betaFeedback

WorksPhysics and Astronomy

Interacting Boson System at Finite Temperatures: The treatment of the lattice QCD results

Abstract

We study an interacting relativistic system of charged bosons at finite temperature and fixed isospin density within a thermodynamically consistent mean-field approach. The interaction is represented by repulsive phi-4 and phi-6 self-interactions, which generate a density-dependent effective mass of the bosonic quasiparticles. We formulate the thermodynamics in the Extended Canonical Ensemble, obtained from the grand canonical description by a Legendre transformation from the isospin chemical potential to the conserved isospin density. In the condensed phase the chemical potential is constrained by the Bose-condensation condition and is therefore determined self-consistently through the effective mass. We apply the formalism to a pion-like system at finite isospin and clarify the relation between the isospin chemical-potential conventions used for pion and quark degrees of freedom. Using the convention adopted in lattice-QCD calculations, we compare the pressure, isospin density, energy density, and trace anomaly with lattice results at T=120 MeV. With one constant fitted to the lattice pressure, each model reproduces the lattice isospin density, energy density, and trace anomaly. Both the phi-4 and phi-6 models reproduce the main features of the lattice data, with the phi-6 interaction providing the better description. The Extended Canonical Ensemble provides a convenient framework for treating interacting particle-antiparticle systems across the transition to the Bose-condensed phase.

Introduction

The thermodynamic properties of strongly interacting matter at finite temperature and density are relevant for relativistic nucleus–nucleus collisions and for the study of QCD matter under extreme conditions. At sufficiently high temperature and density, QCD-motivated effective models and lattice simulations predict substantial modifications of hadronic properties associated with chiral symmetry restoration and deconfinement. In the hadronic phase, pions constitute the lightest degrees of freedom and, because they are bosons carrying isospin, their system can undergo Bose–Einstein condensation at sufficiently large isospin density. Recent years the subject of the asymmetric isospin systems attract a valuable attention, see Refs. [1, 2, 3].

The thermodynamics of interacting bosons differs essentially from that of an ideal gas. In particular, the interaction modifies the quasiparticle mass and therefore affects simultaneously the particle density, pressure, energy density, and the condition for Bose condensation. A thermodynamically consistent mean-field treatment of interacting scalar bosons with repulsive self-interactions was developed in Refs. [4, 5]. In this approach the interaction generates a scalar-density dependent effective mass and an excess-pressure contribution, whose relation guarantees thermodynamic consistency.

In the present work we investigate an interacting relativistic particle–antiparticle system at finite temperature and fixed isospin density. We consider repulsive φ4\varphi^{4} and φ6\varphi^{6} interactions and formulate the thermodynamics within the Extended Canonical Ensemble (ECE). The ECE is obtained from the grand canonical description by a Legendre transformation in which the conserved isospin density, rather than the isospin chemical potential, is taken as the independent variable. This formulation is convenient in the condensed phase, where the chemical potential is no longer an independent variable but is constrained by the condensation condition.

Finite isospin density is one of the few regimes of dense QCD free of the fermion sign problem and therefore directly accessible to lattice Monte Carlo simulations [1]. An additional issue arises when comparing the pion description with lattice-QCD calculations. Different normalizations of the isospin chemical potential are used in the literature. For the pion degrees of freedom, consistency of the isospin number and the corresponding chemical potential gives

μI(π)=μI(q)=μu−μd.(1)\mu_{I}^{(\pi)} = \mu_{I}^{(q)} = \mu_{u} - \mu_{d}. \tag*{(1)}

With this convention the condensation condition is μI=mπ\mu_{I}=m_{\pi} for an ideal pion gas. Lattice calculations commonly employ instead the convention (see Refs. [6, 7, 8, 9, 10])

μIlat=12(μu−μd) ,(2)\mu_{I}^{\mathrm{lat}} = \frac{1}{2}(\mu_{u} - \mu_{d}) \, , \tag*{(2)}

for which the same physical condensation condition is expressed as μIlat=mπ/2\mu_{I}^{\mathrm{lat}} = m_{\pi}/2. We shall use the latter convention when comparing our results with the lattice data [9, 10].

The purpose of this work is therefore twofold. First, we formulate a self-consistent description of the thermal and Bose-condensed phases of an interacting relativistic boson system at fixed isospin density. Second, we apply the resulting equations of state to finite-isospin lattice-QCD data and compare the performance of the φ4\varphi^{4} and φ6\varphi^{6} mean-field models. In particular, we investigate the pressure, isospin density, energy density, and trace anomaly and determine which interaction provides the more adequate description of the lattice results.

A systematic low-energy description of QCD at finite isospin density is provided by chiral perturbation theory, which predicts the onset of pion condensation and the equation of state close to the threshold at low temperatures [1, 2, 11, 12]. The present approach is complementary to it: it is not a systematic low-energy expansion, but a thermodynamically consistent mean-field model with a single interaction constant, applied at finite temperature across the Bose-condensation transition.

Another widely used description of hadronic matter is the hadron resonance gas model [13, 14, 15], which includes the full spectrum of hadrons and resonances. Here we deliberately keep the description minimal and include only the charged pion-like degrees of freedom built of uu and dd quarks. Heavier states that couple to the isospin chemical potential, such as the charged ρ\rho mesons and the kaons, which are present in the 2+12+1-flavor lattice simulations, are neglected: at the considered temperature their thermal contributions are suppressed by their large masses and are expected to be small compared with the pion contribution.

Interacting scalar mesons: The mean-field approximation

Let us start with the Lagrangian density for scalar-boson charged field ϕ^(x)\hat{\phi}(x)

L(x)=L0+Lint=∂μϕ^†(x) ∂μϕ^(x)−m2ϕ^†(x)ϕ^(x)+Lint ,(3)\mathcal{L}(x) = \mathcal{L}_{0} + \mathcal{L}_{\mathrm{int}} = \partial_{\mu}\hat{\phi}^{\dagger}(x)\,\partial^{\mu}\hat{\phi}(x) - m^{2}\hat{\phi}^{\dagger}(x)\hat{\phi}(x) + \mathcal{L}_{\mathrm{int}} \, , \tag*{(3)}

where L0\mathcal{L}_{0} is a free part of the Lagrangian density. The complex scalar field describes two charged states, a particle and an antiparticle, which for the pion system are identified with π+\pi^{+} and π−\pi^{-}; the neutral pion, which does not couple to the isospin chemical potential, is not included. Here x=(t,r)x=(t,\boldsymbol{r}) is the time-space four-vector and ∂μ∂μ=∂2/∂t2−∇2\partial_{\mu}\partial^{\mu}=\partial^{2}/\partial t^{2}-\nabla^{2}.

In the present work we consider the ϕ4\phi^4 and ϕ6\phi^6 models that realize a selfinteracting repulsion between particles in the system, i.e. we assume the following interaction Lagrangian densities, respectively,

Lint(σ^)=−λ2[ϕ^†(x)ϕ^(x)]2,orLint(σ^)=−b3[ϕ^†(x)ϕ^(x)]3,(4)\mathcal{L}_{\mathrm{int}}(\hat{\sigma})=-\frac{\lambda}{2}\left[\hat{\phi}^{\dagger}(x)\hat{\phi}(x)\right]^2,\qquad\text{or}\qquad\mathcal{L}_{\mathrm{int}}(\hat{\sigma})=-\frac{b}{3}\left[\hat{\phi}^{\dagger}(x)\hat{\phi}(x)\right]^3, \tag*{(4)}

where the interaction constants are positive, λ≥0\lambda\ge0, b≥0b \ge0. The repulsion between pions is what stabilizes the condensate at a finite isospin density [1]; here this mechanism is implemented in a thermodynamically consistent mean-field form. Next, we consider the mean-field approximation expanding the interaction Lagrangian around the mean value σ\sigma defined as:

σ=⟨ϕ^†(x)ϕ^(x)⟩,(5)\sigma=\left\langle\hat{\phi}^{\dagger}(x)\hat{\phi}(x)\right\rangle, \tag*{(5)}

where ⟨…⟩\langle\ldots\rangle means a thermal averaging. Indeed, one can use the interaction Lagrangian density in approximate way taking just two first terms of the expansion of Lint(σ^)\mathcal{L}_{\mathrm{int}}(\hat{\sigma}) around the mean value, see [4]. Following this procedure we define

Lint(σ^)≈LintMF(σ^)=Lint(σ)+∂Lint(σ)∂σ(σ^−σ),(6)\mathcal{L}_{\mathrm{int}}(\hat{\sigma})\approx\mathcal{L}_{\mathrm{int}}^{\mathrm{MF}}(\hat{\sigma})=\mathcal{L}_{\mathrm{int}}(\sigma)+\frac{\partial\mathcal{L}_{\mathrm{int}}(\sigma)}{\partial\sigma}(\hat{\sigma}-\sigma), \tag*{(6)}

where σ^=ϕ^†(x)ϕ^(x)\hat{\sigma}=\hat{\phi}^{\dagger}(x)\hat{\phi}(x) is the operator of the scalar density.

