betaFeedback

WorksPhysics and Astronomy

Inverse problem in femtoscopic correlation functions: The Tcc(3875)+T_{cc}(3875)^+ state

Abstract

We study here the inverse problem of starting from the femtoscopic correlation functions of related channels and analyze them with an efficient tool to extract the maximum information possible on the interaction of the components of these channels, and the existence of possible bound states tied to this interaction. The method is flexible enough to accommodate non-molecular components and the effect of missing channels relevant for the interaction. We apply the method to realistic correlation functions for the D∗+D0D^{*+}D^0 and D∗0D+D^{*0}D^+ channels derived consistently from the properties of the Tcc(3875)+T_{cc}(3875)^+ and find that we can extract the existence of a bound state, its nature as a molecular state of the D∗+D0D^{*+}D^0 and D∗0D+D^{*0}D^+ channels, the probabilities of each channel, as well as scattering lengths and effective ranges for the channels, together with the size of the source function, all of them with a relatively good precision.

Introduction

The measurement of correlation functions when studying the hadron-pairs production in heavy ion collision is becoming a useful tool to learn about hadron dynamics, as proved in a number of experimental [1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12] and theoretical papers [13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26]. In particular, the ALICE collaboration has already started to work on the DD sector with the study of the DpDp channel [27] and a systematic study of the correlations function for the exotic mesons is planned by the same collaboration for the future [28].

In a recent paper [29] we studied the correlation functions of the D∗+D0D^{*+}D^{0} and D∗0D+D^{*0}D^{+} channels in connection with the Tcc(3875)+T_{cc}(3875)^{+} state discovered in Refs. [30, 31]. The study took advantage of using an on shell factorization of the Bethe-Salpeter equation which leads to the formulation of Koonin-Prat [22, 32, 33, 34] but in a simplified way. A different method based on coordinate space, rather than momentum space, is used in Ref. [35] obtaining results in qualitative agreement with Ref. [29]. The purpose of the present paper is to do the inverse problem. This is, starting from the correlation functions for D∗+D0D^{*+}D^{0} and D∗0D+D^{*0}D^{+}, to employ an analysis tool to obtain the properties of the Tcc(3875)+T_{cc}(3875)^{+}. We present here three methods, applied to this particular case, but which are general and can be used for many problems in hadron physics. The first method is extremely simple and allows one to obtain approximate scattering lengths. The second method uses a KK-matrix approach that allows us to improve on the first one. Finally, the third method, more sophisticated and consistent with dispersion relations, considers real parts not included in the KK-matrix approach, and allows us to obtain scattering lengths, effective radii, and the binding and width of the Tcc(3875)+T_{cc}(3875)^{+} state, this is, all the information concerning the Tcc(3875)+T_{cc}(3875)^{+} state measured by LHCb collaboration. At the same time, the analysis allows us to determine the molecular compositeness of the Tcc(3875)+T_{cc}(3875)^{+} as has been recently shown in Ref. [36] (see also Ref. [37]).

Formalism: correlation functions

We start with a brief discussion about the correlation functions, and how the D∗DD^{*}D interaction enters into their calculation. We consider the coupled channels D∗+D0D^{*+}D^{0} and D∗0D+D^{*0}D^{+}. As discussed in Ref. [29] (where the reader is referred for further details), for the SS-wave interaction relevant in the limited range of values of the c.m. momentum of the outgoing D∗DD^{*}D system, the correlation functions can be written as:

CD∗0D∗+(pD0)=1+4πθ(Λ−pD0)∫0∞drr2S12(r){∣j0(pD0r)+T11(s)G~1(r;s)∣2+∣T12(s)G~2(r;s)∣2−j02(pD0r)},(1)C_{D^{*0}D^{*+}}(p_{D^{0}})=1+4\pi\theta(\Lambda-p_{D^{0}})\int_{0}^{\infty}dr r^{2}S_{12}(r) \left\{\left|j_{0}(p_{D^{0}}r)+T_{11}(s)\widetilde{G}_{1}(r;s)\right|^{2} +\left|T_{12}(s)\widetilde{G}_{2}(r;s)\right|^{2} -j_{0}^{2}(p_{D^{0}}r)\right\}, \tag*{(1)}
CD+D∗0(pD+)=1+4πθ(Λ−pD+)∫0∞drr2S12(r){∣j0(pD+r)+T22(s)G~2(r;s)∣2+∣T12(s)G~1(r;s)∣2−j02(pD+r)},(2)C_{D^{+}D^{*0}}(p_{D^{+}})=1+4\pi\theta(\Lambda-p_{D^{+}})\int_{0}^{\infty}dr r^{2}S_{12}(r) \left\{\left|j_{0}(p_{D^{+}}r)+T_{22}(s)\widetilde{G}_{2}(r;s)\right|^{2} +\left|T_{12}(s)\widetilde{G}_{1}(r;s)\right|^{2} -j_{0}^{2}(p_{D^{+}}r)\right\}, \tag*{(2)}

where Tij(s)T_{ij}(s) are the TT-matrix elements of the coupled channel system, to be further discussed below, and $s \equiv E2E^2 is the invariant mass squared of the D∗DD^*D pair. Furthermore, pD0p_{D^0} and pD+p_{D^+} are the c.m. momenta of the D0D∗+D^0D^{*+} and D+D∗0D^+D^{*0} channels, respectively, given by:

pD0≡q=λ1/2(s,mD02,mD∗+2)2s,(3)p_{D^0} \equiv q = \frac{\lambda^{1/2}(s,m_{D^0}^2,m_{D^{*+}}^2)}{2\sqrt{s}}, \tag*{(3)}
pD+≡k=λ1/2(s,mD+2,mD∗02)2s(4)p_{D^+} \equiv k = \frac{\lambda^{1/2}(s,m_{D^+}^2,m_{D^{*0}}^2)}{2\sqrt{s}} \tag*{(4)}

being λ(a,b,c)=a2+b2+c2−2ab−2ac−2bc\lambda(a,b,c)=a^2+b^2+c^2-2ab-2ac-2bc the well known Källen function. The zeroth order spherical Bessel function has been denoted by j0(z)j_0(z).

