJ)+Th.\begin{split}&{\partial S_{E}\over\partial\lambda}=-\int d\tau~{}{T^{2}\over 2\left(1-{\alpha J\over\sqrt{\ell}}\right)+{T\over h}},\\ &{\partial S_{E}\over\partial\alpha}=\int d\tau~{}{\alpha T^{2}+{2JT\over\sqrt{\ell}}\over 2\left(1-{\alpha J\over\sqrt{\ell}}\right)+{T\over h}}.\end{split} (36)

This leads to the following differential equations in the deformed spectrum:

2(1Qα)Eλ+EhEλ+E2=0,2(1Qα)Eα+EhEααE22QE=0,\begin{split}&2\left(1-{Q\alpha\over\sqrt{\ell}}\right){\partial E\over\partial\lambda}+{E\over h}{\partial E\over\partial\lambda}+E^{2}=0,\\ &2\left(1-{Q\alpha\over\sqrt{\ell}}\right){\partial E\over\partial\alpha}+{E\over h}{\partial E\over\partial\alpha}-\alpha E^{2}-{2Q\over\sqrt{\ell}}E=0,\end{split} (37)

which solves as (33).

3.3.1 Thermal partition function and its flow equation

Using the same logic as discussed in the case of TT¯T\overline{T} and JT¯J\overline{T} deformations, the flow equations (37) for the deformed spectrum imply

tr[((1Jα)Hλ+HhHλ+H2)eβH+μJ/]=0,tr[(2(1Jα)Hα+HhHααH22JH)eβH+μJ/]=0.\begin{split}&{\rm tr}\left[\left(\left(1-{J\alpha\over\sqrt{\ell}}\right){\partial H\over\partial\lambda}+{H\over h}{\partial H\over\partial\lambda}+H^{2}\right)e^{-\beta H+\mu J/\sqrt{\ell}}\right]=0,\\ &{\rm tr}\left[\left(2\left(1-{J\alpha\over\sqrt{\ell}}\right){\partial H\over\partial\alpha}+{H\over h}{\partial H\over\partial\alpha}-\alpha H^{2}-{2J\over\sqrt{\ell}}H\right)e^{-\beta H+\mu J/\sqrt{\ell}}\right]=0.\end{split} (38)

This leads to the following flow equations of the thermal partition function:

2(αμ1)Zλ+1h(β1β)Zλ+β2Zβ2=0,2(αμ1)Zα+1h(β1β)Zαβ(αβ2μ)Zβ=0.\begin{split}&2\left(\alpha{\partial\over\partial\mu}-1\right){\partial Z\over\partial\lambda}+{1\over h}\left({\partial\over\partial\beta}-{1\over\beta}\right){\partial Z\over\partial\lambda}+\beta{\partial^{2}Z\over\partial\beta^{2}}=0,\\ &2\left(\alpha{\partial\over\partial\mu}-1\right){\partial Z\over\partial\alpha}+{1\over h}\left({\partial\over\partial\beta}-{1\over\beta}\right){\partial Z\over\partial\alpha}-\beta\left(\alpha{\partial\over\partial\beta}-2{\partial\over\partial\mu}\right){\partial Z\over\partial\beta}=0.\end{split} (39)

The above set of differential equations solve as

Z(λ,α,β,μ)=𝒞β𝑑β𝒞μ𝑑μK(λ,α,β,β,μ,μ)Z0(β,μ),Z(\lambda,\alpha,\beta,\mu)=\int_{\mathcal{C}_{\beta}}d\beta^{\prime}\int_{\mathcal{C_{\mu}}}d\mu^{\prime}\ K(\lambda,\alpha,\beta,\beta^{\prime},\mu,\mu^{\prime})\ Z_{0}(\beta^{\prime},\mu^{\prime}), (40)

where, as before Z0(β,μ)Z_{0}(\beta^{\prime},\mu^{\prime}) is the partition function of the undeformed theory at temperature 1/β1/\beta^{\prime} and chemical potential μ\mu^{\prime}. The contour 𝒞μ\mathcal{C}_{\mu} runs along the full imaginary μ\mu^{\prime} axis from i-i\infty to ii\infty and the contour 𝒞β\mathcal{C}_{\beta} runs along the real positive β\beta^{\prime} axis from 0 to \infty. The kernel K(λ,α,β,β,μ,μ)K(\lambda,\alpha,\beta,\beta^{\prime},\mu,\mu^{\prime}) takes the form

K(λ,α,β,β,μ,μ)=β2πiβ2αexp((μμ)24hα2β(ββ)αβ(μμ)).K(\lambda,\alpha,\beta,\beta^{\prime},\mu,\mu^{\prime})={\beta\over 2\pi i\beta^{\prime 2}\alpha}~{}{\rm exp}\bigg{(}{(\mu^{\prime}-\mu)^{2}\over 4h\alpha^{2}\beta^{\prime}}-{(\beta^{\prime}-\beta)\over\alpha\beta^{\prime}}(\mu^{\prime}-\mu)\bigg{)}. (41)

For the sake of convergence of the integrals in (40), hh has to be greater than zero. For h<0h<0, one can define the integral for h>0h>0 and then analytically continue the result for h<0h<0. Alternatively, for h<0h<0, one can choose appropriate contours of integrations to make the integrals finite yielding the desired deformed partition function. From (40), one can easily read off the deformed spectrum (33).

4 Correlation functions and thermodynamics

In this section, as an application of the kernel formula (40) and (41) discussed in the previous section, we are going to compute the correlation functions of the deformed theory and analyze the thermodynamics of the deformed system. The discussion that follows, are going to be made in the most general setup namely in the presence of both TT¯T\overline{T} and JT¯J\overline{T} deformation at the same time. Then we are going to comment on various interesting limiting cases.

4.1 Deformed correlation function

The discussion here is in the same spirit as appears in Gross:2019uxi . Let us consider a set of nn operators 𝒪i(τi)\mathcal{O}_{i}(\tau_{i}) i={1,2,,n}\forall i=\{1,2,\cdots,n\} that doesn’t change upon deformation inserted at τi\tau_{i}. For simplicity lets also assume that τ1>τ2>>τn1>0\tau_{1}>\tau_{2}>\cdots>\tau_{n-1}>0.Then the n-point thermal correlator of the deformed theory is given by