The mean-field approximation

Let us consider the mean-field (MF) Lagrangian density

L≈LMF=L0+LintMF,(7)\mathcal{L}\approx\mathcal{L}_{\mathrm{MF}}=\mathcal{L}_0+\mathcal{L}_{\mathrm{int}}^{\mathrm{MF}}, \tag*{(7)}

where L0\mathcal{L}_0 is defined in eq. (3) and LintMF\mathcal{L}_{\mathrm{int}}^{\mathrm{MF}} is defined in eq. (6). In accordance with these definitions we get

LMF=∂μϕ^†(x) ∂μϕ^(x)−M2(σ)ϕ^†(x)ϕ^(x)+Pex(σ),(8)\mathcal{L}_{\mathrm{MF}}=\partial_{\mu}\hat{\phi}^{\dagger}(x)\,\partial^{\mu}\hat{\phi}(x)-M^2(\sigma)\hat{\phi}^{\dagger}(x)\hat{\phi}(x)+P_{\mathrm{ex}}(\sigma), \tag*{(8)}

where

M2(σ)=m2+U(σ),U(σ)=−∂Lint(σ)∂σ,Pex(σ)=Lint(σ)−σ∂Lint(σ)∂σ.(9)M^2(\sigma)=m^2+U(\sigma),\qquad U(\sigma)=-\frac{\partial\mathcal{L}_{\mathrm{int}}(\sigma)}{\partial\sigma},\qquad P_{\mathrm{ex}}(\sigma)=\mathcal{L}_{\mathrm{int}}(\sigma)-\sigma\frac{\partial\mathcal{L}_{\mathrm{int}}(\sigma)}{\partial\sigma}. \tag*{(9)}

and

Pex(σ)=Lint(σ)−σ∂Lint(σ)∂σ.(10)P_{\mathrm{ex}}(\sigma)=\mathcal{L}_{\mathrm{int}}(\sigma)-\sigma\frac{\partial\mathcal{L}_{\mathrm{int}}(\sigma)}{\partial\sigma}. \tag*{(10)}

As it is seen from eq. (8), in the MF approximation the scalar field “earns” an effective mass M(σ)M(\sigma) and, as it will be seen further, the quantity PexP_{\mathrm{ex}} plays a role of an excess pressure. It is easy to check that in accordance with definitions (9) and (10) we get the following relation: σ(∂U/∂σ)=∂Pex/∂σ\sigma(\partial U/\partial\sigma)=\partial P_{\mathrm{ex}}/\partial\sigma, which is a necessary attribute to provide the thermodynamic consistency of the model. For the two interactions considered here, U(σ)=λσU(\sigma)=\lambda\sigma for the φ4\varphi^4 model and U(σ)=bσ2U(\sigma)=b\sigma^2 for the φ6\varphi^6 model.

The condensed phase

In what follows we use the Bogolyubov’s decomposition of the boson field operator [16]

φ^=Φ0+ψ^with⟨ψ^⟩=0,(11)\hat{\varphi}=\Phi_0+\hat{\psi}\qquad\text{with}\qquad\langle\hat{\psi}\rangle=0, \tag*{(11)}

where Φ0\Phi_0 is the wave function which is responsible for the condensate creation in the system and ψ^\hat{\psi} is the field operator that describes the thermal (kinetic) particles.

The field operators ϕ^(x)\hat{\phi}(x) and π^(x)=∂L/∂(∂0ϕ^)=∂ϕ^†(x)/∂x0\hat{\pi}(x)=\partial\mathcal{L}/\partial(\partial_0\hat{\phi})=\partial\hat{\phi}^{\dagger}(x)/\partial x^0 satisfy the following equal-time commutation relations

[ϕ^(t,r),π^(t,r′)]=iδ3(r−r′)⇒[ψ^(t,r),∂ψ^†(t,r′)∂t]=iδ3(r−r′).(12)\left[\hat{\phi}(t,\mathbf{r}),\hat{\pi}(t,\mathbf{r}')\right]=i\delta^3(\mathbf{r}-\mathbf{r}')\quad\Rightarrow\quad\left[\hat{\psi}(t,\mathbf{r}),\frac{\partial\hat{\psi}^{\dagger}(t,\mathbf{r}')}{\partial t}\right]=i\delta^3(\mathbf{r}-\mathbf{r}'). \tag*{(12)}

Hence, the thermal field ψ^(x)\hat{\psi}(x) satisfies the same equal-time commutation relations as the total scalar field ϕ^(x)\hat{\phi}(x). Using this we can write the Bogolyubov’s decomposition in the relativistic case as

ϕ^(t,r)=Φ0(t)+∫p≠0d3p(2π)32ωp[ape−iωpt+bp†eiωpt]eip⋅r,(13)\hat{\phi}(t,\mathbf{r})=\Phi_0(t)+\int_{p\ne0}\frac{d^3p}{(2\pi)^3 2\omega_p}\left[a_p e^{-i\omega_p t}+b_p^\dagger e^{i\omega_p t}\right]e^{i\mathbf{p}\cdot\mathbf{r}}, \tag*{(13)}

where ωp=M2+p2\omega_p=\sqrt{M^2+\mathbf{p}^2} and we assume that the condensate medium is spatially homogeneous.

Then, in accordance with equal-time commutation relations (12), the boson annihilation and creation operators, aka_k, ak†a_k^\dagger and bkb_k, bk†b_k^\dagger, satisfy the commutation relations

[ak,ak′†]=(2π)32ωkδ3(k−k′),[bk,bk′†]=(2π)32ωkδ3(k−k′).(14)\left[a_k,a_{k'}^\dagger\right]=(2\pi)^3 2\omega_k\delta^3(\mathbf{k}-\mathbf{k}'),\qquad\left[b_k,b_{k'}^\dagger\right]=(2\pi)^3 2\omega_k\delta^3(\mathbf{k}-\mathbf{k}'). \tag*{(14)}

In what follows we investigate the bosonic systems in thermodynamic limit V→∞V\to\infty, that is instructive to fix the following relations

∫d3r ψ^(t,r)=0,∫d3r ∂μψ^(t,r)=0.(15)\int d^3r\,\hat{\psi}(t,\mathbf{r})=0,\qquad\int d^3r\,\partial^\mu\hat{\psi}(t,\mathbf{r})=0. \tag*{(15)}

With account to these relations we integrate the MF Lagrangian density (8) over the spatial volume occupied by the system, LMF(t)=∫d3r LMF(t,r)L_{\mathrm{MF}}(t)=\int d^3r\,\mathcal{L}_{\mathrm{MF}}(t,\mathbf{r}), and obtain

LMF(t)=V[∂tΦ0†∂tΦ0−M2Φ0†Φ0]+∫d3r[∂μψ^†(x)∂μψ^(x)−M2ψ^†(x)ψ^(x)]+VPex(σ).(16)L_{\mathrm{MF}}(t)=V\left[\partial_t\Phi_0^\dagger\partial_t\Phi_0-M^2\Phi_0^\dagger\Phi_0\right]+\int d^3r\left[\partial_\mu\hat{\psi}^\dagger(x)\partial^\mu\hat{\psi}(x)-M^2\hat{\psi}^\dagger(x)\hat{\psi}(x)\right]+VP_{\mathrm{ex}}(\sigma). \tag*{(16)}

All terms in (8) that are linear on the thermal scalar field ψ^\hat{\psi} disappear after integration over the spatial volume in accordance with eqs. (15). However, it will be not true in case if the scalar condensate field Φ0\Phi_0 depends on spatial coordinates (these statements can be also found in Ref. [17]).

The action is a functional of two fields S[Φ0,ψ^]=∫dtLMF(t;σ)S[\Phi_0,\hat{\psi}] = \int dt L_{\mathrm{MF}}(t;\sigma). Using the MF Lagrangian (16) one can get equations of motion for the condensate field Φ0\Phi_0 and thermal field ψ^\hat{\psi} by minimizing the action

[∂0∂0+M2(σ)]Φ0=0,(17)\left[\partial_0\partial^0 + M^2(\sigma)\right]\Phi_0 = 0, \tag*{(17)}
[∂μ∂μ+M2(σ)]ψ^(x)=0.(18)\left[\partial_\mu\partial^\mu+ M^2(\sigma)\right]\hat{\psi}(x) = 0. \tag*{(18)}

There are two solutions of eq. (17). The first one is trivial Φ0=0\Phi_0 = 0, it corresponds to an absence of the condensate in the system. The second solution indicates the presence of the condensate

Φ0=e−iMt2Mψ0(+)+eiMt2M(ψ0(−))∗.(19)\Phi_0 = \frac{e^{-iMt}}{\sqrt{2M}}\psi_0^{(+)} + \frac{e^{iMt}}{\sqrt{2M}}\left(\psi_0^{(-)}\right)^*. \tag*{(19)}

We would like to stress that ψ0(+)\psi_0^{(+)} and ψ0(−)\psi_0^{(-)} are the nonrelativistic wave functions attributed to particles and antiparticles, which are in rest.