S12(r)S_{12}(r) is the so-called source function, assumed to be spherically symmetric and usually parameterized as a gaussian function:

S12(r)=1(4π)3R3exp⁡(−r24R2),(5)S_{12}(r)=\frac{1}{(\sqrt{4\pi})^3R^3}\exp\left(-\frac{r^2}{4R^2}\right), \tag*{(5)}

with RR a parameter representing the size of the system, typically in the range 1 to 5 fm, depending on the transverse mass of the initial colliding system, i.e., pppp or heavy ions collisions. We shall see that in the inverse problem that we solve, we are also able to get this parameter.

Finally, the function G~i(r;s)\widetilde{G}_i(r;s) is given by

G~i(r;s)=∫0ΛdqHi(q)j0(qr)s−(ω1(i)(q)+ω2(i)(q))2+iϵ,(6)\widetilde{G}_i(r;s)=\int_0^\Lambda dq H_i(q)\frac{j_0(qr)}{s-\left(\omega_1^{(i)}(q)+\omega_2^{(i)}(q)\right)^2+i\epsilon}, \tag*{(6)}
Hi(q)=q22π2ω1(i)(q)+ω2(i)(q)2ω1(i)(q)ω2(i)(q),(7)H_i(q)=\frac{q^2}{2\pi^2}\frac{\omega_1^{(i)}(q)+\omega_2^{(i)}(q)}{2\omega_1^{(i)}(q)\omega_2^{(i)}(q)}, \tag*{(7)}

where ωj(i)(q)=mj(i)2+q2\omega_j^{(i)}(q)=\sqrt{m_j^{(i)2}+q^2}, which for practical purposes is better evaluated using

G~(i)(r;s)=∫0ΛdqHi(q)j0(qr)−j0(pir)s−(ω1(i)(q)+ω2(i)(q))2+j0(pir)Gi(s),(8)\widetilde{G}^{(i)}(r;s)=\int_0^\Lambda dq H_i(q)\frac{j_0(qr)-j_0(p_ir)}{s-\left(\omega_1^{(i)}(q)+\omega_2^{(i)}(q)\right)^2}+j_0(p_ir)G_i(s), \tag*{(8)}

with:

Gi(s)=∫0ΛdqHi(q)1s−(ω1(i)(q)+ω2(i)(q))2+iϵ.(9)G_i(s)=\int_0^\Lambda dq H_i(q)\frac{1}{s-\left(\omega_1^{(i)}(q)+\omega_2^{(i)}(q)\right)^2+i\epsilon}. \tag*{(9)}

The advantage of this form is that, since s=ω1(i)(pi)+ω2(i)(pi)\sqrt{s}=\omega_1^{(i)}(p_i)+\omega_2^{(i)}(p_i), the first integral in Eq. (8) has no longer a singularity, while the second integral is a standard two-point loop function regularized with a cutoff Λ\Lambda, that can be evaluated analytically as can be found in Ref. [38] (in the erratum).

Formalism: T-matrix models

Simplified method to obtain the scattering lengths

We start with a transition potential, which we do not need to specify:

Vij=(V11V12V21V22),(10)V_{ij}=\begin{pmatrix}V_{11} & V_{12}\\ V_{21} & V_{22}\end{pmatrix}, \tag*{(10)}

and the scattering amplitude given by:

T=[1−VG]−1V,(11)T=[1-VG]^{-1}V, \tag*{(11)}

with GG a diagonal matrix with elements given by the loop function Eq. (9) of the intermediate pseudoscalar-vector channels, G≡diag⁡[G1(s),G2(s)]G\equiv\operatorname{diag}[G_1(s),G_2(s)]. The value of Λ\Lambda in the Gi(s)G_i(s) function indicates the range of the interaction in momentum space [39, 40].

In this part we ignore the D∗D^* width, which is shown in Ref. [36] to have a very small effect on the scattering lengths. We also ignore in this part the effect of the effective range r0r_0, also shown to be moderate in the study of [29]. With the normalization of the GG function of Eq. (9) suited to the field theoretical formulation one has [36]

T11≃−8πs1a1−iq,(12)T_{11}\simeq-\frac{8\pi\sqrt{s}}{\frac{1}{a_1}-iq}, \tag*{(12)}
T22≃−8πs1a2−ik,(13)T_{22}\simeq-\frac{8\pi\sqrt{s}}{\frac{1}{a_2}-ik}, \tag*{(13)}

where a1a_1, a2a_2 are, respectively, the scattering lengths of the D∗+D0D^{*+}D^0 and D∗0D+D^{*0}D^+ channels and neglecting the D∗D^* width, a1a_1 is real, but a2a_2 is complex because D∗0D+D^{*0}D^+ can decay into D∗+D0D^{*+}D^0. At the D∗0D+D^{*0}D^+ threshold, k=0k=0 and the only imaginary part comes from GD∗+D0G_{D^{*+}D^0} in Eq. (11), Im⁡GD∗+D0=−q/8πs\operatorname{Im}G_{D^{*+}D^0}=-q/8\pi\sqrt{s}. Ignoring terms in q2q^2, consistently with also not considering the terms in r0q2/2r_0q^2/2 of the effective range expansion, we can write

−1a2=b2−ib2′q.(14)-\frac{1}{a_2}=b_2-ib'_2q. \tag*{(14)}

Let us recall that the c.m. differential cross section in the present normalization is given by [41]

dσi←jdΩ=164π2pjpi∣Tij∣2(15)\frac{d\sigma_{i\leftarrow j}}{d\Omega}=\frac{1}{64\pi^2}\frac{p_j}{p_i}|T_{ij}|^2 \tag*{(15)}

with pi,pjp_i,p_j the initial and final momenta respectively, and the optical theorem states [41]

Im⁡T22=−2ksσtot,σtot=σ22+σ1←2.(16)\operatorname{Im}T_{22}=-2k\sqrt{s}\sigma_{\rm tot},\qquad\sigma_{\rm tot}=\sigma_{22}+\sigma_{1\leftarrow2}. \tag*{(16)}