G(n)(β,{τi})=𝑑E1𝑑Q1E1Q1|𝒪1(τ1)𝒪n(0)|Q1E1eβE1+μQ1/=i=1ndEidQiE1Q1|𝒪1|Q2E2EnQn|𝒪n|Q1E1ei=0n1βiEi+1+μQ1,\begin{split}G^{(n)}(\beta,\{\tau_{i}\})&=\int dE_{1}dQ_{1}\langle E_{1}Q_{1}|\mathcal{O}_{1}(\tau_{1})\cdots\mathcal{O}_{n}(0)|Q_{1}E_{1}\rangle~{}e^{-\beta E_{1}+\mu Q_{1}/\sqrt{\ell}}\\ &=\int\prod_{i=1}^{n}dE_{i}dQ_{i}\langle E_{1}Q_{1}|\mathcal{O}_{1}|Q_{2}E_{2}\rangle\cdots\langle E_{n}Q_{n}|\mathcal{O}_{n}|Q_{1}E_{1}\rangle~{}e^{-\sum_{i=0}^{n-1}\beta_{i}E_{i+1}+{\mu Q_{1}\over\sqrt{\ell}}},\end{split} (42)

where βi=τiτi+1\beta_{i}=\tau_{i}-\tau_{i+1} for i=0,1,,n1i=0,1,\cdots,n-1, τ0=β\tau_{0}=\beta and βn1=τn1\beta_{n-1}=\tau_{n-1}. Using the kernel formula (40) and (41), one can express the deformed n-point thermal correlation function in terms of the undeformed correlation function G0(n)(β,{τi})G_{0}^{(n)}(\beta^{\prime},\{\tau^{\prime}_{i}\}) as

G(n)(β,{τi})=(i=0n1dβidμiK(β,βi,δi,0μ,μi))G0(n)(β,{τi}),G^{(n)}(\beta,\{\tau_{i}\})=\int\left(\prod_{i=0}^{n-1}d\beta^{\prime}_{i}d\mu^{\prime}_{i}~{}K(\beta,\beta^{\prime}_{i},\delta_{i,0}\mu,\mu^{\prime}_{i})\right)G_{0}^{(n)}(\beta^{\prime},\{\tau^{\prime}_{i}\}), (43)

where βi=τiτi+1\beta^{\prime}_{i}=\tau^{\prime}_{i}-\tau^{\prime}_{i+1} for i=0,1,,n1i=0,1,\cdots,n-1, τ0=β\tau^{\prime}_{0}=\beta^{\prime} and βn1=τn1\beta^{\prime}_{n-1}=\tau^{\prime}_{n-1}.

As a simple example let us consider the deformation of the 2-point functions of conformal quantum mechanics at zero chemical potential. The 2-point functions of operators of dimension Δ\Delta in conformal quantum mechanics is given by DeAlfaro:1977kq ; Chamon:2011xk

𝒪(τ)𝒪(0)0=1|τ|2Δ.\langle\mathcal{O}(\tau)\mathcal{O}(0)\rangle_{0}={1\over|\tau|^{2\Delta}}. (44)

Thus, using (43) the 2-point function of a conformal quantum mechanics deformed by a general linear combination of TT¯T\overline{T} and JT¯J\overline{T} at zero chemical potential (i.e. μ=0\mu=0) is given by 121212In (45), Kn(x)K_{n}(x) is the modified Bessel function of second kind, not to be confused with the kernel KK.

𝒪(τ)𝒪(0)λ,α=2hπ|τ|2Δ+12e2h|τ|K2Δ+12(2h|τ|).\langle\mathcal{O}(\tau)\mathcal{O}(0)\rangle_{\lambda,\alpha}=2\sqrt{{h\over\pi}}~{}|\tau|^{-2\Delta+{1\over 2}}~{}e^{2h|\tau|}~{}K_{2\Delta+{1\over 2}}(2h|\tau|). (45)

Note that the λ\lambda and α\alpha dependences in (45) come in the combination hh. Thus, in the limit h=h=\infty (i.e. λ=α2/2\lambda=\alpha^{2}/2), the two point function of the deformed theory takes the same form as that of the undeformed theory. This implies that in the absence of chemical potential, there exists a regime in the parameter space (i.e. for all points on the red curve in figure 1) where the theory “looks undeformed”. This is reminiscent of the fact that the deformed energy spectrum in the limit hh\to\infty, at some fixed charge remains unchanged up to some multiplicative factor.

Starting from a conformal quantum mechanics, the TT¯T\overline{T} and the JT¯J\overline{T} deformation break certain symmetries (e.g. scale invarience) of the seed theory. It’s likely that in the regime in the parameter space where h=h=\infty, there is a restoration of some amount of broken symmetries. This is reflected in the fact that the 2-point function “looks undeformed” in this limit. It’s as if that the effects of the TT¯T\overline{T} and JT¯J\overline{T} deformations cancel each other in this regime in the parameter space. In the grand canonical ensemble, where the system is coupled to a finite chemical potential, there we do expect to see some changes.

For h<0h<0, the deformed Euclidean 2-point function (45) becomes complex. This is related to the non-unitarity of the theory for h<0h<0.

4.2 Thermodynamics and Hagedorn behavior

In this subsection we are going to present another application of the kernel formula (40) and (41) in analyzing the thermodynamics of a certain class of quantum systems. This is going to reveal the non-local nature of the theory in the UV.

Let us consider systems with linear specific heat in the presence of a finite chemical potential 131313In d=1d=1 quantum mechanics, systems like Schwarzian models exhibit linear specific heat Stanford:2017thb . The kind of systems we are interested in (46) could be thought of as a charged versions of the usual Schwarzian models. This could also be relevant in understanding the holographic dual of TT¯+JT¯T\overline{T}+J\overline{T} deformed charged SYK model.. The partition function of such a thermodynamic systems schematically takes the following form

logZ0=c1β+c2μ2β,\log Z_{0}={c_{1}\over\beta}+{c_{2}\mu^{2}\over\beta}, (46)

where c1,2c_{1,2} are system dependent positive constants proportional to the number of degrees of freedom of the system. Using the kernel formula, one can derive the deformed partition function and analyze the thermodynamics of the deformed system. Let’s do it in two steps: first lets analyze the TT¯T\overline{T} case and then analyze the more general case namely TT¯+JT¯T\overline{T}+J\overline{T} at zero chemical potential of the deformed theory.

4.2.1 TT¯T\overline{T} case

The leading contribution to the deformed partition function is given by

logZ(λ,β,βh)2c1βh2(ββ2βh2),\displaystyle\log Z(\lambda,\beta,\beta_{h})\sim{{2c_{1}\over\beta_{h}^{2}}\left(\beta-\sqrt{\beta^{2}-\beta_{h}^{2}}\right)}, (47)

where

βh=2c1λ\beta_{h}=\sqrt{2c_{1}\lambda} (48)

is the branch point in β\beta interpretable as the inverse Hagedorn temperature of the system. Let F=1βlogZ(λ,β,βh)F=-{1\over\beta}\log Z(\lambda,\beta,\beta_{h}) be the free energy of the system, then the thermal entropy is given by

S=β2Fβ2c1β2βh2.S=\beta^{2}{\partial F\over\partial\beta}\sim{2c_{1}\over\sqrt{\beta^{2}-\beta_{h}^{2}}}. (49)

Next, we Legendre transform FF to express the energy EE of the system as

E=F+S/β.E=F+S/\beta. (50)

From (49) and (50) β\beta solves as

β(2c1+βh2E)4c1E+βh2E2.\beta\sim{(2c_{1}+\beta_{h}^{2}E)\over\sqrt{4c_{1}E+\beta_{h}^{2}E^{2}}}. (51)

Substituting (51) in (49), one finds the thermal entropy of the system as

S4c1E+βh2E2.\displaystyle S\sim\sqrt{4c_{1}E+\beta_{h}^{2}E^{2}}. (52)

Thus at low energies, the entropy of the system goes as S4c1ES\sim\sqrt{4c_{1}E} whereas at very high energies, the entropy grows as SβhES\sim\beta_{h}E signaling the Hagedorn nature of the short distance physics with inverse Hagedorn temperature βh\beta_{h}.