In accordance with the MF Lagrangian (16) the effective Hamiltonian in the MF approach HMF(σ)=∫d3rHMF(x,σ)H_{\mathrm{MF}}(\sigma) = \int d^3r H_{\mathrm{MF}}(x,\sigma) has a quasi-free form

HMF=V[∂tΦ0∗∂tΦ0+M2(σ)Φ0∗Φ0]+∫d3r[∂tψ^†∂tψ^+∇ψ^†⋅∇ψ^+M2(σ)ψ^†ψ^]−VPex(σ).(20)H_{\mathrm{MF}} = V\left[\partial_t\Phi_0^*\partial_t\Phi_0 + M^2(\sigma)\Phi_0^*\Phi_0\right] + \int d^3r\left[\partial_t\hat{\psi}^\dagger\partial_t\hat{\psi} + \nabla\hat{\psi}^\dagger\cdot\nabla\hat{\psi} + M^2(\sigma)\hat{\psi}^\dagger\hat{\psi}\right] - V P_{\mathrm{ex}}(\sigma). \tag*{(20)}

Applying solution (19), we obtain the condensate contribution to the MF Hamiltonian:

∂0Φ0∗∂0Φ0+M2Φ0∗Φ0=M[(ψ0(+))∗ψ0(+)+(ψ0(−))∗ψ0(−)]=Mncond,(21)\partial_0\Phi_0^*\partial_0\Phi_0 + M^2\Phi_0^*\Phi_0 = M\left[\left(\psi_0^{(+)}\right)^*\psi_0^{(+)} + \left(\psi_0^{(-)}\right)^*\psi_0^{(-)}\right] = M n_{\mathrm{cond}}, \tag*{(21)}

where the condensate particle-number density ncondn_{\mathrm{cond}} reads:

ncond=(ψ0(+))∗ψ0(+)+(ψ0(−))∗ψ0(−).(22)n_{\mathrm{cond}} = \left(\psi_0^{(+)}\right)^*\psi_0^{(+)} + \left(\psi_0^{(-)}\right)^*\psi_0^{(-)}. \tag*{(22)}

In correspondence to eq. (11) the total scalar density σ=⟨φ^†φ^⟩\sigma= \langle\hat{\varphi}^\dagger\hat{\varphi}\rangle also can be decomposed as:

σ=σcond+σth,whereσcond=Φ0∗Φ0,σth=⟨ψ^†(x)ψ^(x)⟩.(23)\sigma= \sigma_{\mathrm{cond}} + \sigma_{\mathrm{th}}, \qquad\text{where} \qquad\sigma_{\mathrm{cond}} = \Phi_0^*\Phi_0,\qquad\sigma_{\mathrm{th}} = \langle\hat{\psi}^\dagger(x)\hat{\psi}(x)\rangle. \tag*{(23)}

The connection between the condensate scalar density and the condensate particle-number density is given by

σcond=12M[(ψ0(+))∗ψ0(+)+(ψ0(−))∗ψ0(−)]=ncond2M.(24)\sigma_{\mathrm{cond}} = \frac{1}{2M}\left[\left(\psi_0^{(+)}\right)^*\psi_0^{(+)} + \left(\psi_0^{(-)}\right)^*\psi_0^{(-)}\right] = \frac{n_{\mathrm{cond}}}{2M}. \tag*{(24)}

Then, one can represent the MF Hamiltonian (20) in terms of creation and annihilation operators as:

HMF(σ)=MVncond+12∫d3k(2π)3[a†(k)a(k)+b†(k)b(k)]−VPex(σ).(25)H_{\mathrm{MF}}(\sigma) = MV n_{\mathrm{cond}} + \frac{1}{2} \int\frac{d^3 k}{(2\pi)^3} \left[a^\dagger(\mathbf{k})a(\mathbf{k}) + b^\dagger(\mathbf{k})b(\mathbf{k})\right] - V P_{\mathrm{ex}}(\sigma). \tag*{(25)}

Averaging of the MF Hamiltonian (20) gives the energy density of the system, ε=⟨H⟩/V\varepsilon= \langle H\rangle/V:

ε=Mncond+1V∫d3r⟨∂tψ^†∂tψ^+∇ψ^†⋅∇ψ^⟩+m2σth+U(σ)σth−Pex(σ).(26)\varepsilon= M n_{\mathrm{cond}} + \frac{1}{V} \int d^3 r \left\langle\partial_t \hat{\psi}^\dagger\partial_t \hat{\psi} + \nabla\hat{\psi}^\dagger\cdot\nabla\hat{\psi} \right\rangle+ m^2 \sigma_{\mathrm{th}} + U(\sigma)\sigma_{\mathrm{th}} - P_{\mathrm{ex}}(\sigma). \tag*{(26)}

Isospin density in bosonic system

The Noether current, which guarantees the isospin conservation, in the operator form reads j^μ(x)=i[ϕ^†(x)∂μϕ^(x)−(∂μϕ^†(x))ϕ^(x)]\hat{j}^{\mu}(x) = i\left[\hat{\phi}^{\dagger}(x)\partial^{\mu}\hat{\phi}(x) - \left(\partial^{\mu}\hat{\phi}^{\dagger}(x)\right)\hat{\phi}(x)\right]. The operator N^I\hat{N}_{I} of the total isospin number of the system is obtained as an integral of the zero component j^0\hat{j}^{0} of the Noether current over the volume of the system:

N^I(t,r)=i∫d3r[ϕ^†(t,r)∂tϕ^(t,r)−(∂tϕ^†(t,r))ϕ^(t,r)].(27)\hat{N}_{I}(t,\mathbf{r}) = i \int d^3 r \left[\hat{\phi}^{\dagger}(t,\mathbf{r})\partial_t\hat{\phi}(t,\mathbf{r}) - \left(\partial_t\hat{\phi}^{\dagger}(t,\mathbf{r})\right)\hat{\phi}(t,\mathbf{r})\right]. \tag*{(27)}

Using decomposition ϕ^=Φ0+ψ^\hat{\phi} = \Phi_0 + \hat{\psi} one can show that in a homogeneous system the isospin-number operator N^I\hat{N}_{I} consists of the condensate NIcondN_{I\mathrm{cond}} and thermal N^Ith\hat{N}_{I\mathrm{th}} contributions

N^I=NIcond+N^Ith,(28)\hat{N}_{I} = N_{I\mathrm{cond}} + \hat{N}_{I\mathrm{th}}, \tag*{(28)}

where using solution (19) we get

NIcond=V[(ψ0(+))∗ψ0(+)−(ψ0(−))∗ψ0(−)]=VnIcond.(29)N_{I\mathrm{cond}} = V\left[\left(\psi_0^{(+)}\right)^* \psi_0^{(+)} - \left(\psi_0^{(-)}\right)^* \psi_0^{(-)}\right] = V n_{I\mathrm{cond}}. \tag*{(29)}

(It is instructive to compare the expression obtained for nIcondn_{I\mathrm{cond}} with ncondn_{\mathrm{cond}} given in eq. (22).) The thermal contribution N^Ith\hat{N}_{I\mathrm{th}} to the isospin density in eq. (28) is given by

N^Ith=N^th(+)−N^th(−),whereN^th(+)=∫d3k(2π)32ωkak†ak,N^th(−)=∫d3k(2π)32ωkbk†bk.(30)\hat{N}_{I\mathrm{th}} = \hat{N}_{\mathrm{th}}^{(+)} - \hat{N}_{\mathrm{th}}^{(-)}, \quad\text{where} \quad\hat{N}_{\mathrm{th}}^{(+)} = \int\frac{d^3 k}{(2\pi)^3 2\omega_k} a_k^\dagger a_k, \quad\hat{N}_{\mathrm{th}}^{(-)} = \int\frac{d^3 k}{(2\pi)^3 2\omega_k} b_k^\dagger b_k. \tag*{(30)}

Bosonic system of particles and antiparticles within the Extended Canonical Ensemble

We are going to describe thermodynamic properties of the boson system of particles and antiparticles in the framework of the Extended Canonical Ensemble (ECE) that is obtained after movement from the Grand Canonical Ensemble (GCE) to Canonical Ensemble (CE) by means of the Legendre transformation, see Ref. [5]. Our starting point is the partition function in the GCE, which in the thermal phase and in the condensed phase is given by, respectively,

