1S nucleon-nucleon scattering in the modified Weinberg approach 0 E. Epelbaum,1 A. M. Gasparyan,1,2 J. Gegelia,1,3 and H. Krebs1 1Institut fu¨r Theoretische Physik II, Fakult¨at fu¨r Physik und Astronomie, Ruhr-Universit¨at Bochum 44780 Bochum, Germany 2SSC RF ITEP, Bolshaya Cheremushkinskaya 25, 117218 Moscow, Russia 3Tbilisi State University, 0186 Tbilisi, Georgia (Dated: 12 July, 2014) 5 Abstract 1 0 Nucleon-nucleon scattering in the 1S partial wave is considered in chiral effective field theory 0 2 within the renormalizable formulation of Ref. [1] beyond the leading-order approximation. By n applying subtractive renormalization, the subleading contact interaction in this channel is taken a J into account non-perturbatively. For a proper choice of renormalization conditions, the predicted 6 energy dependenceof the phase shiftand the coefficients in the effective range expansion are found to be in a good agreement with the results of the Nijmegen partial wave analysis. ] h t - PACS numbers: 11.10.Gh,12.39.Fe,13.75.Cs l c u n [ 1 v 1 9 1 1 0 . 1 0 5 1 : v i X r a 1 I. INTRODUCTION The seminal work of Weinberg [2] has triggered a renewed interest to the nuclear force problem in the framework of effective field theory (EFT). In this approach, nuclear forces are defined as kernels of the corresponding dynamical equations and can be derived order- by-order making use of the systematic chiral expansion. Starting from the pioneering work of Ref. [3], this approach has developed rapidly and is nowadays commonly employed in studies of low-energy few- and many-nucleon dynamics and nuclear structure calculations, see [4–6] for recent review articles. While offering many attractive features, Weinberg’s approach was criticized for being non-renormalizable. The main difficulty is related to the fact that iterations of the truncated NN potential within the Lippmann-Schwinger (LS) equation generate ultraviolet (UV) divergencies which cannot be absorbed by counter terms (contact interactions) included in the truncated potential. In particular, infinitelymanycountertermsareneededtoabsorbUVdivergencesemergingfrom iterations of the leading-order (LO) one-pion exchange (OPE) potential [7]. This feature is sometimes referred to as inconsistency of Weinberg’s approach. TheUVcutoffΛcanberemoved fromtheLSequationbyenforcing thelimitΛ → ∞non- perturbatively, see e.g. [8, 9]. It is possible to obtain a finite, manifestly non-perturbative solution of the LS equation with a singular 1/r3-potential by including one/no contact operator in each attractive/repulsive channel [9]. However, such a procedure is incompatible with the principles of EFT which require that all UV divergences emerging from iterations of the LS equation are absorbed by counter terms [10]. It is not surprising that such an approach fails to reproduce experimental data even at N3LO [11]. Treating the exchange of pions perturbatively as suggested by Kaplan, Savage and Wise (KSW) [12] allows one to avoid the above-mentioned inconsistency. However, the perturba- tive series fail to converge within this framework [13–16]. Presently, there exist different views and formulations of organizing the chiral expansion in the few-nucleon sector [8–10, 17–32]. A novel approach to the NN scattering problem in EFT has been formulated in Refs. [1, 33–36] and is referred to as the modified Weinberg ap- proach. Within this framework, the leading order (LO) NN scattering amplitude is obtained bysolvingtheKadyshevsky equation[37]fortheLOpotentialconsisting ofthecontactinter- action part and the OPE potential. This equation provides an example of three-dimensional integral equations which satisfy relativistic elastic unitarity. An important feature of the Kadyshevsky equation is that it is renormalizable for the LO