One can also read of the Hagedorn temperature from the deformed density of states ρ(λ,E)\rho(\lambda,E)

ρ(λ,E)e4c1E+βh2E2.\rho(\lambda,E)\sim e^{\sqrt{4c_{1}E+\beta_{h}^{2}E^{2}}}. (53)

4.2.2 TT¯+JT¯T\overline{T}+J\overline{T} case

The leading contribution to the deformed partition function, in this case, is given by

logZ(λ,α,β,βh)2c1βh2(ββ2βh2).\displaystyle\log Z(\lambda,\alpha,\beta,\beta_{h})\sim{2c_{1}\over\beta_{h}^{2}}\left(\beta-\sqrt{\beta^{2}-\beta_{h}^{2}}\right). (54)

Surprisingly the functional form of the deformed partition function at leading order (54) is identical to that of the TT¯T\overline{T} case. The inverse Hagedorn temperature, as read off from the branch point of the deformed partition function is given by

βh\displaystyle\beta_{h} =\displaystyle= c1h(1+4c2hα2).\displaystyle\sqrt{{c_{1}\over h}(1+4c_{2}h\alpha^{2})}. (55)

The inverse temperature and thermal entropy as a function of the energy takes the following form

β(2c1+βh2E)4c1E+βh2E2,S4c1E+βh2E2.\begin{split}&\beta\sim{(2c_{1}+\beta_{h}^{2}E)\over\sqrt{4c_{1}E+\beta_{h}^{2}E^{2}}},\\ &S\sim\sqrt{4c_{1}E+\beta_{h}^{2}E^{2}}.\end{split} (56)

Note that at very high energies, ββh\beta\to\beta_{h} and SβhES\sim\beta_{h}E, whereas at low energies S4c1ES\sim\sqrt{4c_{1}E}. The density of states goes as

ρ(λ,α,E)e4c1E+βh2E2.\displaystyle\rho(\lambda,\alpha,E)\sim e^{\sqrt{4c_{1}E+\beta_{h}^{2}E^{2}}}. (57)

At this point one might wonder which states are contributing to the Hagedorn (55). To answer this question let us analyze the deformed spectrum (33). The undeformed energy as a function of the deformed energy and the couplings is given by

E0=(1αQ)E+E24h.\displaystyle E_{0}=\left(1-{\alpha Q\over\sqrt{\ell}}\right)E+{E^{2}\over 4h}. (58)

The entropy (at fixed charge) of the undeformed theory with partition function (46) is given by (see appendix A)

SQ4c1E0c1Q2c2.\displaystyle S_{Q}\sim\sqrt{4c_{1}E_{0}-{c_{1}Q^{2}\over c_{2}\ell}}. (59)

As one varies λ\lambda and α\alpha, the energy of the states changes according to (33), but the number of states remains unchanged. Thus the entropy remains independent of the couplings λ\lambda and α\alpha upon deformation. Substituting (58) in (59) one obtains

SQ4c1E0c1Q2c2=4c1(1αQ)E+c1E2hc1Q2c2.\displaystyle S_{Q}\sim\sqrt{4c_{1}E_{0}-{c_{1}Q^{2}\over c_{2}\ell}}=\sqrt{4c_{1}\left(1-{\alpha Q\over\sqrt{\ell}}\right)E+{c_{1}E^{2}\over h}-{c_{1}Q^{2}\over c_{2}\ell}}. (60)

An interesting fact is that the inverse Hagedorn temperature at fixed charge

βhQ=c1h,\displaystyle\beta^{Q}_{h}=\sqrt{{c_{1}\over h}}, (61)

is different from the inverse Hagedorn temperature (55). This is reminiscent of the fact that the grand canonical ensemble exhibits different Hagedorn behavior compared to fixed charge ensemble. Note that the fixed charge Hagedorn temperature goes to infinity in the limit hh\to\infty. The thermodynamics (at fixed charge) in this regime is somewhat intermediate between a local theory (i.e. theories with entropy that goes like E\sqrt{E}) and a theory with a Hagedorn density of states.

At energy EE, the leading contribution to the fixed charge entropy (60) comes from states with charge

Q=2c2αE.\displaystyle Q=-2c_{2}\alpha\sqrt{\ell}E. (62)

for which the entropy takes the form

SQ4c1E+c1h(1+4c2hα2)E2.\displaystyle S_{Q}\sim\sqrt{4c_{1}E+{c_{1}\over h}(1+4c_{2}h\alpha^{2})E^{2}}. (63)

This is in precise agreement with the second equation in (56) with the inverse Hagedorn temperature given by (55).

The particular thermodynamic systems that we are interested in, namely systems with linear specific heat (46) exhibit Hagedorn density of states at high energies. It is however important to stress that systems, which do not have linear specific heat (i.e. are not described by (46)), are not likely to exhibit Hagedorn growth. The density of states given in (57) is not a universal property of TT¯T\overline{T} and JT¯J\overline{T} deformations in d=1d=1; rather this is typical to systems with linear specific heat. This is an important difference between two-dimensional and one-dimensional TT¯+JT¯T\overline{T}+J\overline{T} deformation. In d=2d=2, the partition function of a generic CFT2, in the thermodynamic limit, exhibits linear specific heat. Thus, the Hagedorn growth of the density of states of a TT¯T\overline{T} deformed CFT2 at high energies is a universal property unlike d=1d=1.

4.3 Comments on thermodynamic stability

In this subsection, we are going to discuss about the thermodynamic stability of a quantum mechanical system with linear specific heat, upon deformation by a general linear combination of TT¯T\overline{T} and JT¯J\overline{T}. We will closely follow the argument in Barbon:2020amo . 141414We thank E. Rabinovici for raising the issue of thermodynamic stability which eventually lead to this subsection.

The temperature of a thermodynamic system at fixed charge is given by

1T(E)=SQE.\displaystyle{1\over T(E)}={\partial S_{Q}\over\partial E}. (64)

From the monotonicity property of T(E)T(E), one can determine the thermodynamic stability of such a thermodynamic system. The sign of TE{\partial T\over\partial E} determines the sign of the specific heat and systems with positive specific heat (i.e. TT is a monotonically increasing function of EE) are thermodynamically stable. Systems with negative specific heat (i.e. TT is a monotonically decreasing function of EE) however, are thermodynamically unstable. In the discussion that follows, we will use this concept to comment of the thermodynamic stability of the deformed system we are interested in.