Z(T,μI,V)=eβVPex(σ)Tr⁡{e−β[H^MF−μIN^I]th}(31)Z(T,\mu_I,V)=e^{\beta V P_{\mathrm{ex}}(\sigma)}\operatorname{Tr}\left\{e^{-\beta[\hat{H}_{\mathrm{MF}}-\mu_I\hat{N}_I]_{\mathrm{th}}}\right\} \tag*{(31)}
Z(T,μI,V)=e−β[Ncond(M−μI)−VPex(σ)]Tr⁡{e−β[H^MF−μIN^I]th}(32)Z(T,\mu_I,V)=e^{-\beta[N_{\mathrm{cond}}(M-\mu_I)-V P_{\mathrm{ex}}(\sigma)]}\operatorname{Tr}\left\{e^{-\beta[\hat{H}_{\mathrm{MF}}-\mu_I\hat{N}_I]_{\mathrm{th}}}\right\} \tag*{(32)}

where we introduce notation for thermal quantities:

[H^MF−μIN^I]th=∫d3k(2π)32ωk[(ωk−μI)ak†ak+(ωk+μI)bk†bk].(33)\left[\hat{H}_{\mathrm{MF}}-\mu_I\hat{N}_I\right]_{\mathrm{th}}=\int\frac{d^3k}{(2\pi)^3 2\omega_k}\left[(\omega_k-\mu_I)a_k^\dagger a_k+(\omega_k+\mu_I)b_k^\dagger b_k\right]. \tag*{(33)}

Note, the C-number contributions to the Hamiltonian (25) and particle-number operator (28) associated with the presence of the condensate do not affect the mean value of some operator A^\hat{A} in comparison with the statistical averaging of this operator in the thermal phase.

After calculation of the partition function one can write the thermodynamic potential Ω=−Tln⁡Z(T,μI;σ)\Omega=-T\ln Z(T,\mu_I;\sigma) in a homogeneous thermal system and in condensate system, respectively,

Ω(T,μI,V)=Ωth(T,μI,V)−VPex(σ),thermal phase(34)\Omega(T,\mu_I,V)=\Omega_{\mathrm{th}}(T,\mu_I,V)-VP_{\mathrm{ex}}(\sigma),\qquad\text{thermal phase} \tag*{(34)}
Ω(T,μI,V)=Ncond(M−μI)+Ωth(T,μI,V)−VPex(σ),condensed phase(35)\Omega(T,\mu_I,V)=N_{\mathrm{cond}}(M-\mu_I)+\Omega_{\mathrm{th}}(T,\mu_I,V)-VP_{\mathrm{ex}}(\sigma),\qquad\text{condensed phase} \tag*{(35)}

where

Ωth(T,μI,V)=VT∫d3k(2π)3{ln⁡[1−e−(ωk(σ)−μI)/T]+ln⁡[1−e−(ωk(σ)+μI)/T]}.(36)\Omega_{\mathrm{th}}(T,\mu_I,V)=VT\int\frac{d^3k}{(2\pi)^3}\left\{\ln\left[1-e^{-(\omega_k(\sigma)-\mu_I)/T}\right]+\ln\left[1-e^{-(\omega_k(\sigma)+\mu_I)/T}\right]\right\}. \tag*{(36)}

The condensate term in eq. (35) vanishes identically once the onset condition μI=M\mu_I=M is imposed, but it has to be retained in the variational procedure that determines ncondn_{\mathrm{cond}}.

In the GCE, the isospin number nI=NI/Vn_I=N_I/V is related to the corresponding chemical potential μI\mu_I as:

nI=−1V∂Ω∂μI.(37)n_I=-\frac{1}{V}\frac{\partial\Omega}{\partial\mu_I}. \tag*{(37)}

We use this relation as a basis for the Legendre transform: after solving this equation with respect to the chemical potential one obtains dependence μI=μI(T,nI)\mu_I=\mu_I(T,n_I), i.e., instead of the variable μI\mu_I we go to the new variable nIn_I. It is analog of the definition of the “effective potential” that is the standard transformation in the quantum field theory. In our case the “effective potential” will be the extended free energy. Making the Legendre transformation we move from the Grand Canonical Ensemble to Extended Canonical Ensemble, where we define the free energy as

F(T,NI,V)=Ω−μI∂Ω∂μI.(38)F(T,N_I,V)=\Omega-\mu_I\frac{\partial\Omega}{\partial\mu_I}. \tag*{(38)}

Here we imply that the chemical potential is determined by TT and nIn_I. From now on, the statistical operator ρ^\hat{\rho} and the corresponding partition function in the ECE are defined as

ρ^=e−β[H^MF−μI(T,nI)N^I],Z(T,nI)=Tr⁡ρ^.(39)\hat{\rho}=e^{-\beta[\hat{H}_{\mathrm{MF}}-\mu_I(T,n_I)\hat{N}_I]},\qquad Z(T,n_I)=\operatorname{Tr}\hat{\rho}. \tag*{(39)}

And in place of the connection between the partition function and the free energy in the Canonical Ensemble, F=−Tln⁡Tr⁡e−βHF=-T\ln\operatorname{Tr}e^{-\beta H}, in the Extended Canonical Ensemble in accordance with transformation (38) the free energy is given by

F(T,NI,V)=NIμI(T,nI)−Tln⁡Tr⁡{e−β[H−μI(T,nI)N^I]}.(40)F(T,N_I,V)=N_I\mu_I(T,n_I)-T\ln\operatorname{Tr}\left\{e^{-\beta[H-\mu_I(T,n_I)\hat{N}_I]}\right\}. \tag*{(40)}

where μI=μI(T,nI)\mu_I=\mu_I(T,n_I) is solution of eq. (37).

Using eq. (35) for the condensed phases we can rewrite eq. (37) in explicit form

nI=ncond+nIth,(41)n_I=n_{\mathrm{cond}}+n_{I\mathrm{th}}, \tag*{(41)}
nIth(T,μI)=∫d3k(2π)3[fBE(ωk(σ),μI)−fBE(ωk(σ),−μI)],(42)n_{I\mathrm{th}}(T,\mu_I)=\int\frac{d^3k}{(2\pi)^3}\left[f_{\mathrm{BE}}(\omega_k(\sigma),\mu_I)-f_{\mathrm{BE}}(\omega_k(\sigma),-\mu_I)\right], \tag*{(42)}

where ωk(σ)=M(σ)2+k2\omega_k(\sigma)=\sqrt{M(\sigma)^2+k^2}, M2=m2+U(σ)M^2=m^2+U(\sigma) and the Bose-Einstein distribution function is given by fBE(ω,μ)={exp⁡[(ω−μ)/T]−1}−1f_{\mathrm{BE}}(\omega,\mu)=\{\exp[(\omega-\mu)/T]-1\}^{-1}.

At this point we have to make a very important note. Let us assume that both components of the Bose gas are in the condensed phase. Then, in accordance with the condition for condensate formation, for the arguments of the distribution functions in eq. (41) should be valid two equalities

M(σ)−μI=0,M(σ)+μI=0.(43)M(\sigma)-\mu_I=0,\qquad M(\sigma)+\mu_I=0. \tag*{(43)}

Evidently this set of equations leads to M(σ)=0M(\sigma)=0 and μI=0\mu_I=0, simultaneously. The first equality, i.e. a zero effective mass of the quasiparticles, cannot be realized for the system with only a repulsion between particles, M2=m2+U(σ)M^2=m^2+U(\sigma), where U(σ)≥0U(\sigma)\geq0, see eq. (9). Therefore, only one component of the system can develop a condensate when we keep nI≠0n_I\ne0.1 Let it be the particles, i.e. the component that carries the positive isospin charge and enters with +μI+\mu_I (for the pion system these are the π+\pi^+ mesons), for which the first equation in (43) is valid, μI=M(σ)\mu_I=M(\sigma) in the condensed phase. The latter means that the antiparticles (π−\pi^- mesons) exist only in the thermal phase for all temperature range, hence, they cannot be found in the condensed phase, i.e., the antiparticle amplitude vanishes, ψ0(−)=0\psi_0^{(-)}=0 in eq. (19).

So, making the choice of the positive isospin charge of the dominant component in the pion gas we obtain equations nI=nIth(T,μI)n_I = n_{I\mathrm{th}}(T,\mu_I) for the thermal phase (see eq. (41)), and nI=ncond+nIth(T,M)n_I = n_{\mathrm{cond}} + n_{I\mathrm{th}}(T,M), where μI=M(σ)\mu_I = M(\sigma), for the condensed phase. We use these equations separately in both phases to determine the chemical potential as a function of the temperature and isospin number. Therefore, in the thermal phase the chemical potential as the solution of equation nI=nIth(T,μI)n_I = n_{I\mathrm{th}}(T,\mu_I) is a function of the temperature and the isospin density, μI=μI(T,nI)\mu_I = \mu_I(T,n_I). However, in the condensed phase the value of the chemical potential, μI=M(σ)\mu_I = M(\sigma), depends on the temperature and the isospin density via a dependence on the scalar density, σ=σ(T,nI)\sigma= \sigma(T,n_I).