Using Eqs. (11), (14), (15), and (16), we immediately find that

σ1←2=σtot−σ22=1k4πb2′qb22+(b2′q+k)2=164π24πsqk∣T12∣2.(17)\sigma_{1\leftarrow2}=\sigma_{\rm tot}-\sigma_{22}=\frac{1}{k}\frac{4\pi b'_2q}{b_2^2+(b'_2q+k)^2}=\frac{1}{64\pi^2}\frac{4\pi}{s}\frac{q}{k}|T_{12}|^2. \tag*{(17)}

We see that by having T11T_{11} and T22T_{22} one cannot get T12T_{12}, but we can deduce ∣T12∣2\lvert T_{12}\rvert^2, which is exactly the magnitude we need for the correlation function. Thus, one can express ∣T12∣2\lvert T_{12}\rvert^2 as

∣T12∣2=64π2sb2′2b22+(b2′q+k)2(18)\lvert T_{12}\rvert^2 = \frac{64\pi^2 s b_2'^2}{b_2^2 + (b_2' q + k)^2} \tag*{(18)}

Below the second threshold kk is purely imaginary (k→iκk \to i\kappa) and then the denominator of Eq. (18) becomes (b2+κ)2+b2′2q2(b_2 + \kappa)^2 + b_2'^2 q^2.

K-matrix formalism

A quite general parameterization of the TT-matrix is the KK-matrix parameterization,

T−1(s)=K−1(s)−iρ(s)(19)T^{-1}(s) = K^{-1}(s) - i\rho(s) \tag*{(19)}

where ρ(s)\rho(s) is a diagonal matrix with elements

ρii(s)=−pi(s)8πs=Im⁡Gi(s)(20)\rho_{ii}(s) = -\frac{p_i(s)}{8\pi\sqrt{s}} = \operatorname{Im} G_i(s) \tag*{(20)}

This parameterization fulfills unitarity as long as the KK-matrix elements do not have a right-hand cut. Therefore, they can be rational functions, or functions with left-hand cuts, or a combination thereof. For our particular problem, we consider the simplest case in which the KK-matrix elements are just constants.

Determination of the complete information on the interaction

For this part we follow the strategy used in Ref. [36] to determine the compositeness of the Tcc(3875)+T_{cc}(3875)^+ state from the mass distribution of Ref. [31] (Fig. 8 of supplementary material).

We start from the potential of Eq. (10) and the TT-matrix of Eq. (11) and assume that isospin is a good symmetry of the potential (it is a bit broken at the level of the TT-matrix because of the different masses of the two channels). With the isospin doublets (D+,−D0)(D^+, -D^0), (D∗+,−D∗0)(D^{*+}, -D^{*0}) we write the I=0,1I = 0, 1 states as

∣D∗D,I=0⟩=−12(∣D∗+D0⟩−∣D∗0D+⟩)(21)\lvert D^*D, I = 0\rangle= -\frac{1}{\sqrt{2}}\left(\lvert D^{*+}D^0\rangle- \lvert D^{*0}D^+\rangle\right) \tag*{(21)}
∣D∗D,I=1⟩=−12(∣D∗+D0⟩+∣D∗0D+⟩)(22)\lvert D^*D, I = 1\rangle= -\frac{1}{\sqrt{2}}\left(\lvert D^{*+}D^0\rangle+ \lvert D^{*0}D^+\rangle\right) \tag*{(22)}

Demanding isospin conservation, i.e. ⟨I=0∣V∣I=1⟩=0\langle I = 0\lvert V\lvert I = 1\rangle= 0, we obtain V22=V11V_{22} = V_{11}.

In addition, in order to allow for the general case in which the Tcc(3875)+T_{cc}(3875)^+ could be a compact tetraquark state, a pure molecular state of the D∗+D0D^{*+}D^0 and D∗0D+D^{*0}D^+ states, or contain a mixture of both, or other missing two hadron components [42, 43, 44, 45], we write:

V11=V11′+αmV2(s−sth)(23)V_{11} = V'_{11} + \frac{\alpha}{m_V^2}(s - s_{\mathrm{th}}) \tag*{(23)}
V12=V12′+βmV2(s−sth)(24)V_{12} = V'_{12} + \frac{\beta}{m_V^2}(s - s_{\mathrm{th}}) \tag*{(24)}

with V11′V'_{11}, V12′V'_{12} constants, and where sths_{\mathrm{th}} is the D∗+D0D^{*+}D^0 threshold energy squared,1 α\alpha, β\beta are free parameters and mV=800 MeVm_V = 800\,\mathrm{MeV} a typical vector meson mass included to make α\alpha and β\beta dimensionless.

We should note that the formalism of using Eqs. (10) and (11) with the GG function of Eq. (9) is valid when all the involved particles are stable. In Refs. [36, 46] the D∗D^* width is considered and, as shown in Ref. [47], this is easily done replacing the GG function in Eq. (9) by:

G^i(s)=∫0Λdq Hi(q)1s+ω1+ω2×1s−ω1−ω2+is′2mD∗ΓD∗(s′)(25)\widehat{G}_i(s) = \int_0^\Lambda dq\,H_i(q)\frac{1}{\sqrt{s}+\omega_1+\omega_2}\times\frac{1}{\sqrt{s}-\omega_1-\omega_2+i\frac{\sqrt{s'}}{2m_{D^*}}\Gamma_{D^*}(s')} \tag*{(25)}

with s′=(s−ωD)2−q⃗ 2s' = (\sqrt{s} - \omega_D)^2 - \vec{q}^{\,2} and ΓD∗(s′)\Gamma_{D^*}(s') as given in Eqs. (16) and (15) of Ref. [46]. In the limit in which the total width of the vector meson goes to zero, G^i(s)\widehat{G}_i(s) reduces to Gi(s)G_i(s).

The new formalism has now more parameters: V11′V'_{11}, V12′V'_{12}, α\alpha, β\beta and Λ\Lambda plus the parameter RR if it is also unknown. In total six parameters to describe the two correlation functions in the energy range of the experiment. Once these parameters are fixed, one can calculate all properties related to the Tcc(3875)+T_{cc}(3875)^+ state:

  1. aD∗+D0a_{D^*+D^0} and aD∗0D+a_{D^{*0}D^+}.