We argued in subsubsection 4.2.2 that (see eq. (60))

SQ(E0(E))=SQ(E0).\displaystyle S_{Q}(E_{0}(E))=S_{Q}(E_{0}). (65)

Differentiating (65), with respect to the deformed energy EE, one can write

T(E)=E(E0)T0(E0),\displaystyle T(E)=E^{\prime}(E_{0})T_{0}(E_{0}), (66)

where E(E0)E^{\prime}(E_{0}) is the derivative of the deformed energy with respect to the undeformed one and T0(E0)=(SQ/E0)1T_{0}(E_{0})=(\partial S_{Q}/\partial E_{0})^{-1} is the temperature of the undeformed theory. For theories with linear specific heat (46) 151515For thermodynamic stability of systems with non-linear specific heat in the case of TT¯T\overline{T} deformation see Barbon:2020amo . It would be interesting to study stability of such systems in the case of TT¯+JT¯T\overline{T}+J\overline{T} deformation., the undeformed temperature takes the following form

T0(E0)=12c14c1E0c1Q2c2.\displaystyle T_{0}(E_{0})={1\over 2c_{1}}\sqrt{4c_{1}E_{0}-{c_{1}Q^{2}\over c_{2}\ell}}. (67)

Note that dT0dE0{dT_{0}\over dE_{0}} is always positive implying that the seed theory is thermodynamically stable. Substituting (67) in (66) and using the deformed spectrum (33), one can write

TE=Q2+4c2h(1αQ)24c24c1E0c1Q2c2(E0+h(1αQ)2).\displaystyle{\partial T\over\partial E}={Q^{2}+4c_{2}h\ell\left(1-\alpha{Q\over\sqrt{\ell}}\right)^{2}\over 4c_{2}\ell\sqrt{4c_{1}E_{0}-{c_{1}Q^{2}\over c_{2}\ell}}\left(E_{0}+h\left(1-\alpha{Q\over\sqrt{\ell}}\right)^{2}\right)}. (68)

It is easy to see that for h>0h>0, (i.e. for the regime in the parameter space where the deformed theory is unitary), TE>0{\partial T\over\partial E}>0 (remember that c2>0c_{2}>0). Thus the deformed theory has a positive specific heat implying that the theory is thermodynamically stable.

Next, let us discuss what happens for h<0h<0 where the theory is non-unitary. Assuming reality of (59) (i.e. assuming E0>Q24c2E_{0}>{Q^{2}\over 4c_{2}\ell}), the deformed system has positive specific heat for

0>h>Q24c2(1αQ)2 OR h<E0(1αQ)2,\displaystyle 0>h>-{Q^{2}\over 4c_{2}\ell\left(1-\alpha{Q\over\sqrt{\ell}}\right)^{2}}\ \ \ \text{ OR }\ \ \ h<-{E_{0}\over\left(1-\alpha{Q\over\sqrt{\ell}}\right)^{2}}, (69)

for a specified E0E_{0} and QQ. Thus, for h<0h<0, the system is thermodynamically stable provided hh satisfies (69). Note that, for non-unitary theories, the stability is highly state dependent.

5 UV completion

It was shown in Gross:2019ach that the theory defined by the flow equation (14) has a non-perturbative UV completion: quantum mechanics coupled to gravity in d=1d=1. The idea of such UV completion has its origin in Dubovsky:2017cnj ; Dubovsky:2018bmo in the case of TT¯T\overline{T} deformation in d=2d=2 and was further generalized in Aguilera-Damia:2019tpe for a deformation by a general linear combination of TT¯,JT¯T\overline{T},J\overline{T} and TJ¯T\overline{J} also in d=2d=2. In the discussion that follows, we will briefly describe the non-perturbative UV completion of TT¯T\overline{T} deformation in d=1d=1 discussed in Gross:2019ach and analogously propose a non-perturbative UV completion for a more general deformation of a quantum mechanical system defined by the flow equations (LABEL:ttbjtbfloweq).

5.1 TT¯T\overline{T} case

The UV completion proposed in Gross:2019ach is the following:

Z(λ,β)=DeDXDΦVdiffexp(S0[e,Φ]S[λ;e,X]),\displaystyle Z(\lambda,\beta)=\int{DeDXD\Phi\over V_{\rm{diff}}}~{}{\rm exp}\left(-S_{0}[e,\Phi]-S[\lambda;e,X]\right)~{}, (70)

where S0[e,Φ]S_{0}[e,\Phi] is the action of the seed theory which contains some matter fields, collectively denoted by Φ\Phi, coupled to one-dimensional gravity with einbein e(τ)e(\tau). The Euclidean time is compact ττ+β\tau\sim\tau+\beta^{\prime} with periodicity β\beta^{\prime}. The full deformed theory is invariant under reparametrization of τ\tau. Thus, in the path integral, one needs to divide by the volume of the group of diffeomorphisms denoted by VdiffV_{\rm{diff}}. The reparametrization invariant d=1d=1 gravity action is given by

S[λ;e,X]=12λ0β𝑑τe(X˙e1)2,\displaystyle S[\lambda;e,X]={1\over 2\lambda}\int_{0}^{\beta^{\prime}}d\tau e\left({\dot{X}\over e}-1\right)^{2}, (71)

where XX (with X˙τX\dot{X}\equiv\partial_{\tau}X) is a compact bosonic field with periodicity β\beta that satisfies X(τ+β)=X(τ)+mβX(\tau+\beta^{\prime})=X(\tau)+m\beta with mm\in\mathbb{Z} is the winding of XX along the τ\tau circle. On fixing the einbein gauge by introducing Faddeev-Popov ghosts and choosing the gauge e=1e=1 and finally dividing by the volume of the residual symmetry, the deformed partition function takes the following form:

Z(λ,β)=β2πλ0dββ3/2mexp((mββ)22βλ)Z0(β),\displaystyle Z(\lambda,\beta)={\beta\over\sqrt{2\pi\lambda}}\int_{0}^{\infty}{d\beta^{\prime}\over\beta^{\prime 3/2}}\sum_{m\in\mathbb{Z}}{\rm exp}\bigg{(}-{(m\beta-\beta^{\prime})^{2}\over 2\beta^{\prime}\lambda}\bigg{)}~{}Z_{0}(\beta^{\prime}), (72)

where the undeformed partition function Z0(β)Z_{0}(\beta^{\prime}) is given by

Z0(β)=DΦexp(S0[e=1,Φ]).\displaystyle Z_{0}(\beta^{\prime})=\int D\Phi~{}{\rm exp}\left(-S_{0}[e=1,\Phi]\right). (73)

The partition function (72) of the deformed theory in the unit winding sector matches exactly with the deformed partition function (22) and (23) obtained from the kernel analysis.