potential, i.e. all ultraviolet divergences generated by iterations can be explicitly absorbed into redefinition of the NN derivative-less contact interaction. The scattering amplitude can still be renormalized if higher-order corrections to the potential are taken into account perturbatively. If higher order corrections to the potential indeed provide small contributions to the amplitude, their perturbative and non-perturbative inclusions are expected to lead to small differences in the results which are beyond the accuracy one is working at. However, this observation is only meaningful if a proper renormalization is carried out in both cases. In general, we are not able to subtract all divergences from amplitudes if higher-order contributions in the potential are treated non-perturbatively. In the 1S partial wave, one observes a very large 0 discrepancy between the LO EFT results and the experimental data already at rather low energies [1]. This large discrepancy signals that at least a part of the higher-order contribu- tions in the effective potential is likely to require a non-perturbative treatment within our 2 approach.1 In this paper we study in detail the role of the next-to-leading order (NLO) short- range contribution to the potential which can be included both perturbatively and non- perturbatively. Specifically, we will express the solution to the integral equation in a closed form following the lines of Ref. [40] and apply the BPHZ-type subtractive renormalization [41]. After subtracting all ultraviolet divergences, we will calculate the remaining finite expressions numerically, fit the available two low-energy constants (LECs) to the data and compare the obtained results with the phase shifts for various choices of the renormalization scale parameter. Here and in what follows, the resulting NN amplitude will be referred to as NLO as opposed to the LO result of Ref. [1]. A more complete calculation including the corresponding two-pion exchange potential to first order in perturbationtheory is postponed for a future study. Our paper is organized as follows: In section II we provide the formal expression for the scattering amplitude by making use of the standard two-potential formalism. Subtrac- tive renormalization of the amplitude is discussed in detail in section III. Next, section IV addresses the issue of the appropriate choice of the renormalization conditions (i.e. subtrac- tion scale) and also presents the results of our calculation. Our findings are summarized in section V. II. FORMAL EXPRESSION FOR THE SCATTERING AMPLITUDE In the framework of the modified Weinberg approach, the NLO 1S partial wave NN 0 scatteringamplitude2 canbeobtainedbyextractingtheS-wavecomponentfromthesolution to the integral equation (for the fully off-shell amplitude T) T (p ,p~′,~p) = V (~p′,~p)+ d3k V ~p′,~k G(p ,k) T p ,~k,p~ , (2.1) 0 0 0 Z (cid:16) (cid:17) (cid:16) (cid:17) m2 1 G(p ,k) = , (2.2) 0 2(2π)3 ~k2 +m2 p − ~k2 +m2 +iǫ 0 (cid:16) (cid:17)(cid:16) p (cid:17) ′ where~p(p~ )istheincoming(outgoing)three-momentumofthenucleoninthecenter-of-mass frame, p = ~q2 +m2 with m denoting the nucleon mass and ~q being the corresponding 0 three-momenptum of an incoming (on-mass-shell) nucleon. Further, the potential is given by g2M2 1 V (p~′,~p) = C +C ~p′2 +p~2 − A π 2 4F2 (p~′ −~p)2 +M2 (cid:2) (cid:0) (cid:1)(cid:3) π π ≡ V +V , C π g2 C = C −3C + A +DM2. (2.3) S T 4F2 π π Here g , F and M are the nucleon axial-vector coupling, pion decay constant and the pion A π π mass, respectively. The parameters C , C , C and D refer to the LECs of the effective S T 2 1 Notice, however, that the LO calculations reported in Refs. [38, 39] within the standard nonrelativistic approach using a coordinate-space regularization for the OPE potential yields a superior description of the phase shift. 