Using eqs. (34) and (35) one can write the Legendre transformation (38) in explicit form for the thermal and condensed phases, respectively:

F(T,nI)=μInI+Fth(T,nI)−Pex,(44)\mathcal{F}(T,n_I) = \mu_I n_I + \mathcal{F}_{\mathrm{th}}(T,n_I) - P_{\mathrm{ex}}, \tag*{(44)}
F(T,nI)=MnI+Fth(T,nI)−Pex,(45)\mathcal{F}(T,n_I) = M n_I + \mathcal{F}_{\mathrm{th}}(T,n_I) - P_{\mathrm{ex}}, \tag*{(45)}

where the thermal (kinetic) contribution is

Fth(T,nI)=T∫d3k(2π)3{ln⁡[1−e−(ωk−μI)/T]+ln⁡[1−e−(ωk+μI)/T]},(46)\mathcal{F}_{\mathrm{th}}(T,n_I) = T \int\frac{d^3 k}{(2\pi)^3} \left\{ \ln\left[1-e^{-(\omega_k-\mu_I)/T}\right] + \ln\left[1-e^{-(\omega_k+\mu_I)/T}\right] \right\}, \tag*{(46)}

where F(T,nI)=F(T,NI,V)/V\mathcal{F}(T,n_I) = F(T,N_I,V)/V is the density of the free energy, and we recall that in the thermal phase μI=μI(T,nI)\mu_I = \mu_I(T,n_I) and in the condensed phase μI=M(σ(T,nI))\mu_I = M(\sigma(T,n_I)).

Self-consistent equations

For a closed description of the system it is necessary to formulate the self-consistent equations that determine a dependence of the thermodynamic quantities on the free variables TT and nIn_I. In thermodynamically equilibrium case we use the Bose-Einstein distribution functions that were determined above. In the thermal phase we return to equation nI=nIth(T,μI)n_I = n_{I\mathrm{th}}(T,\mu_I), see (41). We calculate also the thermal scalar density, see eq. (23), as a derivative of the Ωth\Omega_{\mathrm{th}}, see eq. (36), with respect to M2M^2: σ=⟨ψ^†ψ^⟩=(1/V)∂Ωth/∂M2\sigma= \langle\hat{\psi}^{\dagger}\hat{\psi}\rangle= (1/V)\partial\Omega_{\mathrm{th}}/\partial M^2. Or calculate it directly using the averagings: ⟨a^k†a^k⟩=2VωkfBE(ωk,μI)\langle\hat{a}^{\dagger}_k\hat{a}_k\rangle= 2V\omega_k f_{\mathrm{BE}}(\omega_k,\mu_I) and ⟨b^k†b^k⟩=2VωkfBE(ωk,−μI)\langle\hat{b}^{\dagger}_k\hat{b}_k\rangle= 2V\omega_k f_{\mathrm{BE}}(\omega_k,-\mu_I). As a result we formulate self-consistent set of equations

nI=∫d3k(2π)3[fBE(ωk,μI)−fBE(ωk,−μI)],(47)n_I = \int\frac{d^3 k}{(2\pi)^3}\left[f_{\mathrm{BE}}(\omega_k,\mu_I)-f_{\mathrm{BE}}(\omega_k,-\mu_I)\right], \tag*{(47)}
σ=∫d3k(2π)32ωk[fBE(ωk,μI)+fBE(ωk,−μI)],(48)\sigma= \int\frac{d^3 k}{(2\pi)^3 2\omega_k}\left[f_{\mathrm{BE}}(\omega_k,\mu_I)+f_{\mathrm{BE}}(\omega_k,-\mu_I)\right], \tag*{(48)}

where σ\sigma is the total thermal scalar density, ωk=M(σ)2+k2\omega_k = \sqrt{M(\sigma)^2 + k^2} and M2(σ)=m2+U(σ)M^2(\sigma) = m^2 + U(\sigma). This set of equations must be solved for a fixed temperature T≥TcT \ge T_c and isospin density nIn_I. Finally, we get the functions μI(T,nI)\mu_I(T,n_I) and σ(T,nI)\sigma(T,n_I).

In the condensed phase the thermodynamic state of the system is determined by a self-consistent equation for the scalar density, σ=σcond+σth\sigma= \sigma_{\rm cond} + \sigma_{\rm th}, which looks like eq. (48) that is completed by the contribution of the condensate scalar density. And we also remember, that according to eq. (24) there is relation σcond=ncond/2M\sigma_{\rm cond} = n_{\rm cond}/2M (recall that antiparticles, π−\pi^- mesons, are only in the thermal phase). Therefore, in the condensed phase the thermodynamic state of the system is determined by a set of two equations for σ\sigma and ncondn_{\rm cond}:

nI=ncond(T,nI)+∫d3k(2π)3[fBE(ωk,M(σ))−fBE(ωk(σ),−M(σ))](49)n_I = n_{\rm cond}(T,n_I) + \int\frac{d^3 k}{(2\pi)^3}\left[f_{\rm BE}(\omega_k,M(\sigma)) - f_{\rm BE}(\omega_k(\sigma),-M(\sigma))\right] \tag*{(49)}
σ=ncond2M+∫d3k(2π)32ωk[fBE(ωk,M(σ))+fBE(ωk,−M(σ))](50)\sigma= \frac{n_{\rm cond}}{2M} + \int\frac{d^3 k}{(2\pi)^3 2\omega_k}\left[f_{\rm BE}(\omega_k,M(\sigma)) + f_{\rm BE}(\omega_k,-M(\sigma))\right] \tag*{(50)}

where, as in the thermal phase, ωk=M2(σ)+k2\omega_k = \sqrt{M^2(\sigma) + k^2} and M2(σ)=m2+U(σ)M^2(\sigma) = m^2 + U(\sigma). This set of equations must be solved for a fixed temperature, TT, and isospin density nIn_I, then we obtain functions: σ(T,nI)\sigma(T,n_I), μI(T,nI)=M(σ(T,nI))\mu_I(T,n_I) = M(\sigma(T,n_I)), ncond(T,nI)n_{\rm cond}(T,n_I) and σcond(T,nI)\sigma_{\rm cond}(T,n_I).

The pressure and energy density are related to the free energy density as:

p=nIμI(T,nI)−F(T,nI),ϵ=F(T,nI)−T[∂F(T,nI)∂T]nI.(51)p = n_I\mu_I(T,n_I) - \mathcal{F}(T,n_I), \qquad\epsilon= \mathcal{F}(T,n_I) - T\left[\frac{\partial\mathcal{F}(T,n_I)}{\partial T}\right]_{n_I}. \tag*{(51)}

Using eqs. (44) and (45) we get the pressure

p(T,nI)=−Fth(T,nI)+Pex(σ),(52)p(T,n_I) = -\mathcal{F}_{\rm th}(T,n_I) + P_{\rm ex}(\sigma), \tag*{(52)}

where Fth(T,nI)\mathcal{F}_{\rm th}(T,n_I) is given in eq. (46). The integral on the r.h.s. of this equation gives the pressure of thermal quasi particles and antiparticles, what together can be treated as the kinetic pressure of the quasi-ideal gas. The additional pressure PexP_{\rm ex} is due to interaction in the system. From this decomposition we see that the quantity PexP_{\rm ex} is indeed the excess pressure. The condensate itself does not produce an independent kinetic-pressure contribution.

For calculation of the energy density we use eq. (51) and obtain for the thermal and condensed phases, respectively:

ϵ(T,nI)=∫d3k(2π)3ωk[fBE(ωk,μI)+fBE(ωk,−μI)]−Pex(σ).(53)\epsilon(T,n_I) = \int\frac{d^3 k}{(2\pi)^3}\omega_k\left[f_{\rm BE}(\omega_k,\mu_I) + f_{\rm BE}(\omega_k,-\mu_I)\right] - P_{\rm ex}(\sigma). \tag*{(53)}
ϵ(T,nI)=Mncond+∫d3k(2π)3ωk[fBE(ωk,M)+fBE(ωk,−M)]−Pex(σ).(54)\epsilon(T,n_I) = M n_{\rm cond} + \int\frac{d^3 k}{(2\pi)^3}\omega_k\left[f_{\rm BE}(\omega_k,M) + f_{\rm BE}(\omega_k,-M)\right] - P_{\rm ex}(\sigma). \tag*{(54)}

where μI=μI(T,nI)\mu_I = \mu_I(T,n_I) in the thermal phase.

The quark content of the pionic phase

Our aim in this section is to put in correspondence the definitions of the isospin chemical potentials μI(π)\mu_{I}^{(\pi)} and μI(q)\mu_{I}^{(q)} for pion and for quark degrees of freedom, respectively.