5.2 TT¯+JT¯T\overline{T}+J\overline{T} case

Following the discussion in the previous section, one may guess that the non-perturbative definition of a TT¯+JT¯T\overline{T}+J\overline{T} deformed quantum mechanical system governed by the flow equation (LABEL:ttbjtbfloweq) could be obtained by coupling the theory to gravity as well as a U(1)U(1) gauge field in d=1d=1. For deatailed discussion on gauge and gravity theory in d=1d=1 see Elitzur:1992bf . In the following discussion we will perform the path integral over the additional fields (i.e. the einbein, the gauge field and two compact real scalar fields) that will lead to the Kernel formula (40) and (41)

We propose

Z(λ,α,β)=DeDXDfDYDΦVdiff×Vgaugeexp(S0[e,f,Φ]S[λ,α;e,X,f,Y]),\displaystyle Z(\lambda,\alpha,\beta)=\int{DeDXDfDYD\Phi\over V_{\rm{diff}}\times V_{\rm{gauge}}}~{}{\rm exp}\left(-S_{0}[e,f,\Phi]-S[\lambda,\alpha;e,X,f,Y]\right)~{}, (74)

as the non-perturbative UV completion of the deformed theory in the most general case. The fields ee and XX are, as before, the einbein and the compact scalar field with periodicity β\beta, ff is the U(1)U(1) gauge field and YY is a compact scalar with periodicity μ\mu^{\prime} satisfying Y(τ+β)=Y(τ)+m~μY(\tau+\beta^{\prime})=Y(\tau)+\widetilde{m}\mu^{\prime} with m~\widetilde{m}\in\mathbb{Z}. The action of the seed theory coupled to gravity and gauge field is denoted by S0[e,f,Φ]S_{0}[e,f,\Phi]. This coupling is done in a way such that S0[e,f,Φ]S_{0}[e,f,\Phi] respects invariance under reparametrization in τ\tau and U(1)U(1) gauge symmetry. In fact, the full deformed theory should be invariant under reparametrization and gauge symmetry as a result of which, one needs to divide by the volume of the diffeomorphism (denoted by VdiffV_{\rm{diff}}) and gauge group (denoted by VgaugeV_{\rm{gauge}}). The reparametrization and gauge invariant d=1d=1 gravity plus gauge theory action is given by 161616Note that, pure gauge and gravity theory in d=1d=1 have negative degrees of freedom. To make the theories physically meaningful one needs to add the scalar fields XX and YY resulting in a theory with 0 degrees of freedom Elitzur:1992bf . Thus one may view the theory solely given by (75) as topological.

S[e,X,f,Y]=0βdτe[14hα2(Y˙f)21α(Y˙f)(X˙e)].\displaystyle S[e,X,f,Y]=\int_{0}^{\beta^{\prime}}{d\tau\over e}\left[-{1\over 4h\alpha^{2}}(\dot{Y}-f)^{2}-{1\over\alpha}(\dot{Y}-f)(\dot{X}-e)\right]. (75)

Under reparametrization ττ~\tau\to\widetilde{\tau}, the fields (e,X˙,f,Y˙)(e,\dot{X},f,\dot{Y}) transform as

e(τ)e~(τ~)=(ττ~)e(τ),τX(τ)τ~X~(τ~)=(ττ~)τX(τ),f(τ)f~(τ~)=(ττ~)f(τ),τY(τ)τ~Y~(τ~)=(ττ~)τY(τ),\displaystyle\begin{split}&e(\tau)\to\widetilde{e}(\widetilde{\tau})=\left({\partial\tau\over\partial\widetilde{\tau}}\right)e(\tau),\\ &\partial_{\tau}X(\tau)\to\partial_{\widetilde{\tau}}\widetilde{X}(\widetilde{\tau})=\left({\partial\tau\over\partial\widetilde{\tau}}\right)\partial_{\tau}X(\tau),\\ &f(\tau)\to\widetilde{f}(\widetilde{\tau})=\left({\partial\tau\over\partial\widetilde{\tau}}\right)f(\tau),\\ &\partial_{\tau}Y(\tau)\to\partial_{\widetilde{\tau}}\widetilde{Y}(\widetilde{\tau})=\left({\partial\tau\over\partial\widetilde{\tau}}\right)\partial_{\tau}Y(\tau),\\ \end{split} (76)

and under gauge transformation the fields (f,Y˙)(f,\dot{Y}) transform as

ff+g,Y˙Y˙+g,\displaystyle\begin{split}&f\to f+g,\\ &\dot{Y}\to\dot{Y}+g,\end{split} (77)

where gg is some arbitrary function of τ\tau. One can easily check that the action (75) is invariant under reparametrization symmetry (LABEL:rept) and gauge transformation (LABEL:gt).

Next let’s fix gauge. We set e=1e=1 with ττ+β\tau\sim\tau+\beta^{\prime}. This fixes almost all the reparametrization symmetry except constant shifts in τ\tau and ττ\tau\to-\tau symmetry. To fix this residual symmetry we will divide the path integral by the volume of the this residual gauge group which turns out to be 2β2\beta^{\prime}.171717The volume of the residual gauge symmetry is simply twice the circumference of the τ\tau circle. The 2 factor comes form ττ\tau\to-\tau symmetry. Similarly let’s choose f=0f=0 with βfβf+μ\beta^{\prime}f\sim\beta^{\prime}f+\mu^{\prime} to fix the U(1)U(1) gauge symmetry. As in the case of reparametrization symmetry, this gauge fixing keeps unfixed the symmetry due to constant shifts of the zero mode of ff and its 2\mathbb{Z}_{2} sign flip. To fix this residual gauge symmetry we divide by 2μ2\mu^{\prime}.181818It is easy to convince oneself from dimensional analysis that 2μ2\mu^{\prime} is indeed the appropriate factor to divide the path integral by in order to kill the residual gauge symmetry. For a more accurate treatment one may need to adapt the techniques used in string worldsheet theory to fix residual worldsheet gauge symmetry. This has been illuminated in footnote 3 of Gross:2019ach .

To fix the einbein gauge and the U(1)U(1) gauge symmetry we adopt the Faddeev-Popov procedure. The Faddeev-Popov measures for the reparametrization and U(1)U(1) gauge symmetry is defined as

1=ΔFP(e)0𝑑βDζδ(e1ζ),1=ΔFP(f)𝑑μDηδ(f0η),\displaystyle\begin{split}&1=\Delta_{FP}(e)\int_{0}^{\infty}d\beta^{\prime}\int D\zeta~{}\delta(e-1^{\zeta}),\\ &1=\Delta_{FP}(f)\int_{-\infty}^{\infty}d\mu^{\prime}\int D\eta~{}\delta(f-0^{\eta}),\end{split} (78)

where ζ\zeta is the diffeomorphism in τ\tau and 1ζ1^{\zeta} is a diffeomorphism transformation of the fudicial einbein e=1e=1 and similarly η\eta is the U(1)U(1) gauge transformation and 0η0^{\eta} is the gauge transformed fudicial gauge field f=0f=0. Next, we follow the usual procedure of writing the delta functions in (LABEL:fpd) as Fourier integrals and in introducing Grassmann fields ai,bia_{i},b_{i} and cic_{i} for i=1,2i=1,2 to invert the Faddeev-Popov measure. The Faddeev-Popov measures can be expressed as 191919For the ghost fields, we use the same normalization as in Gross:2019ach .