  2. r0,D∗+D0r_{0,D^*+D^0} and r0,D∗0D+r_{0,D^{*0}D^+} from the effective range expansion kcot⁡δ(k)≃−1a+12r0k2k\cot\delta(k) \simeq-\frac{1}{a} + \frac{1}{2}r_0 k^2.

  3. The existence of a bound state and the binding energy.

  4. The couplings gD∗+D0g_{D^*+D^0}, gD∗0D+g_{D^{*0}D^+} of the state to the D∗+D0D^{*+}D^0 and D∗0D+D^{*0}D^+ channels.

  5. The width of the state, ΓTcc\Gamma_{T_{cc}}.

  6. The probabilities P1P_1, P2P_2 of the channels D∗+D0D^{*+}D^0, D∗0D+D^{*0}D^+ in the wave function of the Tcc(3875)+T_{cc}(3875)^+, and Z=1−P1−P2Z = 1-P_1-P_2 the probability of other components than the former ones in the wave function.

In order to evaluate the magnitudes enumerated above, we use the formulas of sections III and IV of Ref. [36], which we do not reproduce here. The aD∗+D0a_{D^{*+}D^{0}}, aD∗0D+a_{D^{*0}D^{+}} scattering lengths are obtained from T11T_{11} and T22T_{22} at their respective thresholds and they are complex. They are very similar to those obtained in the limit of ΓD∗→0\Gamma_{D^{*}} \to0 except that aD∗+D0a_{D^{*+}D^{0}} is real in such a case, however, Im⁡aD∗+D0\operatorname{Im} a_{D^{*+}D^{0}} with finite D∗D^{*} width is different from zero but quite smaller than Re⁡aD∗+D0\operatorname{Re} a_{D^{*+}D^{0}}. The radii and couplings are evaluated in the limit of ΓD∗→0\Gamma_{D^{*}} \to0. Then, the probabilities P1P_1, P2P_2 are given by [42, 43, 44]

P1=−gD∗+D02∂GD∗+D0∂s∣s0,(26)P_1=-g_{D^{*+}D^{0}}^{2}\left.\frac{\partial G_{D^{*+}D^{0}}}{\partial s}\right|_{s_0}, \tag*{(26)}
P2=−gD∗0D+2∂GD∗0D+∂s∣s0.(27)P_2=-g_{D^{*0}D^{+}}^{2}\left.\frac{\partial G_{D^{*0}D^{+}}}{\partial s}\right|_{s_0}. \tag*{(27)}

If P1+P2P_1+P_2 is unity or very close, this indicates that we have a molecular state made from the D∗+D0D^{*+}D^{0}, D∗0D+D^{*0}D^{+} channels.

Results

To study the inverse problem of obtaining the TT-matrix from correlation function data, and in the absence of real data for the D∗+D0D^{*+}D^{0}–D∗0D+D^{*0}D^{+} system, we generate synthetic data by taking the correlation functions computed in Ref. [29] for R=1R=1 fm and picking N=50N=50 points in the momentum variable for each channel. We take errors of 10% of the central value as is common in most present experiments. Furthermore, we take more data in the region k≲50k \lesssim50 MeV, to take into account that this is the region closest to threshold and thus more interesting. The data are shown in Figure 1, together with the fits that we carry on with the different formalisms described in the previous section. For each formalism, we perform a best fit with MINUIT [48, 49], and obtain the errors of the parameters (as well as of every other quantity, e.g. scattering lengths) by means of the bootstrap technique [50]. The parameter errors that we obtain in our final fits with the bootstrap method are in good agreement with those derived with MINUIT both with MIGRAD (from the χ2\chi^2 hessian) and with MINOS. In what follows, we discuss the different fits.

Comparison of the synthetic data for the $D^{*+}D^0$ and $D^{*0}D^+$ correlation functions

Figure 1. Comparison of the synthetic data for the D∗+D0D^{*+}D^0 (top) and D∗0D+D^{*0}D^+ (bottom) correlation functions with the results obtained in Subsects. III A, III B, and III C. The xx-axis variable is the c.m. momentum of each pair. The right panels show a zoom into the lower momentum region of the corresponding left ones.

Scattering length

We start with the first method described in Subsec. III A. The (real) parameters that we have are a1a_1, b2b_2, b2′b'_2, Λ\Lambda, and RR. The best fit has χ2/d.o.f.=0.08\chi^2/\mathrm{d.o.f.}=0.08.2 The parameters are given by:

a1=3.85(21) fm,(28)a_1=3.85(21)\,\mathrm{fm}, \tag*{(28)}
b2=−0.3149(26) fm−1,(29)b_2=-0.3149(26)\,\mathrm{fm}^{-1}, \tag*{(29)}
b2′=0.410(56),(30)b'_2=0.410(56), \tag*{(30)}
Λ=375(26) MeV,(31)\Lambda=375(26)\,\mathrm{MeV}, \tag*{(31)}
R=1.038(25) fm.(32)R=1.038(25)\,\mathrm{fm}. \tag*{(32)}

The results for the obtained correlation functions are shown in Figure 1, where the good agreement with the synthetic data can be observed. The scattering length a1a_1 in this method is directly a free parameter, but we can also obtain the scattering length a2a_2 [via Eq. (14)], which is shown in Table I. These results should be compared with those extracted from the model employed in Ref. [31] in the analysis of the data in the limit ΓD∗→0\Gamma_{D^{*}} \to0, which are also shown in Table I. As we can see in Table I, the values of aD∗+D0a_{D^{*+}D^{0}} (a1a_1) and aD∗0D+a_{D^{*0}D^{+}} (a2a_2) can be considered only in rough qualitative agreement with the experiment. It is also rewarding to see that the method provides the value of RR assumed for the data with great precision.3

Results with the K-matrix formalism

We now fit the synthetic data employing the TT-matrix described in Subsec. III B, taking as free (and real) parameters K11K_{11}, K12K_{12}, K22K_{22}, Λ\Lambda, and RR. The fit has χ2/d.o.f.=0.09\chi^2/\mathrm{d.o.f.}=0.09, and the comparison between the obtained correlation functions with the synthetic data is shown in Figure 1. For the parameters we obtain the values:

K11=1368(110),(33)K_{11}=1368(110), \tag*{(33)}
K22=1227(40),(34)K_{22}=1227(40), \tag*{(34)}
K12=−1308(55),(35)K_{12}=-1308(55), \tag*{(35)}
R=0.97(4) fm,(36)R=0.97(4)\,\mathrm{fm}, \tag*{(36)}
Λ=342(38) MeV.(37)\Lambda=342(38)\,\mathrm{MeV}. \tag*{(37)}
Modela1a_1 [fm]r1r_1 [fm]a2a_2 [fm]r2r_2 [fm]
Subsec. III A3.85(25)—2.837(98) −i-i0.972(92)—
Subsec. III B8.22(43)—1.600(170) −i-i1.206(62)—
Subsec. III C7.83(26)(31) −i-i1.84(13)(15)−2.840-2.840(47)(280)2.084(61)(62) −i-i1.219(51)(97)−0.08-0.08(12)(30) −i-i2.56(23)(33)
Exp. [30, 31]7.10(51) −i-i1.85(28)—1.760(300) −i-i1.820(300)—
ΓD∗→0\Gamma_{D^*} \to0 [30, 31]6.13(51)−3.52-3.52(50)1.710(300) −i-i1.070(300)0.26(30) −i-i3.27(30)

Table 1. Comparison of the D∗+D0D^{*+}D^0 (a1a_1, r1r_1) and D∗0D+D^{*0}D^+ (a2a_2, r2r_2) scattering lengths and effective ranges, as obtained with the different methods employed to fit the synthetic correlation function data in Figure 1, as well as with their experimental determinations. In the last row we also show the results of the experimental analysis model in the limit ΓD∗→0\Gamma_{D^*} \to0, which is the limit in which we compute the effective ranges r1,2r_{1,2}. For the results of Subsec. III C, the first (second) uncertainty represent the statistical (systematic) one, computed as described in the text.

We note that K11≃K22K_{11} \simeq K_{22}, as expected from isospin symmetry. On the other hand, the cutoff is around 2σ2\sigma with respect to the original value, Λ=418\Lambda= 418 MeV. The model of Subsec. III B for the TT-matrix is different from that used to generate the synthetic data, and, furthermore, the former does not depend on the cutoff. Therefore, the dependence of G~\widetilde{G} on Λ\Lambda tends to compensate this effect. Since the dependence is mild, because the j0(qr)j_0(qr) in G~\widetilde{G} already acts as a regulator inEq. (6), this somewhat large deviation of Λ\Lambda is reasonable.

We can also compute the scattering lengths and the Tcc(3875)+T_{cc}(3875)^+ binding energy, which are shown in Tables I and II, respectively. As can be seen, there is a clear improvement with respect to the previous method in the scattering length determinations, which are now in better agreement with the experimental ones, in particular a1a_1. The binding energy is also close to the experimental one, although around 3σ3\sigma away from it.

One could think of improving the parameterization by including higher-order (p2p^2) terms in the KK-matrix elements and/or imposing isospin symmetry (forcing K11=K22K_{11}=K_{22}). These modifications would lead to parameterizations closer to the full model and, since we employ the latter in the next section, we refrain here from doing such an exercise.

ModelBB [keV]Γ\Gamma [keV]
Subsec. III B218(17)—
Subsec. III C341(16)(11)38(2)(1)
Exp. [30, 31]360(40)48(2)

Table 2. Comparison of the Tcc(3875)+T_{cc}(3875)^{+} binding energy (BB) and width (Γ\Gamma), as obtained with the different methods employed to fit the synthetic correlation function data in Figure 1, as well as with their experimental determinations. For the results of Subsec. III C, the first (second) uncertainty represent the statistical (systematic) one, computed as described in the text.

Results with the general method

We discuss now the results obtained with the full method, described in Subsec. III C. In a first step, we consider as free parameters V11′V'_{11}, V12′V'_{12}, α\alpha, β\beta, Λ\Lambda, and RR.5 From the fits obtained with this set of free parameters, two issues can be observed. First, the parameters α\alpha and β\beta are difficult to constrain, and furthermore, they are compatible with zero.6 Therefore, without any loss of generality, we can neglect them. Second, the cutoff Λ\Lambda is largely correlated (with a correlation coefficient r≃0.99r \simeq0.99) with the parameters V11′V'_{11} and V12′V'_{12} and, therefore, they cannot be simultaneously fitted in a proper manner. This correlation is of course well known, and it corresponds to the fact that, just taking a single I=0I=0 channel for simplicity, one would have T−1=(V11−V12)−1−GT^{-1}=(V_{11}-V_{12})^{-1}-G, and there is always the possibility of a trade off between V11−V12V_{11}-V_{12} and Λ\Lambda, such that choosing δ(V11−V12)=δG\delta(V_{11}-V_{12})=\delta G at the pole one does not change its position. Then, the amplitude close to the pole position (which in turn is close to the threshold) barely changes. The dependence on Λ\Lambda appears not only in the TT-matrix, as just discussed, but also in the G~i(s,r)\widetilde{G}_i(s,r) functions. However, the dependence of the latter on the cutoff is mild, as shown in Refs. [23, 29].

As a consequence of the previous considerations, in a second step we set α=β=0\alpha=\beta=0, and take a fixed value for the cutoff, namely Λ=418\Lambda=418 MeV (which is the value used in Ref. [46] to compute the TT-matrix, from which

we have generated the synthetic data), such that the free parameters are now V11′V'_{11}, V12′V'_{12}, and RR. This final set of parameters are fitted to the synthetic data, and the results of the fit are shown in Figure 1, compared with the data and the results obtained with the rest of the formalisms. The best fit has χ2/d.o.f.∼10−4\chi^2/\mathrm{d.o.f.}\sim10^{-4}, since this is a model for the TT-matrix almost identical to the one employed to generate the synthetic data. The parameter values are:

VI=0=V11′−V12′=−435.88(88) ,(38)V_{I=0}=V'_{11}-V'_{12}=-435.88(88)\,, \tag*{(38)}
VI=1=V11′+V12′=600(330) ,(39)V_{I=1}=V'_{11}+V'_{12}=600(330)\,, \tag*{(39)}
R=1.008(24)(11) fm ,(40)R=1.008(24)(11)\ \mathrm{fm}\,, \tag*{(40)}

with a correlation matrix:

\begin{pmatrix} 1 & -0.65 & 0.95 \\ & 1 & -0.70 \\ & & 1 \end{pmatrix}\,, \tag*{(41)}\end{pmatrix}

being rows and columns ordered as in Eq. (25).