2 Note the different overall sign in comparison with the Feynman amplitude considered in Ref. [1]. 3 Lagrangian. Below, we work with the S-wave component of Eq. (2.1) and denote the 1S 0 ′ partial wave projected OPE potential by V (p,p) with π g2M2 (p−p′)2 +M2 V (p′,p) = A π ln π . (2.4) π 16F2pp′ (p+p′)2 +M2 π π ′ For the analysis of divergent integrals, it is useful to have the asymptotics of V (p,p) at π large values of momenta g2M2 g2M2 V (p′,p) ≈ − A π , V (p+l,p) ≈ − A π lnp. (2.5) π p→∞,p′<∞ 4F2p2 π p→∞,|l|<∞ 8F2p2 (cid:12) π (cid:12) π (cid:12) (cid:12) The contact-interaction part of the potential V is separable. Therefore, it is possible C to write the solution to Eq. (2.1) in a form, which allows one to carry out the subtractive renormalization explicitly. This can be achieved by making use of the well-known two- potential formalism. For this purpose, we write Eq. (2.1) symbolically as T = V +V GT, (2.6) and express its solution as T = T +(1+T G)T (1+GT ), (2.7) π π C π where T and T satisfy the equations π C T = V +V GT , (2.8) π π π π T = V +V G(1+T G)T . (2.9) C C C π C For a separable contact-interaction potential, V (p′,p) = ξ(p′)T Cξ(p), (2.10) C where C and ξ(p) are 2 × 2 and 2 × 1 matrices, respectively, whose explicit form will be specified below, the solution to Eq. (2.9) is also given in a separable form T (p ,p′,p) = ξT(p′)Xξ(p), (2.11) C 0 where X is a 2×2 matrix, X = C−1 −Σ −1 , (2.12) (cid:2) (cid:3) and the 2×2 “selfenergy” matrix Σ reads Σ(p ) = ξGξT +ξGT GξT (2.13) 0 π ≡ d3k ξ(k)G(p ,k)ξT(k)+ d3k d3k ξ(k )G(p ,k )T (p ,k ,k )G(p ,k )ξT(k ). 0 1 2 1 0 1 π 0 1 2 0 2 2 Z Z Thus, the final expression of the amplitude T has the form T = T +ΞT X Ξ. (2.14) π with Ξ(p ,p) = ξ(1+GT ) ≡ ξ(p)+ d3kξ(k)G(p ,k)T (p ,k,p) . (2.15) 0 π 0 π 0 Z 4 ... = + + + = + + + ... + = + + + ... FIG. 1: Building blocks of the scattering amplitude. The first, second and third lines represent T , π Ξ and Σ, respectively. The solid and dashed lines correspond to nucleons and pions, respectively. The filled circles represent ξ and ξT. III. RENORMALIZATION OF THE SCATTERING AMPLITUDE The expression for the scattering amplitude in Eq. (2.14) contains UV divergences. We perform renormalization by applying the BPHZ procedure, i.e. we subtract all divergences and sub-divergences of the loop diagrams and replace the LECs with their renormalized, finite values. In general, in renormalizable theories, subtractive renormalization can be realized by counter terms in the Lagrangian. Chiral effective field theory is renormalizable in the sense of effective field theories, i.e. all divergences can be absorbed into redefinition of aninfinite number of counter terms. To realize subtractive renormalizationin theconsidered problem, we would need to include the contributions of an infinite number of counter terms of the effective Lagrangian. Although this is possible for the case at hand by considering energy-dependent counter terms, here we only show explicitly one momentum- and energy- independent counter term δz and write the contact interaction potential in a separable form V (p′,p) = C +C p′2 +p2 = 1, p′2 +δz C˜ C2 1 . (3.1) C 2 (cid:18)C 0 (cid:19)(cid:18) p2 +δz (cid:19) 2 (cid:0) (cid:1) (cid:0) (cid:1) The new parameter is expressed as ˜ C = C −2C δz. (3.2) 2 Thus, the contact-interaction potential has the form C = C˜ C2 , ξ(p) ≡ (ξ (p),ξ (p))T = 1, p2 +δz T . (3.3) 1 2 (cid:18)C 0 (cid:19) 2 (cid:0) (cid:1) The various terms contributing to the amplitude T are visualized diagrammatically in Fig. 1 in terms of the corresponding building blocks, where in the first line the amplitude T π is shown. The second line represents Ξ, while the analogous diagrams for ΞT are not shown explicitly. The third line depicts the quantity