ΔFP(e)=0𝑑βDa1Db1Dc1exp(40β𝑑τ(b1c˙1a1b1/β)),ΔFP(f)=0𝑑βDa2Db2Dc2exp(40β𝑑τ(b2c˙2a2b2/β)).\displaystyle\begin{split}&\Delta_{FP}(e)=\int_{0}^{\infty}d\beta^{\prime}\int Da_{1}Db_{1}Dc_{1}~{}{\rm exp}\left(-4\int_{0}^{\beta^{\prime}}d\tau(b_{1}\dot{c}_{1}-a_{1}b_{1}/\beta^{\prime})\right),\\ &\Delta_{FP}(f)=\int_{0}^{\infty}d\beta^{\prime}\int Da_{2}Db_{2}Dc_{2}~{}{\rm exp}\left(-4\int_{0}^{\beta^{\prime}}d\tau(b_{2}\dot{c}_{2}-a_{2}b_{2}/\beta^{\prime})\right).\end{split} (79)

Next we substitute (LABEL:fpd) and (LABEL:fpg) in (74) and perform the path integrals over the aia_{i} ghost fields. This gives

Z(λ,α;β)=0𝑑β𝑑μi=12[DbiDci]DΦDXDYi=12(4β0β𝑑τbi)×exp(S0[e=1,f=0,Φ]S[λ,α;e=1,X,f=0,Y]4i=120βdτbic˙i).\displaystyle\begin{split}Z(\lambda,\alpha;\beta)&=\int_{0}^{\infty}d\beta^{\prime}\int_{-\infty}^{\infty}d\mu^{\prime}\int\prod_{i=1}^{2}[Db_{i}Dc_{i}]D\Phi DXDY\prod_{i=1}^{2}\left({4\over\beta^{\prime}}\int_{0}^{\beta^{\prime}}d\tau~{}b_{i}\right)\\ &\times{\rm exp}\left(-S_{0}[e=1,f=0,\Phi]-S[\lambda,\alpha;e=1,X,f=0,Y]-4\sum_{i=1}^{2}\int_{0}^{\beta^{\prime}}d\tau~{}b_{i}\dot{c}_{i}\right).\end{split} (80)

Fourier expansion of XX and YY fields are given by

X(τ)=τ(mββ)+1βn=exp(2πinτ/β)Xn,Y(τ)=τ(m~μβ)+1βn=exp(2πinτ/β)Yn,\displaystyle\begin{split}&X(\tau)=\tau\left({m\beta\over\beta^{\prime}}\right)+{1\over\sqrt{\beta^{\prime}}}\sum_{n=-\infty}^{\infty}{\rm exp}\left(2\pi in\tau/\beta^{\prime}\right)X_{n},\\ &Y(\tau)=\tau\left({\widetilde{m}\mu^{\prime}\over\beta^{\prime}}\right)+{1\over\sqrt{\beta^{\prime}}}\sum_{n=-\infty}^{\infty}{\rm exp}\left(2\pi in\tau/\beta^{\prime}\right)Y_{n},\end{split} (81)

where the reality condition on the fields XX and YY demands Xn=XnX_{n}=X_{-n}^{\ast} and Yn=YnY_{n}=Y_{-n}^{\ast} Since XX and YY are compact bosons, the zero modes X0X_{0} and Y0Y_{0} satisfies the periodicity condition X0X0+ββX_{0}\sim X_{0}+\sqrt{\beta^{\prime}}\beta and Y0Y0+βμY_{0}\sim Y_{0}+\sqrt{\beta^{\prime}}\mu^{\prime}. Next, we substitute the mode expansions (LABEL:xym) in the gauge fixed action (75), which gives

S[e=1,X,f=0,Y]=Sw+Sosc,\displaystyle S[e=1,X,f=0,Y]=S_{w}+S_{osc}, (82)

where SwS_{w} and SoscS_{osc} denote respectively the contribution from the winding mode and the oscillators given by

Sw=m~2μ24hα2βμαβ(mββ),Sosc=2π24hα2β2n=n2(YnYn+4hαYnXn).\displaystyle\begin{split}&S_{w}=-{\widetilde{m}^{2}\mu^{\prime 2}\over 4h\alpha^{2}\beta^{\prime}}-{\mu^{\prime}\over\alpha\beta^{\prime}}(m\beta-\beta^{\prime}),\\ &S_{osc}=-{2\pi^{2}\over 4h\alpha^{2}\beta^{\prime 2}}\sum_{n=-\infty}^{\infty}n^{2}\left(Y_{n}Y_{-n}+4h\alpha Y_{n}X_{-n}\right).\end{split} (83)

The path integral of the oscillator part of the action (LABEL:wosc) can be evaluated exactly:

DXDYexp(Sosc)=0ββ𝑑X00μβ𝑑Y0n=1(hα2β2πn2)n=1(β24πhn2).\displaystyle\int DXDY~{}{\rm exp}\left(-S_{osc}\right)=\int_{0}^{\beta\sqrt{\beta^{\prime}}}dX_{0}\int_{0}^{\mu^{\prime}\sqrt{\beta^{\prime}}}dY_{0}\prod_{n=1}^{\infty}\left(-{h\alpha^{2}\beta^{\prime 2}\over\pi n^{2}}\right)\prod_{n=1}^{\infty}\left({\beta^{\prime 2}\over 4\pi hn^{2}}\right). (84)

The integrals over the zero modes namely X0X_{0} and Y0Y_{0} are ordinary definite integrals and can easily be evaluated. The infinite products in (84) can be explicitly computed using zeta-function regularization techniques (see appendix B for details) giving rise to

DXDYexp(Sosc)=βμ2πiβα.\displaystyle\int DXDY~{}{\rm exp}\left(-S_{osc}\right)={\beta\mu^{\prime}\over 2\pi i\beta^{\prime}\alpha}. (85)

The path integral over the non-zero modes of the ghost fields are given by

DbiDciexp(40β𝑑τbic˙i)=n=18πnβ=β2.\displaystyle\int Db_{i}Dc_{i}~{}{\rm exp}\left(-4\int_{0}^{\beta^{\prime}}d\tau b_{i}\dot{c}_{i}\right)=\prod_{n=1}^{\infty}{8\pi n\over\beta^{\prime}}={\sqrt{\beta^{\prime}}\over 2}. (86)

In the ghost path integral, we have excluded the integrals over the zero modes of the cic_{i} ghost fields. The integrals of the zero modes of the cic_{i} ghost fields takes care of redundancies associated with the residual gauge symmetries. Since have excluded the integrals over the zero modes of the cic_{i} ghost fields, one needs to divide out by the volume of the residual gauge groups (both residual reparametrization symmetry and residual U(1)U(1) gauge symmetry).