As discussed earlier, we have taken a fixed value of the cutoff Λ\Lambda, but we nonetheless take into consideration the uncertainty induced by the value of Λ\Lambda in the following manner. We perform fits of the free parameters to the synthetic data for values of Λ\Lambda in the range [350, 500] MeV. We obtain the maximum and the minimum value of every quantity, such as the scattering lengths or the Tcc(3875)+T_{cc}(3875)^{+} binding energy, and we take half the difference between both values as an estimation of a systematic uncertainty. Thus, in this section, whenever two uncertainties are given [for instance in Eq. (25c)], the first one refers to the statistical uncertainty obtained with the bootstrap method, and the second one the systematic uncertainty from the cutoff variation just described. This systematic uncertainty is not shown for VI=0,1V_{I=0,1} [Eq. (25a) and (39)] because, as has been also discussed, the correlation of these parameters with Λ\Lambda is almost one. Therefore, we just quote the fitted values that we obtain for Λ=350\Lambda=350 MeV, VI=0=−532V_{I=0}=-532 and VI=1=300V_{I=1}=300, and for Λ=500\Lambda=500 MeV, VI=0=−355V_{I=0}=-355 and VI=1=−2001V_{I=1}=-2001. As said, the spread here is much larger than the statistical uncertainty.

As can be seen [Eqs. (25a) and (39)], the isoscalar matrix element VI=0V_{I=0} is much more constrained than its isovector counterpart VI=1V_{I=1}. This is due to the isoscalar nature of the Tcc(3875)+T_{cc}(3875)^{+} state. The pole of this state, being below and very close to the threshold, enhances the correlation function near the D∗+D0D^{*+}D^0 threshold, such that CD∗+D0(0)≃30C_{D^{*+}D^0}(0)\simeq30. We note that, as discussed right after Eq. (5), we are able to recover in Eq. (25c) the original value R=1R=1 fm for the size of the source function with which the synthetic data were generated. The large correlation between RR and VI=0V_{I=0} in Eq. (41) can be easily understood. In Ref. [23] it was shown that, for a single channel, in the Lednicky-Lyuboshits approximation [51], one obtains that the deviation from 1 of C(k=0)C(k=0) depends on the ratio a/Ra/R, with the scattering length, which is determined from the matrix element VI=0V_{I=0}. In a system different from D∗DD^{*}D, with a smaller scattering length or, equivalently, without the prominent enhancement induced by the Tcc(3875)+T_{cc}(3875)^{+}, we would expect this correlation to be smaller. Furthermore, we have tested that using MINOS, which estimates the (negative and positive) errors of a parameter by varying it and fitting the other free parameters until the χ2\chi^{2} deviates from the minimum in one unit, we get uncertainties very similar to the statistical one shown in Eq. (25c), obtained from the bootstrap method. Likewise, the (parabolic) error obtained from the χ2\chi^{2} hessian (with MIGRAD) is very similar, too. Therefore, despite the large correlation coefficient between RR and VI=0V_{I=0} in Eq. (41), the method proposed here allows one to determine RR with about 2–3% of statistical uncertainty. Hence, we expect that this will also be the case with real data.

With this TT-matrix parameterization we can also compute the scattering lengths and effective ranges,7 shown in Table I, as well as the Tcc(3875)+T_{cc}(3875)^{+} binding energy and its width, shown in Table II. It can be seen that there is quite good agreement with the experimental determination [30, 31] of the same quantities, also shown in Tables I and II. We also note that the method allows for a determination of the uncertainty of these quantities similar or smaller than the experimental ones. Shall the D∗+D0D^{*+}D^{0} and D∗0D+D^{*0}D^{+} correlation functions be measured with an uncertainty of around 10%, as we have imposed in our synthetic data, the D∗+D0D^{*+}D^{0} and D∗0D+D^{*0}D^{+} interactions, as well as the Tcc(3875)+T_{cc}(3875)^{+} parameters, could be greatly constrained.

We can also compute the Tcc(3875)+T_{cc}(3875)^{+} couplings to the two channels, obtaining:

g1=3853(36)(33) MeV,(42)g_{1} = 3853(36)(33)\ \mathrm{MeV}, \tag*{(42)}
g2=−4131(73)(35) MeV,(43)g_{2} = -4131(73)(35)\ \mathrm{MeV}, \tag*{(43)}

which in turn allow for the determination of the D∗+D0D^{*+}D^{0} and D∗0D+D^{*0}D^{+} probabilities,

P1=−g12∂G1∂s∣s=s0=0.691(9)(15),(44)P_{1} = -g_{1}^{2}\left.\frac{\partial G_{1}}{\partial s}\right|_{s=s_{0}} = 0.691(9)(15), \tag*{(44)}
P2=−g22∂G2∂s∣s=s0=0.309(9)(15).(45)P_{2} = -g_{2}^{2}\left.\frac{\partial G_{2}}{\partial s}\right|_{s=s_{0}} = 0.309(9)(15). \tag*{(45)}

The values obtained above fulfill P1+P2=1P_{1}+P_{2}=1 by construction, indicating a fully molecular state [39], since we have neglected α\alpha and β\beta. We recall here that in our first step of the analysis we obtained α\alpha and β\beta values largely compatible with zero. This result, in turn, is an unavoidable consequence given the fact that the synthetic data are generated with a model [46] for the TT-matrix in which the Tcc(3875)+T_{cc}(3875)^{+} state turns out to be almost purely molecular (similar results are obtained in Refs. [37, 52, 53]). However, this is not the point that we want to emphasize

here, rather, the possibility of obtaining these quantities with relative fine precision, say of a few percent, shall these correlation functions be measured. This proves the ability of the method to extract the nature of states from the analysis of correlation data with great precision.