Let us assume that the hadron system is represented by pions as the most light entities, and the baryon charge of the system is zero, NB=0N_{B}=0. Then, for the pion hadron phase the isospin number reads (recall Iπ=1I_{\pi}=1, with three projections 0,±10,\pm1)

NI=Nπ(+)−Nπ(−),(55)N_{I}=N_{\pi}^{(+)}-N_{\pi}^{(-)}, \tag*{(55)}

where Nπ(+)N_{\pi}^{(+)} and Nπ(−)N_{\pi}^{(-)} are the particle numbers of π+\pi^{+} and π−\pi^{-} mesons, respectively. In what follows we identify the condensing component with the positively charged pion-like state.

We consider “hadron” temperatures, T<Tqgp≈160T<T_{\mathrm{qgp}}\approx160 MeV, where the quarks can be considered as a content of the hadrons, and assume that the vast majority of quark configurations reflect the pion structure, π+=udˉ\pi^{+}=u\bar{d} and π−=duˉ\pi^{-}=d\bar{u}. With account for the isospin projections of the quarks, +1/2+1/2 for uu and dˉ\bar{d} and −1/2-1/2 for uˉ\bar{u} and dd, and for the relations Nu=Ndˉ=Nπ(+)N_{u}=N_{\bar{d}}=N_{\pi}^{(+)}, Nuˉ=Nd=Nπ(−)N_{\bar{u}}=N_{d}=N_{\pi}^{(-)}, eq. (55) can be rewritten as

NI=(12Nu+12Ndˉ)−(12Nuˉ+12Nd)=Nu−Nd.(56)N_{I}=\left(\frac{1}{2}N_{u}+\frac{1}{2}N_{\bar{d}}\right)-\left(\frac{1}{2}N_{\bar{u}}+\frac{1}{2}N_{d}\right)=N_{u}-N_{d}. \tag*{(56)}

The free energy of the pion system with account for the quark content reads

Fπ=−pπV+μuNu+μdNd+μuˉNuˉ+μdˉNdˉ=−pπV+μI(q)12(Nu−Nd−Nuˉ+Ndˉ),(57)\begin{aligned} F_{\pi}&=-p_{\pi}V+\mu_{u}N_{u}+\mu_{d}N_{d}+\mu_{\bar{u}}N_{\bar{u}}+\mu_{\bar{d}}N_{\bar{d}}\\ &=-p_{\pi}V+\mu_{I}^{(q)}\frac{1}{2}\left(N_{u}-N_{d}-N_{\bar{u}}+N_{\bar{d}}\right), \tag*{(57)} \end{aligned}

where μI(q)\mu_{I}^{(q)} is the isospin chemical potential that characterizes the quark configurations, i.e. μu=−μuˉ=μI(q)/2\mu_{u}=-\mu_{\bar{u}}=\mu_{I}^{(q)}/2 and μd=−μdˉ=−μI(q)/2\mu_{d}=-\mu_{\bar{d}}=-\mu_{I}^{(q)}/2, or

μI(q)=μu−μd.(58)\mu_{I}^{(q)}=\mu_{u}-\mu_{d}. \tag*{(58)}

With Nu=NdˉN_{u}=N_{\bar{d}}, Nuˉ=NdN_{\bar{u}}=N_{d} and eq. (56) the free energy takes the form Fπ=−pπV+μI(q)NIF_{\pi}=-p_{\pi}V+\mu_{I}^{(q)}N_{I}. Comparing it with the free energy of the pion system, where only π+\pi^{+} and π−\pi^{-} exist,

Fπ=−pπV+μπ(+)Nπ(+)+μπ(−)Nπ(−)=−pπV+μI(π)NI,(59)F_{\pi}=-p_{\pi}V+\mu_{\pi}^{(+)}N_{\pi}^{(+)}+\mu_{\pi}^{(-)}N_{\pi}^{(-)}=-p_{\pi}V+\mu_{I}^{(\pi)}N_{I}, \tag*{(59)}

we conclude that, if the quark configurations correspond to the pion degrees of freedom, then

μI(q)=μI(π)=μI.(60)\mu_{I}^{(q)}=\mu_{I}^{(\pi)}=\mu_{I}. \tag*{(60)}

Since in an ideal gas the condensate is created when the chemical potential equals the particle mass, the condensation of pions in terms of the quark content begins at μI(q)=μI=mπ\mu_I^{(q)} = \mu_I = m_\pi. This is also the convention of Son and Stephanov [1], who assign to the light quarks chemical potentials of equal magnitude ∣μI∣/2|\mu_I|/2 and opposite sign, and find the onset of pion condensation at ∣μI∣=mπ|\mu_I| = m_\pi (see also [5] and references therein).

In the lattice calculations [6, 7, 8, 9, 10] the isospin chemical potential of quarks is defined as

μI=12(μu−μd),(61)\mu_I = \frac{1}{2}\left(\mu_u-\mu_d\right), \tag*{(61)}

while the isospin number is NI=Nu−NdN_I = N_u - N_d [8] and the baryon density is zero. Then the free energy (57) takes the form

Fπ=−Vpπ+2μINI,(62)F_\pi= -V p_\pi+ 2\mu_I N_I, \tag*{(62)}

i.e. the pion chemical potential entering the Bose distribution is 2μI2\mu_I, and the condition for the condensate creation in an ideal gas becomes

μI=12mπ.(63)\mu_I = \frac{1}{2}m_\pi. \tag*{(63)}

The difference between the two thresholds, μI=mπ\mu_I = m_\pi and μI=mπ/2\mu_I = m_\pi/2, is thus a normalization of the isospin chemical potential.

Description of the lattice data

To describe the results of the lattice calculations [6, 7, 8, 9, 10], we use the mean-field approach presented in section III. We consider the pion-like system within two models, φ4\varphi^4 and φ6\varphi^6, with a repulsive self-interaction. We stress that the quasiparticles of the model are not the physical pions: at the condensation threshold their effective mass is M=mπ/2M = m_\pi/2, see eq. (64) below, and they couple directly to the isospin chemical potential in the lattice normalization (61). The model should therefore be regarded as a phenomenological parametrization of the dependence of the lattice thermodynamics on μI\mu_I. We aim to show the competition between these two models in fitting the lattice data.

To fit the lattice data we use the same definitions that were adopted in Refs. [8, 9, 10]: the pion mass value mπ=135m_\pi= 135 MeV, μI=12(μu−μd)\mu_I = \frac{1}{2}(\mu_u-\mu_d) and NI=Nu−NdN_I = N_u - N_d. That is, for the comparison below we use the lattice convention, eq. (61). We fit the data at the fixed temperature T=120T = 120 MeV. To be adequate to the lattice calculations [6, 7, 8, 9, 10] we adopt that at this temperature the phase transition to the condensed phase occurs when μI=mπ/2\mu_I = m_\pi/2. In accordance with our model, this means that at this temperature the effective (thermal) mass MM of pions is

M(σcr)=12mπ,(64)M(\sigma_{\mathrm{cr}}) = \frac{1}{2}m_\pi, \tag*{(64)}

where the value of the scalar density σcr(T)\sigma_{\mathrm{cr}}(T) belongs to the critical curve. The value σcr\sigma_{\mathrm{cr}} can be found from self-consistent equation (50) when ncond=0n_{\mathrm{cond}} = 0:

σcr(M,T)=∫d3k(2π)32ωk[fBE(ωk,M)+fBE(ωk,−M)],(65)\sigma_{\mathrm{cr}}(M,T) = \int\frac{d^3 k}{(2\pi)^3 2\omega_k}\left[f_{\mathrm{BE}}(\omega_k,M) + f_{\mathrm{BE}}(\omega_k,-M)\right] , \tag*{(65)}

where ωk=mπ2/4+k2\omega_k = \sqrt{m_\pi^2/4 + k^2} and we use condition of the condensate creation of the positively charged particles, μI=M\mu_I = M, which is valid in the condensed phase, starting from the critical point. Here, in correspondence to Refs. [9, 10], in eq. (65) we use the value of the effective mass M=mπ/2M = m_{\pi}/2 at T=120T = 120 MeV. With these two values eq. (65) transforms into a formula for the numerical calculation of the critical scalar density σcr\sigma_{\mathrm{cr}}.