Next, one needs to compute the integrals over the zero modes of the bib_{i} ghost fields. Since bib_{i} is periodic in τ\tau with periodicity β\beta^{\prime}, the integrated bib_{i} insertions in (80) picks out contribution only from the zero modes bi0b_{i}^{0}. Thus the contribution from the integral over the zero mode of bib_{i} ghost fields is given by

𝑑bi0(4β0β𝑑τbi)=4β𝑑bi0bi0=4β.\displaystyle\int db_{i}^{0}\left({4\over\beta^{\prime}}\int_{0}^{\beta^{\prime}}d\tau~{}b_{i}\right)={4\over\sqrt{\beta^{\prime}}}\int db_{i}^{0}b_{i}^{0}={4\over\sqrt{\beta^{\prime}}}. (87)

Thus putting together all the pieces and dividing by the volume of the residual symmetry groups one obtains

Z(λ,α;β)=β2πiα𝑑μ0dββ2m,m~exp(m~2μ24hα2β+m~μαβ(mββ))Z0(β,μ).\displaystyle Z(\lambda,\alpha;\beta)={\beta\over 2\pi i\alpha}\int_{-\infty}^{\infty}d\mu^{\prime}\int_{0}^{\infty}{d\beta^{\prime}\over\beta^{\prime 2}}\sum_{m,\widetilde{m}}{\rm exp}\left({\widetilde{m}^{2}\mu^{\prime 2}\over 4h\alpha^{2}\beta^{\prime}}+{\widetilde{m}\mu^{\prime}\over\alpha\beta^{\prime}}(m\beta-\beta^{\prime})\right)Z_{0}(\beta^{\prime},\mu^{\prime}).

Note that the above integral is strictly convergent for h<0h<0. For h>0h>0 we choose the μ\mu^{\prime}-contour to run along the imaginary μ\mu^{\prime} axis from i-i\infty to +i+i\infty. With this choice of the μ\mu^{\prime}-contour and for unit windings, m=m~=1m=\widetilde{m}=1, (5.2) agrees precisely with the kernel formula (40) with (41) at μ=0\mu=0. For non-zero μ\mu all one needs to do is to send μ\mu^{\prime} to μμ\mu^{\prime}-\mu in (5.2).

6 Discussion

The main goal of this paper is to continue the study of an analogue of TT¯T\overline{T} deformation of a d=1d=1 quantum mechanical system introduced in Gross:2019ach and further generalize such solvable deformations in d=1d=1 to those that have properties similar to JT¯J\overline{T} deformation and deformation by an arbitrary linear combination of TT¯T\overline{T} and JT¯J\overline{T} in d=2d=2. Unlike the study of TT¯T\overline{T} in quantum mechanics, which has its motivation from JT gravity in AdS2AdS_{2} with a Dirichlet wall at a finite radial distance, we draw motivation from the brunch cut singularity of the spectrum in d=2d=2 to define solvable deformations in quantum mechanics that can be thought of as an analogue of JT¯J\overline{T} and TT¯+JT¯T\overline{T}+J\overline{T} deformations in d=1d=1. We have shown that the thermal partition function of such deformed theories satisfy a flow equation whose solution gives the deformed partition function as an integral of the undeformed partition function weighted by some kernel. We argued that this can be realized as an alternative definition of the deformations valid even at quantum level.

The kernel formula has various interesting applications. It has been used to compute the deformed two-point functions. One intriguing observation is that in the presence of TT¯+JT¯T\overline{T}+J\overline{T} deformation, there is a regime (i.e. 2λ=α22\lambda=\alpha^{2}) in the parameter space where the two-point function looks undeformed. It’s likely that in this regime in the parameter space there is an enhancement of symmetries. We would like to leave the detailed analysis of the symmetries of the system in this limit as a future exercise. The kernel formula can further be used in analyzing the thermodynamics of the deformed system. We concluded that for a deformed quantum mechanical system with undeformed partition function given by (46), the microscopic origin of the states with Hagedorn density (55) arises from the charged states (62). We also studied the thermodynamic stability of such deformed systems and found that if is the deformed theory is unitary, it is also thermodynamically stable. For the regime in the parameter space where the deformed theory is non-unitary, thermodynamic stability is highly state dependent.

Our study of irrelevant deformations of one-dimensional quantum mechanics pave the way to various direction of future research, since the arena of TT¯T\overline{T} and JT¯J\overline{T} deformation in d=1d=1 is highly unexplored. An obvious extension of our work would be to extend the computation of deformed correlation functions in section 4 to higher point functions. We studied two-point function of operators of the deformed theory that do not depend on the couplings explicitly. It would be interesting to extend our analysis to those operators that have explicit dependence on the couplings e.g. it would be nice to calculate two-point functions of the stress tensor operator and the U(1)U(1) current.

Perhaps the most interesting extension of our work would be to understand the holographic realization of JT¯J\overline{T} deformation in quantum mechanics and its relation to charged SYK model. Given that the flow equation of the partition function of a JT¯J\overline{T} deformed quantum mechanical system is known, it would be interesting to understand, what kind of two-dimensional bulk gravity yields a similar flow equation. A possible way to proceed would be to understand JT¯J\overline{T} deformation of charged SYK model. This may lead us to some deformation of JT gravity (possibly with some matter fields turned on) with a cutoff and mixed boundary conditions, the flow equation of which resembles the flow equation of JT¯J\overline{T} deformed quantum mechanics. Eventually one may be interested in understanding the bulk dual upon turning on TT¯+JT¯T\overline{T}+J\overline{T}.

Other interesting open problems include understanding the entanglement structure of these deformed theories. One may also be interested in computing the deformed Lagrangian of the some simple quantum mechanical systems (e.g. single real boson with polynomial potential etc.). It would also be interesting to perform supersymmetric extensions of these deformations.

Acknowledgements

We would like to thank Eliezer Rabinovici for helpful comments on the manuscript. The work of SC and AM are supported by the Infosys Endowment for the study of the Quantum Structure of Spacetime.

Appendix A Entropy and partition function

In this appendix we are going to show that if the fixed charge thermal entropy is given by

S(E,Q)4c1Ec1Q2c2,S(E,Q)\sim\sqrt{4c_{1}E-{c_{1}Q^{2}\over c_{2}\ell}}, (89)

then this must correspond to a theory with grand canonical partition function given by (46).