We would like to mention here that in the first step of the calculations where α\alpha and β\beta are free parameters, the results that we obtain for the observables are very similar to and compatible within errors with those in Table I. In the case where α\alpha and β\beta are free, P1+P2P_{1}+P_{2} is not exactly 1 and one obtains P1+P2=1.002(49)P_{1}+P_{2}=1.002(49), which is indeed compatible with P1+P2=1P_{1}+P_{2}=1, and the uncertainty is a little bigger than the sum in quadratures of the errors of P1P_{1} and P2P_{2} in Eqs. (28).

Conclusions

In this work we investigated the potential of the inverse problem in femtoscopic correlation functions, in which starting from the correlation functions of some related channels we determine the interaction between these channels. The data are analyzed by means of a coupled channel unitary scheme which has the freedom to accommodate missing channels relevant to the interaction, as well as components of some genuine states of non-molecular nature. This leaves a few parameters to be determined from the data, together with the parameter that gives the source size. We applied the method to pseudo-data of the D∗+D0D^{*+}D^{0} and D∗0D+D^{*0}D^{+} correlation functions and extracted from them all the possible information. The results obtained are remarkable and we found that using these data, all above the threshold of the two channels, we could determine that there was a bound state of the system, and that this state was of molecular nature, with negligible room for a non-molecular state. We determined the scattering length and the effective range for the two channels, the probabilities to have each of the channels in the wave function of the bound state and also, remarkably, the size of the source parameter, RR, all of them with a good precision and in agreement with the experimental information used to constrain the TT-matrix employed in the generation of the synthetic data (correlation functions).

In order to better appreciate the precision in the magnitudes obtained with the general method, we also used two other different methods, one very simple which does not require to use any intermediate interaction potentials between the channels, and which provides the scattering amplitudes at a qualitative level. Another one, based on the common KK-matrix approach, provides much better results for the scattering amplitudes, but however does not allow to obtain the molecular probabilities which rely upon ∂G/∂s\partial G/\partial s, and Re⁡G\operatorname{Re} G is not fully accounted for in the KK-matrix approach.

In conclusion, the full method proposed here is ideal to analyze correlation functions and extract from them valuable information on hadron interactions, which will be useful in the future to study the interaction of hadron states which cannot be formed otherwise, as well as to determine the nature of bound states or resonances.

Acknowledgments

This work was supported by the Spanish Ministerio de Ciencia e Innovación (MICINN) and European FEDER funds under Contracts No. PID2020-112777GB-I00, and by Generalitat Valenciana under contract PROMETEO/2020/023. This project has received funding from the European Union Horizon 2020 research and innovation programme under the program H2020-INFRAIA-2018-1, grant agreement No. 824093 of the STRONG-2020 project. M. A. and A. F. are supported through Generalitat Valencia (GVA) Grants Nos. CIDEGENT/2020/002 and APOSTD-2021-112, respectively. M. A. and A. F. thank the warm support of ACVJLI.