As was declared in Section II we fit the lattice data using two approaches which correspond to two interaction Lagrangians presented in eq. (4). In accordance with the formalism given in Section II A every type of interaction Lagrangian determines its own mean field (see eq. (9)): U(σ)=−∂Lint(σ)/∂σU(\sigma) = -\partial L_{\mathrm{int}}(\sigma)/\partial\sigma. And further the effective mass MM of quasiparticles is determined as: M2=m∗2+U(σ)M^{2} = m_{*}^{2} + U(\sigma), where m∗m_{*} is the bare mass and M(σcr)=mπ/2M(\sigma_{\mathrm{cr}}) = m_{\pi}/2. Hence for every approach at the critical point we get:

M2=m∗2+λσcrfor ϕ4 model,(66)M^{2} = m_{*}^{2} + \lambda\sigma_{\mathrm{cr}} \qquad\text{for }\phi^{4}\text{ model}, \tag*{(66)}
M2=m∗2+bσcr2for ϕ6 model.(67)M^{2} = m_{*}^{2} + b\sigma_{\mathrm{cr}}^{2} \qquad\text{for }\phi^{6}\text{ model}. \tag*{(67)}

These relations give the connection between the value of the bare mass m∗m_{*} and the interaction constant λ\lambda or bb, respectively. To find interaction constant, we use eq. (52) to fit the lattice pressure [10] as Δp(T,μI)=p(T,μI)−p(T,0)\Delta p(T,\mu_{I}) = p(T,\mu_{I}) - p(T,0). For the best fit of the Δp\Delta p we find: λ≈0.4934\lambda\approx0.4934, bmπ2≈0.5673b m_{\pi}^{2} \approx0.5673. In accordance with eqs. (66) and (67) the corresponding fitted bare masses are m∗=mπ/2.22m_{*} = m_{\pi}/2.22 for the ϕ4\phi^{4} model and m∗=mπ/2.021m_{*} = m_{\pi}/2.021 for the ϕ6\phi^{6} model. This fit of the lattice pressure is shown in Figure 1, in the left panels. To our understanding, the result, that the bare mass of particles in both approaches is approximately two times smaller than the pion mass mπ=135m_{\pi} = 135 MeV, is a consequence of the normalization of the chemical potential μI\mu_{I} for quarks adopted in [6, 7, 8, 9, 10], see eq. (61).

Figure 1. Fit of lattice calculations results [10] at T=120 MeVT = 120\ \mathrm{MeV} for the quantity Δp\Delta p in the ϕ4\phi^4 model (top left panel) and the ϕ6\phi^6 model (bottom left panel). Description of lattice results [10] at T=122 MeVT = 122\ \mathrm{MeV} for nIn_I in the ϕ4\phi^4 model (top right panel) and the ϕ6\phi^6 model (bottom right panel). The red segment of the approximation curve refers to the thermal phase, and the blue segment to the condensed phase.

In the thermal region, the chemical potential is obtained by solving the thermal self-consistency equations at fixed TT and nIn_{I}. In the condensed region it is constrained by μI=M(T,nI)\mu_{I} = M(T,n_{I}), with the effective mass determined from the coupled condensate equations. For this reason, at a fixed temperature T=120T = 120 MeV, we parameterize all quantities in terms of the isospin density nIn_{I}: nI<nIcrn_{I} < n_{I\mathrm{cr}} for the thermal phase and nI≥nIcrn_{I} \ge n_{I\mathrm{cr}} for the condensed phase, where nIcrn_{I\mathrm{cr}} is obtained by performing the integration in eq. (49) under the conditions ncond=0n_{\mathrm{cond}} = 0 and M=mπ/2M = m_{\pi}/2. In all panels in Figure 1 and Figure 2 the X-axis is a chemical potential, which is taken as solution of eqs. (47), (48) for fixed temperature in the thermal phase; and in the condensed phase a chemical potential equals to the effective mass, μI=M(T,nI)\mu_{I} = M(T,n_{I}), which is taken as solution of eqs. (49), (50). In this sense the calculations are carried out in the Extended Canonical Ensemble.

Figure 2. Description of lattice calculation results [9, 10] at T=120T = 120 MeV for the quantities Δε\Delta\varepsilon and ΔI\Delta I in the ϕ4\phi^4 model (top panels) and the ϕ6\phi^6 model (bottom panels). The red segment of the fit curve corresponds to the thermal phase, and the blue segment to the condensed phase.

The calculation of all thermodynamic quantities was carried out using the interaction constants λ≈0.4934\lambda\approx0.4934 and bmπ2≈0.5673b m_{\pi}^{2} \approx0.5673 obtained by fitting the pressure. The results of the calculation of the isospin density nIn_I (see, eqs. (47), (49)) at T=122T = 122 MeV together with the lattice data [10] are shown in Figure 1, in the right panels. One can see a satisfactory description of the lattice calculation results, where the ϕ6\phi^6 being more accurate. The results of the calculation at T=120T = 120 MeV of the energy density in the form

Δε(T,μI)=ε(T,μI)−ε(T,0),(68)\Delta\varepsilon(T,\mu_I) = \varepsilon(T,\mu_I) - \varepsilon(T,0), \tag*{(68)}

and the trace-anomaly

I=ε−3p=I(T,0)+ΔI(T,μI),(69)I = \varepsilon- 3p = I(T,0) + \Delta I(T,\mu_I), \tag*{(69)}

are shown in Figure 2 in left and right panels together with the lattice data [10]. In all figures the red segment of the curve relates to the thermal phase and the blue segment of the curve relates to the condensed phase. We see quite good description of the lattice data. However, comparison of two model results leads to conclusion, that the φ6\varphi^6 model describes better the results of the lattice calculations.

The agreement is nontrivial because the quantities shown in the different panels are not independently fitted: once the interaction parameter λ\lambda or bb is fixed from the pressure, the isospin density, energy density, and trace anomaly follow from the same thermodynamically consistent framework.

The comparison therefore provides a test of both the interaction model and the thermodynamic formulation. In particular, the thermal and condensed branches match continuously at the transition.

It is necessary to point out, that for a boson system with repulsion only, when M≥m∗→μI≥m∗M \ge m_* \to\mu_I \ge m_* in the condensed phase, the thermodynamic quantities obtained in this way can be presented as functions of μI\mu_I, since at fixed temperature the correspondence between nIn_I and μI\mu_I is one-to-one. In this case it is necessary to account for a transformation of the dispersion law in the condensed phase:

ωk=M2+k2→ωk=μI2+k2.(70)\omega_k = \sqrt{M^2 + \mathbf{k}^2} \rightarrow\omega_k = \sqrt{\mu_I^2 + \mathbf{k}^2}. \tag*{(70)}

And for the same reason the set of eqs. (49), (50) must be rewritten as

nI=ncond(T,nI)+∫d3k(2π)3[fBE(ωk,μI)−fBE(ωk,−μI)],(71)n_I = n_{\mathrm{cond}}(T,n_I) + \int\frac{d^3 k}{(2\pi)^3}\left[f_{\mathrm{BE}}(\omega_k,\mu_I) - f_{\mathrm{BE}}(\omega_k,-\mu_I)\right], \tag*{(71)}
σ=ncond2M+∫d3k(2π)32ωk[fBE(ωk,μI)+fBE(ωk,−μI)],(72)\sigma= \frac{n_{\mathrm{cond}}}{2M} + \int\frac{d^3 k}{(2\pi)^3 2\omega_k}\left[f_{\mathrm{BE}}(\omega_k,\mu_I) + f_{\mathrm{BE}}(\omega_k,-\mu_I)\right], \tag*{(72)}

where, σ=(μI2−m2)/λ\sigma= (\mu_I^2 - m^2)/\lambda for φ4\varphi^4 model and σ=(μI2−m2)/b\sigma= \sqrt{(\mu_I^2 - m^2)/b} for φ6\varphi^6 model. It is seen that such a transformation of the dynamical variable, σ→μI\sigma\to\mu_I, is possible only when μI≥m\mu_I \ge m, i.e. it is allowed only for the repulsive interaction between particles, M≥mM \ge m.

In the boson system in the case of attraction between particles it is impossible to parameterize the thermodynamic quantities by the chemical potential, since the correspondence between μI\mu_I and the thermodynamic states becomes two-valued: two different thermodynamic states correspond to one value of μI\mu_I. This was schematically shown in Figure 3. It is seen that for the top curve, i.e. there is only repulsion, all states can be characterized unequivocally by the pair (T,μI)(T,\mu_I). It is not the case for the ideal gas, where in the condensed phase all states in the interval 0≤T≤Tc0 \le T \le T_c have the same chemical potential μI=mπ\mu_I = m_\pi. When an attraction is “switched on”, for all curves below the ideal gas one definite value of the chemical potential corresponds to a couple of the thermodynamic states – one state in the condensed phase (blue segment) and another in the thermal phase (red segment), see Figure 3.

Figure 3. Dependence of the μI\mu_I on the temperature at nI=0.1 fm−3n_I = 0.1\ \mathrm{fm}^{-3} for different intensities of the attraction between particles that is parameterized by the dimensionless parameter κ\kappa, the coefficient that determines the strength of the attraction (see Ref. [4]): a. κ=0\kappa= 0 – no attraction, there is only repulsion between particles (top curve); b. an ideal gas – the black dashed curve; c. an increasing intensity of attraction corresponds to the curves labeled by an increasing value of the κ=0.1\kappa= 0.1, 0.2, 0.3. The blue segment of the curves corresponds to the condensed phase, the red segment to the thermal phase.