The grand canonical partition function of a quantum mechanical system with U(1)U(1) global symmetry is given by

Z=𝑑E𝑑Qρ(E,Q)exp(βE+μQ/),Z=\int dE\int dQ~{}\rho(E,Q)~{}{\rm exp}\left(-\beta E+\mu Q/\sqrt{\ell}\right), (90)

where ρ(E,Q)=eS(E,Q)\rho(E,Q)=e^{S(E,Q)} is the density of sates at fixed energy EE and charge QQ. We perform the EE and QQ integrals by saddle point approximation. The saddles are located at

E=c1+c2μ2β2,Q=2c2μβ.\displaystyle\begin{split}&E={c_{1}+c_{2}\mu^{2}\over\beta^{2}},\\ &Q={2c_{2}\sqrt{\ell}\mu\over\beta}.\end{split} (91)

Substituting the saddles (LABEL:saddle) in (90) one obtains

Zexp(1β(c1+c2μ2)).Z\sim{\rm exp}\left({1\over\beta}(c_{1}+c_{2}\mu^{2})\right). (92)

Appendix B Zeta function regularization

For a positive non-decreasing infinite sequence 0<λ1λ20<\lambda_{1}\leq\lambda_{2}\leq\dots, the zeta-regularized product of the infinite sequence is defined as Mizuno

n=1λn=exp(ζλ(0)),\prod_{n=1}^{\infty}\lambda_{n}={\rm exp}\left(-\zeta^{\prime}_{\lambda}(0)\right), (93)

where

ζλ(s)=n=11λns\zeta_{\lambda}(s)=\sum_{n=1}^{\infty}~{}{1\over\lambda_{n}^{s}} (94)

is the generalized zeta function and ζλ(s)dζλ(s)/ds\zeta^{\prime}_{\lambda}(s)\equiv d\zeta_{\lambda}(s)/ds. In the case λn=n\lambda_{n}=n where n{1,2,}n\in\{1,2,\cdots\}, (94) reduces to the ordinary Riemann zeta function ζ(s)\zeta(s).

Using the zeta-regularized product formula (93), we would like to compute the following infinite product:

Ns=n=1(nsa),{N}_{s}=\prod_{n=1}^{\infty}({n}^{s}a), (95)

where aa is a c-number. Taking logarithm on either side of (95) one can write

logNs=ζ(0)loga+slog(n=1n).\log{N}_{s}=\zeta(0)\log a+s\log\left(\prod_{n=1}^{\infty}{n}\right). (96)

Next, we use the fact that ζ(0)=12\zeta(0)=-{1\over 2} and ζ(0)=12log2π\zeta^{\prime}(0)=-{1\over 2}\log 2\pi, and use (93) to write

n=1n=2π.\prod_{n=1}^{\infty}{n}=\sqrt{2\pi}. (97)

Substituting (97) in (96), one obtains

Ns=(2π)sa.N_{s}=\sqrt{{(2\pi)^{s}\over a}}. (98)

References

  • (1) F. Smirnov and A. Zamolodchikov, On space of integrable quantum field theories, Nucl. Phys. B 915 (2017) 363 [1608.05499].
  • (2) A. Cavaglià, S. Negro, I. M. Szécsényi and R. Tateo, TT¯T\overline{T}-deformed 2D Quantum Field Theories, JHEP 10 (2016) 112 [1608.05534].
  • (3) M. Guica, An integrable Lorentz-breaking deformation of two-dimensional CFTs, SciPost Phys. 5 (2018) 048 [1710.08415].
  • (4) S. Chakraborty, A. Giveon and D. Kutasov, JT¯J\overline{T} deformed CFT2 and string theory, JHEP 10 (2018) 057 [1806.09667].
  • (5) S. Chakraborty, A. Giveon and D. Kutasov, TT¯T\overline{T}, JT¯J\overline{T}, TJ¯T\overline{J} and String Theory, J. Phys. A 52 (2019) 384003 [1905.00051].
  • (6) B. Le Floch and M. Mezei, Solving a family of TT¯T\overline{T}-like theories, 1903.07606.
  • (7) L. McGough, M. Mezei and H. Verlinde, Moving the CFT into the bulk with TT¯T\overline{T}, JHEP 04 (2018) 010 [1611.03470].
  • (8) A. Giveon, N. Itzhaki and D. Kutasov, TT¯\mathrm{T}\overline{\mathrm{T}} and LST, JHEP 07 (2017) 122 [1701.05576].
  • (9) S. Chakraborty, A. Giveon and D. Kutasov, TT¯T\overline{T}, Black Holes and Negative Strings, 2006.13249.
  • (10) M. Guica and R. Monten, TT¯T\overline{T} and the mirage of a bulk cutoff, 1906.11251.
  • (11) G. Bonelli, N. Doroud and M. Zhu, TT¯T\overline{T}-deformations in closed form, JHEP 06 (2018) 149 [1804.10967].
  • (12) M. Taylor, TT deformations in general dimensions, 1805.10287.
  • (13) T. Hartman, J. Kruthoff, E. Shaghoulian and A. Tajdini, Holography at finite cutoff with a T2T^{2} deformation, JHEP 03 (2019) 004 [1807.11401].
  • (14) A. Belin, A. Lewkowycz and G. Sarosi, Gravitational path integral from the T2T^{2} deformation, 2006.01835.
  • (15) A. B. Zamolodchikov, Expectation value of composite field T anti-T in two-dimensional quantum field theory, hep-th/0401146.
  • (16) D. J. Gross, J. Kruthoff, A. Rolph and E. Shaghoulian, TT¯T\overline{T} in AdS2 and Quantum Mechanics, Phys. Rev. D 101 (2020) 026011 [1907.04873].
  • (17) D. J. Gross, J. Kruthoff, A. Rolph and E. Shaghoulian, Hamiltonian deformations in quantum mechanics, TT¯T\overline{T}, and SYK, 1912.06132.
  • (18) L. V. Iliesiu, J. Kruthoff, G. J. Turiaci and H. Verlinde, JT gravity at finite cutoff, 2004.07242.
  • (19) J. Barbon and E. Rabinovici, Remarks on the thermodynamic stability of TT-bar deformations, 2004.10138.
  • (20) S. Chakraborty and A. Hashimoto, Thermodynamics of TT¯T\overline{T}, JT¯J\overline{T}, TJ¯T\overline{J} deformed conformal field theories, 2006.10271.
  • (21) D. Stanford and E. Witten, Fermionic Localization of the Schwarzian Theory, JHEP 10 (2017) 008 [1703.04612].
  • (22) S. Dubovsky, V. Gorbenko and G. Hernández-Chifflet, TT¯T\overline{T} partition function from topological gravity, JHEP 09 (2018) 158 [1805.07386].
  • (23) V. de Alfaro, S. Fubini and G. Furlan, Conformal Invariance in Field Theory, in 2nd Conference on Differential Geometrical Methods in Mathematical Physics., pp. 255–293, 1, 1977.
  • (24) C. Chamon, R. Jackiw, S.-Y. Pi and L. Santos, Conformal quantum mechanics as the CFT1 dual to AdS2, Phys. Lett. B 701 (2011) 503 [1106.0726].
  • (25) S. Dubovsky, V. Gorbenko and M. Mirbabayi, Asymptotic fragility, near AdS2 holography and TT¯T\overline{T}, JHEP 09 (2017) 136 [1706.06604].
  • (26) J. Aguilera-Damia, V. I. Giraldo-Rivera, E. A. Mazenc, I. Salazar Landea and R. M. Soni, A Path Integral Realization of Joint JT¯J\overline{T}, TJ¯T\overline{J} and TT¯T\overline{T} Flows, 1910.06675.
  • (27) S. Elitzur, A. Forge and E. Rabinovici, Comments on the importance of being overconstrained, Phys. Lett. B 289 (1992) 45.
  • (28) Y. Mizuno, Generalized Lerch formulas: Examples of zeta-regularized products, Journal of Number Theory 118 (2006) 155.