References

References

  1. [1]L. Adamczyk et al. (STAR), Phys. Rev. Lett. 114, 022301 (2015), arXiv:1408.4360 [nucl-ex].
  2. [2]S. Acharya et al. (ALICE), Phys. Lett. B 774, 64 (2017), arXiv:1705.04929 [nucl-ex].
  3. [3]J. Adam et al. (STAR), Phys. Lett. B 790, 490 (2019), arXiv:1808.02511 [hep-ex].
  4. [4]S. Acharya et al. (ALICE), Phys. Rev. C 99, 024001 (2019), arXiv:1805.12455 [nucl-ex].
  5. [5]S. Acharya et al. (ALICE), Phys. Rev. Lett. 123, 112002 (2019), arXiv:1904.12198 [nucl-ex].
  6. [6]S. Acharya et al. (ALICE), Phys. Lett. B 797, 134822 (2019), arXiv:1905.07209 [nucl-ex].
  7. [7]S. Acharya et al. (ALICE), Phys. Lett. B 805, 135419 (2020), arXiv:1910.14407 [nucl-ex].
  8. [8]S. Acharya et al. (ALICE), Phys. Rev. Lett. 124, 092301 (2020), arXiv:1905.13470 [nucl-ex].
  9. [9]A. Collaboration et al. (ALICE), Nature 588, 232 (2020), [Erratum: Nature 590, E13 (2021)], arXiv:2005.11495 [nucl-ex].
  10. [10]S. Acharya et al. (ALICE), Phys. Lett. B 822, 136708 (2021), arXiv:2105.05683 [nucl-ex].
  11. [11]S. Acharya et al. (ALICE), Phys. Rev. Lett. 127, 172301 (2021), arXiv:2105.05578 [nucl-ex].
  12. [12]L. Fabbietti, V. Mantovani Sarti, and O. Vazquez Doce, Ann. Rev. Nucl. Part. Sci. 71, 377 (2021), arXiv:2012.09806 [nucl-ex].DOI
  13. [13]K. Morita, T. Furumoto, and A. Ohnishi, Phys. Rev. C 91, 024916 (2015), arXiv:1408.6682 [nucl-th].DOI
  14. [14]A. Ohnishi, K. Morita, K. Miyahara, and T. Hyodo, Nucl. Phys. A 954, 294 (2016), arXiv:1603.05761 [nucl-th].DOI
  15. [15]K. Morita, A. Ohnishi, F. Etminan, and T. Hatsuda, Phys. Rev. C 94, 031901 (2016), [Erratum: Phys.Rev.C 100, 069902 (2019)], arXiv:1605.06755 [hep-ph].
  16. [16]T. Hatsuda, K. Morita, A. Ohnishi, and K. Sasaki, Nucl. Phys. A 967, 856 (2017), arXiv:1704.05225 [nucl-th].DOI
  17. [17]D. L. M. Matouschek, V. Mantovani Sarti, O. W. Arnold, L. Fabbietti, B. Hohlweger, and A. M. Mathis, Eur. Phys. J. C 78, 394 (2018), arXiv:1802.08481 [hep-ph].
  18. [18]J. Haidensbauer, Nucl. Phys. A 981, 1 (2019), arXiv:1808.05049 [hep-ph].
  19. [19]K. Morita, S. Gongyo, T. Hatsuda, T. Hyodo, Y. Kamiya, and A. Ohnishi, Phys. Rev. C 101, 015201 (2020), arXiv:1908.05414 [nucl-th].DOI
  20. [20]Y. Kamiya, T. Hyodo, K. Morita, A. Ohnishi, and W. Weise, Phys. Rev. Lett. 124, 132501 (2020), arXiv:1911.01041 [nucl-th].DOI
  21. [21]Y. Kamiya, K. Sasaki, T. Fukui, T. Hyodo, K. Morita, K. Ogata, A. Ohnishi, and T. Hatsuda, Phys. Rev. C 105, 014915 (2022), arXiv:2108.09644 [hep-ph].DOI
  22. [22]Z.-W. Liu, J.-X. Lu, and L.-S. Geng, Phys. Rev. D 107, 074019 (2023), arXiv:2302.01046 [hep-ph].DOI
  23. [23]M. Albaladejo, J. Nieves, and E. Ruiz-Arriola, (2023), arXiv:2304.03107 [hep-ph].DOI
  24. [24]N. Ikeno, K. Toledо, and E. Oset, (2023), arXiv:2305.16431 [hep-ph].arxiv.org/abs/2305.16431
  25. [25]J. M. Torres-Rincon, A. Ramos, and L. Tolos, (2023), arXiv:2307.02102 [hep-ph].
  26. [26]Z.-W. Liu, J.-X. Lu, M.-Z. Liu, and L.-S. Geng, (2023), arXiv:2305.19048 [hep-ph].arxiv.org/abs/2305.19048
  27. [27]S. Acharya et al. (ALICE), Phys. Rev. D 106, 052010 (2022), arXiv:2101.05352 [nucl-ex].
  28. [28]ALICE Collaboration, (2022), arXiv:2211.02491 [physics.ins-det].
  29. [29]I. Vidaña, A. Feijoo, M. Albaladejo, J. Nieves, and E. Oset, (2023), arXiv:2303.06079 [hep-ph].arxiv.org/abs/2303.06079
  30. [30]R. Aaij et al. (LHCb), Nature Phys. 18, 751 (2022), arXiv:2109.01038 [hep-ex].
  31. [31]R. Aaij et al. (LHCb), Nature Commun. 13, 3351 (2022), arXiv:2109.01056 [hep-ex].
  32. [32]S. E. Koonin, Phys. Lett. B 70, 43 (1977).
  33. [33]S. Pratt, T. Csorgo, and J. Zimanyi, Phys. Rev. C 42, 2646 (1990).
  34. [34]W. Bauer, C. K. Gelbke, and S. Pratt, Ann. Rev. Nucl. Part. Sci. 42, 77 (1992).
  35. [35]Y. Kamiya, T. Hyodo, and A. Ohnishi, Eur. Phys. J. A 58, 131 (2022), arXiv:2203.13814 [hep-ph].DOI
  36. [36]L. R. Dai, L. M. Abreu, A. Feijoo, and E. Oset, (2023), arXiv:2304.01870 [hep-ph].arxiv.org/abs/2304.01870
  37. [37]M. Albaladejo and J. Nieves, Eur. Phys. J. C 82, 724 (2022), arXiv:2203.04864 [hep-ph].DOI
  38. [38]J. A. Oller, E. Oset, and J. R. Pelaez, Phys. Rev. D 59, 074001 (1999), [Erratum: Phys.Rev.D 60, 099906 (1999), Erratum: Phys.Rev.D 75, 099903 (2007)], arXiv:hep-ph/9804209.arxiv.org/abs/hep-ph/9804209
  39. [39]D. Gamermann, J. Nieves, E. Oset, and E. Ruiz Arriola, Phys. Rev. D 81, 014029 (2010), arXiv:0911.4407 [hep-ph].DOI
  40. [40]J. Song, L. R. Dai, and E. Oset, Eur. Phys. J. A 58, 133 (2022), arXiv:2201.04414 [hep-ph].DOI
  41. [41]F. Mandl and G. Shaw, Quantum Field Theory (1985).
  42. [42]T. Hyodo, Int. J. Mod. Phys. A 28, 1330045 (2013), arXiv:1310.1176 [hep-ph].DOI
  43. [43]F. Aceti, L. R. Dai, L. S. Geng, E. Oset, and Y. Zhang, Eur. Phys. J. A 50, 57 (2014), arXiv:1301.2554 [hep-ph].
  44. [44]A. Martínez Torres, E. Oset, S. Prelovsek, and A. Ramos, JHEP 05, 153 (2015), arXiv:1412.1706 [hep-lat].arxiv.org/abs/1412.1706
  45. [45]V. Montesinos, M. Albaladejo, J. Nieves, and L. Tolos, (2023), arXiv:2306.17673 [hep-ph].arxiv.org/abs/2306.17673
  46. [46]A. Feijoo, W. H. Liang, and E. Oset, Phys. Rev. D 104, 114015 (2021), arXiv:2108.02730 [hep-ph].arxiv.org/abs/2108.02730
  47. [47]L. R. Dai, E. Oset, A. Feijoo, R. Molina, L. Roca, A. M. Torres, and K. P. Khemchandani, Phys. Rev. D 105, 074017 (2022), [Erratum: Phys.Rev.D 106, 099904 (2022)], arXiv:2201.04840 [hep-ph].DOI
  48. [48]F. James and M. Roos, Comput. Phys. Commun. 10, 343 (1975).
  49. [49]F. James, “MINUIT Function Minimization and Error Analysis: Reference Manual Version 94.1,” (1994).
  50. [50]B. Efron and R. Tibshirani, Statist. Sci. 57, 54 (1986).
  51. [51]R. Lednicky and V. L. Lyuboshits, Yad. Fiz. 35, 1316 (1981), Sov. J. Nucl. Phys. 35, 770 (1982).
  52. [52]M. Albaladejo, Phys. Lett. B 829, 137052 (2022), arXiv:2110.02944 [hep-ph].DOI
  53. [53]M.-L. Du, V. Baru, X.-K. Dong, A. Filin, F.-K. Guo, C. Hanhart, A. Nefediev, J. Nieves, and Q. Wang, Phys. Rev. D 105, 014024 (2022), arXiv:2110.13765 [hep-ph].DOI

Paper details

vlnc:W_9tdh3fmp
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