Σ which contributes to X, see Eq. (2.12). To obtain the amplitude using Eq. (2.14), we first perform subtractive renormalization and 5 afterwards calculate numerically the remaining finite expressions for the quantities T , Ξ, π and X. In the following, we describe in detail how these quantities are renormalized. Since the amplitude T is finite (the ultraviolet regularity of the equation for T (2.8) follows from π π the asymptotics (2.5)), we begin our discussion with the subtractive renormalization of Ξ. By writing Ξ(p ,p) as a perturbative series as shown in Fig. 1, 0 Ξ = ξ +ξGV +ξGV GV +··· , (3.4) π π π it is easily seen that Ξ(p ,p) = (Ξ (p ,p),Ξ (p ,p))T can be obtained by solving the integral 0 1 0 2 0 equation Ξ = ξ +ΞGV . (3.5) π This expression defines a system of equations for the quantities Ξ (p ,p), which, using the 1,2 0 explicit form of ξ(p) from Eq. (3.3), can be written as Ξ (p ,p) = 1+ d3kΞ (k)G(p ,k)V (k,p), (3.6) 1 0 1 0 π Z Ξ (p ,p) = p2 +δz + d3kΞ (k)G(p ,k)V (k,p). (3.7) 2 0 2 0 π Z The equation for Ξ (p ,p) is free of ultraviolet divergences, see Eq. (2.5), and has an ul- 1 0 p→∞ traviolet behavior Ξ (p ,p) ≈ const. On the other hand, to identify the divergences in 1 0 Ξ (p ,p), it is convenient to consider iterations of Eq. (3.7) 2 0 Ξ (p) = p2 +δz + d3k(k2 +δz)G(p ,k)V (k,p) 2 0 π Z + d3kd3l(k2 +δz)G(p ,k)V (k,l)G(p ,l)V (l,p)+··· . (3.8) 0 π 0 π Z Remembering the definition of G(p ,k) in Eq. (2.2), we simplify 0 m2 p + ~k2 +m2 k2G(p ,k) = q2G(p ,k)− 0 0 0 2(2π)3 ~kp2 +m2 ≡ q2G(p ,k)+G˜(p ,k). (3.9) 0 0 Substituting the above expression into Eq. (3.8) and re-organizing the perturbative series we obtain Ξ (p ,p) = p2 +q2 d3kG(p ,k)V (k,p) 2 0 0 π Z + q2 d3kd3lG(p ,k)V (k,l)G(p ,l)V (l,p)+··· 0 π 0 π Z + δz + d3kG˜(p ,k)V (k,p) 0 π Z + d3kd3l δz +G˜(p ,k)V (k,l) G(p ,l)V (l,p)+··· . (3.10) 0 π 0 π Z h i From this equation it is easily seen that Ξ (p ,p) can be written in the form 2 0 Ξ (p ,p) = p2 +q2[Ξ (p ,p)−1]+Ψ (p ,p), (3.11) 2 0 1 0 π 0 6 where the quantity Ψ satisfies the equation π Ψ (p ,p) = ξ (p ,p)+ d3kΨ (p ,k)G(p ,k)V (k,p), (3.12) π 0 π 0 π 0 0 π Z with ξ (p ,p) = d3kG˜(p ,k)V (k,p)+δz, (3.13) π 0 0 π Z or symbolically Ψ = ξ +Ψ GV , ξ = ξ G˜V +δz. (3.14) π π π π π 1 π ˜ The GV term in ξ contains logarithmic divergence, which can be removed by adjusting π π the one-loop counter-term δz to δz = − d3kG˜(m,k)V (k,0), (3.15) π Z so that the quantities ξ , Ψ and Ξ become finite. Moreover from Eq. (2.5) it follows that π π 2 p→∞ p→∞ ξ = O(lnp), Ψ = O(lnp). π π Wenow proceedwith the renormalizationof X, which in our scheme reduces toa subtrac- tive renormalization of Σ. The term in ξGξT (the first diagram in the right-hand side of the third line of Fig. 1), which is δz-independent, contains divergences with energy-dependent coefficients. These divergences can be consistently subtracted using the BPHZ prescrip- tion. Those terms in ξGξT which contain δz linearly cancel the sub-divergences in the δz-independent part of the two-loop diagram ξGV GξT contained in the ξGT GξT part of π π the quantity Σ (second diagram in the righthand side of the third line of Fig. 1). The overall divergence of the two-loop diagram ξGV GξT requires an additional BPHZ subtraction. π Terms in ξGξT which contain δz quadratically cancel the two-loop sub-divergence in the δz-independent part of the three-loop diagram ξGV GV GξT (the last explicitly shown π π diagram in the third line of Fig. 1). In addition, all δz-dependent parts of ξGξT require an additional subtraction of overall divergences. All other divergences appearing in the loop expansion of ξGT GξT are canceled automatically by contributions of the δz counter π term. For example, the one-loop sub-divergences of the δz-independent part of the three loop diagram ξGV GV GξT are