For parametrization of the pressure, energy density, and trace anomaly, for the same interaction constant and in the same framework as it was done before, one can use the real physical quantity – isospin density nIn_I. The results of the calculation for φ4\varphi^4 and φ6\varphi^6 models are shown in Figure 4. In this figure the beginning of dependencies on every panel starts from the following values:

Figure 4. Dependence of the pressure (left panel), energy density (central panel), and trace anomaly (right panel) at T=120T = 120 MeV on the isospin density nIn_I calculated for φ4\varphi^4 and φ6\varphi^6 models. The red segment of the each curve corresponds to the thermal phase, and the blue segment to the condensed phase.

p(T,μI=0)=p(T,nI=0)≈8.2T4,p(T,\mu_I=0)=p(T,n_I=0)\approx8.2T^4,
ε(T,μI=0)=ε(T,nI=0)≈25.8T4,\varepsilon(T,\mu_I=0)=\varepsilon(T,n_I=0)\approx25.8T^4,
I(T,μI=0)=I(T,nI=0)≈1.2T4.I(T,\mu_I=0)=I(T,n_I=0)\approx1.2T^4.

These values were set to zero for the dependencies of the same thermodynamic quantities on μI\mu_I in Figure 1, Figure 2.

Conclusions

We have studied an interacting relativistic system of charged bosons at finite temperature and fixed isospin density within a thermodynamically consistent mean-field approach. Repulsive φ4\varphi^4 and φ6\varphi^6 self-interactions generate a density-dependent quasiparticle mass and an excess-pressure contribution. The thermodynamic properties are formulated in the Extended Canonical Ensemble, in which the isospin density is the independent conserved variable.

A characteristic feature of the condensed phase is that the isospin chemical potential is constrained by the Bose-condensation condition, μI=M(σ)\mu_I=M(\sigma), where the effective mass is determined self-consistently by the scalar density. For purely repulsive interactions the results can equally be presented as functions of μI\mu_I, since at fixed temperature the correspondence between nIn_I and μI\mu_I is one-to-one. The ECE formulation provides a convenient description of the system across the thermal-to-condensed transition and becomes necessary for an ideal gas or in the presence of attraction, where the value of μI\mu_I does not determine the thermodynamic state uniquely.

We have also analyzed the relation between the isospin chemical potentials defined for pion and quark degrees of freedom. The numerical value of the chemical potential at the condensation threshold depends on its normalization. In the convention μI=μu−μd\mu_I=\mu_u-\mu_d the ideal-gas condition is μI=mπ\mu_I=m_\pi, whereas in the convention used in lattice-QCD calculations [6, 7, 8, 9, 10], μI=(μu−μd)/2\mu_I=(\mu_u-\mu_d)/2, it becomes μI=mπ/2\mu_I=m_\pi/2. Making this distinction explicit is essential for a consistent comparison between the pion and quark descriptions.

Finally, we applied the mean-field equations, regarded as a phenomenological parametrization, to lattice-QCD results at T=120T=120 MeV. Both the φ4\varphi^4 and φ6\varphi^6 models reproduce the main behavior of the pressure, isospin density, energy density, and trace anomaly. In each model only one parameter, the interaction constant, is fitted to the lattice pressure; the isospin density, energy density, and trace anomaly then follow without additional fitting. Within the present parametrization, the φ6\varphi^{6} model provides the better overall description of the lattice results [10].

At the same time, the success of the presented phenomenological interacting pion-like models suggests that, at “hadron” temperatures, at least at T=120T = 120 MeV, and at the considered isospin densities, the vast majority of quark configurations reflect pion structure, with nd=nun_{d} = n_{u} and nˉu=nd\bar{n}_{u} = n_{d}.

These results support the use of a thermodynamically consistent interacting-boson model as an effective description of finite-isospin hadronic matter across the Bose-condensation transition.

Acknowledgements

The work is supported by the National Research Foundation of Ukraine under the grant number 2025.07/0461. Authors thank A. Korchin for useful and fruitful discussions.

References

References

  1. [1]D.T. Son and M.A. Stephanov, QCD at Finite Isospin Density, Phys. Rev. Lett. 86, 592 (2001); [DOI: 10.1103/PhysRevLett.86.592]; [arXiv:hep-ph/0005225].DOI
  2. [2]J. Kogut, and D. Toublan, QCD at small non-zero quark chemical potentials (15p), Phys. Rev. D64, 034007 (2001); [arXiv:hep-ph/0103271].
  3. [3]D. Toublan, and J. Kogut, Isospin Chemical Potential and the QCD Phase Diagram at Nonzero Temperature and Baryon Chemical Potential (4p), Phys. Lett. B564, 212 (2001); [arXiv:hep-ph/0301183].
  4. [4]I.N. Mishustin, D.V. Anchishkin, L.M. Satarov, O.S. Stashko, and H. Stoecker, Condensation of interacting scalar bosons at finite temperatures, Phys. Rev. C 100, 022201(R) (2019); [DOI: 10.1103/PhysRevC.100.022201]; arXiv:1905.09567 [nucl-th].DOI
  5. [5]D. Anchishkin, V. Gnatovskyy, and I. Kondakova, Ideal Boson Particle-Antiparticle System at Finite Temperatures, J. Phys. G: Nucl. Part. Phys. 52, 115002, No.11 (2025); [DOI 10.1088/1361-6471/ae1152] arXiv: 2507.10752 [nucl-th] (2025).DOI
  6. [6]B.B. Brandt, G. Endrődi, QCD phase diagram with isospin chemical potential (14p.), PoS LAT-TICE2016 039 (2016); [arXiv:1611.06758 [hep-lat]].
  7. [7]B.B. Brandt, G. Endrődi, S. Schmalzbauer, QCD at finite isospin chemical potential (16p), EPJ Web Conf. 175, 07020 (2018); [DOI: 10.1051/epjconf/201817507020]; arXiv:1709.10487 [hep-lat].DOI
  8. [8]B.B. Brandt, G. Endrődi, S. Schmalzbauer, QCD phase diagram for nonzero isospin-asymmetry, Phys. Rev. D 97, 054514 (2018); arXiv:1712.08190 [hep-lat].DOI
  9. [9]Bastian B. Brandt, Francesca Cuteri and Gergely Endrődi, QCD thermodynamics at non-zero isospin asymmetry, PoS LATTICE2021, 132 (2022); [DOI: 10.22323/1.396.0132]; arXiv:2110.14750 [hep-lat].
  10. [10]B. B. Brandt, F. Cuteri, and G. Endrődi, Equation of state and speed of sound of isospin-asymmetric QCD on the lattice, JHEP 07, 055 (2023); [DOI: 10.1007/JHEP07(2023)055]; arXiv:2212.14016 [hep-lat].
  11. [11]A. Mammarella, and M. Mannarelli, Phys. Rev. D92, 085025 (2015); [1507.02934 [hep-ph]], 2015.
  12. [12]S. Carignano, L. Lepori, A. Mammarella, M. Mannarelli, G. Pagliaroli, Eur. Phys. J. A53, 35 (2017); [arXiv:1610.06097 [hep-ph]].DOI
  13. [13]L.M. Satarov, M.N. Dmitriev, I.N. Mishustin, Equation of state of hadron resonance gas and the phase diagram of strongly interacting matter, Phys. Atom. Nucl. 72, 1390-1415 (2009); arXiv:0901.1430 [hep-ph].DOI
  14. [14]Volodymyr Vovchenko, Anton Motornenko, Paolo Alba, Mark I. Gorenstein, Leonid M. Satarov, and Horst Stoecker, Multicomponent van der Waals equation of state: Applications in nuclear and hadronic physics, Phys. Rev. C 96, 045202 (2017); [DOI: 10.1103/PhysRevC.96.045202]; arXiv:1707.09215 [nucl-th].
  15. [15]D. Anchishkin, V. Vovchenko, J. Phys. G: Nucl. Part. Phys. 42, 105102: 1-27 (2015); arXiv:1411.1444 [nucl-th].DOI
  16. [16]N. Bogolubov, On the theory of superfluidity, Sov. J. Phys. 11, 23 (1947). M.M. Bogolyubov, Lekciji z kvantovoyi statystyky, Kyiv, 1947 (Ukrainian). N.N. Bogoliubov, Lectures on Quantum Statistics, Gordon and Breach, New York, 1967.
  17. [17]V.G. Kiselev, Ya.M. Shnir, A.Ya. Tregubovich, Introduction to Quantum Field Theory, CRC Press, NY 2000.

Paper details

vlnc:W_d59xrbt7
Version
v1 · latest
Published
Fresh — added to the Hub within the last 30 days
Added
License
CC BY 4.0
Usage

views · 0

downloads · 0

A view counts once per browser per day, a download each time the PDF is requested.

  • Physics and Astronomy

Not recorded for this paper

View statement
Looking for related papers…
Contents