canceled by those expressions generated by the diagram π π ξGV GξT, which are linear in δz. In the following, we provide the explicit expressions π needed to compute the quantity Σ in Eq. (2.13) and define the corresponding subtractions. It is convenient to split Σ into three terms Σ = Σ +Σfinite +Σdiv. (3.16) 0 π π The term Σ contains only “pionless” contributions: 0 I (q) I (q) Σ = ξGξT| ≡ 0 2 , (3.17) 0 δz=0 (cid:18)I2(q) I4(q) (cid:19) where we have introduced the integrals {I (q),I (q),I (q)} = d3k{1,~k2,(~k2)2}G(p ,k). (3.18) 0 2 4 0 Z 7 We subtract the infinite local (polynomial in p − m) terms from these integrals to make 0 them finite, so that I (q) (i = 0,2,4) are replaced with the subtracted IR(µ,q) defined as i i m2 q IR(µ,q) = I (q)−I (iµ) = 2q sinh−1 −iπ −πm 0 0 0 8π2p m 0 h (cid:16) (cid:17) i m2 µ + 2µ sin−1 −π +πm , 8π2 m2 −µ2 h (cid:16) m (cid:17) i p IR(µ,q) = I (q)−q2I (iµ)− d3kG˜(p ,k) = q2IR(µ,q), 2 2 0 0 0 Z IR(µ,q) = I (q)−q4I (iµ)− d3k(k2 +q2)G˜(p ,k) = q4IR(µ,q). (3.19) 4 4 0 0 0 Z The subtracted integrals depend on the choice of the subtraction point µ. In principle, one has an additional freedom in fixing finite terms polynomial in p −m. However, it leads to 0 higher-order effects, therefore we will only study the µ dependence of obtained results. The remaining terms of Σ are split into the finite and divergent parts, Σfinite and Σdiv, π π which are given by 1 q2 0 1 Σfinite = ξ GV GΞ + ξ GV GΨ π (cid:18)q2 q4 (cid:19) 1 π 1 (cid:18)1 2q2 (cid:19) 1 π π 0 0 + ξ GV GΨ , (3.20) (cid:18)0 1(cid:19) π π π and 0 1 0 0 Σdiv = ξ Gξ + ξ G˜V G˜ξ +2δzξ G˜ξ +ξ Gξ . (3.21) π (cid:18)1 2q2 (cid:19) 1 π (cid:18)0 1(cid:19) 1 π 1 1 1 π π n o It is straightforward to show using Eqs. (3.9), (3.11) and (3.14) that two expressions for Σ given by Eq. (2.13) and Eq. (3.16) are identical. Note that the divergent part Σdiv contains π only afinite number of iterationsof theOPE potential (upto three loopsasshown inFig. 1). All nonperturbative effects due to OPE are included in Σfinite. π Again, following the BPHZ procedure, we subtract the infinite local terms containing overall divergencies from Σdiv of the following form π 0 1 0 0 δΣ = ξ G ξ + ξ G˜V G˜ξ +2δzξ G˜ξ +ξ G ξ , (3.22) π (cid:18)1 2q2 (cid:19) 1 0 π (cid:18) 0 1 (cid:19) 1 π 1 1 1 π 0 π n o where G (k) ≡ G(p = m,k), so that the full subtracted result for Σ reads 0 0 1 q2 0 1 ΣR = IR(µ,q)+ξ GV GΞ + ξ (G−G )ξ +ξ GV GΨ (cid:18) q2 q4 (cid:19) 0 1 π 1 (cid:18) 1 2q2 (cid:19) 1 0 π 1 π π n o n o 0 0 + ξ (G−G )ξ +ξ GV GΨ . (3.23) (cid:18) 0 1 (cid:19) π 0 π π π π n o The finiteness of ΣR can be shown using ultraviolet behavior of V , Ξ , ξ and Ψ considered π 1 π π above. Performing subtractions in the spirit of chiral effective field theory we were supposed to also expand in powers of the pion mass, which we have not done here. However, the 8 non-analytic dependence of the resulting subtraction terms on the pion mass is of a higher order relative to the accuracy of our calculation. Note also that our final perturbative result which will be discussed in section IV depends on the choice of the renormalization scheme. This dependence is also of higher order. In final finite expressions for X we substitute the finite renormalized couplings for C˜ and C 3: 2 XR = CR −1 −ΣR −1 . (3.24) h i (cid:0) (cid:1) Note that XR is not equal to X, because not all of the divergencies can be absorbed by means of redefinition of two available low-energy constants. We parameterize effectively the dependence of the result on the renormalization scheme by exploiting the freedom to choose the subtraction point µ. IV. THE CHOICE OF RENORMALIZATION CONDITIONS AND NUMERICAL RESULTS We are now in the position to specify the choice of renormalization conditions which, in the case at hand, translates into specifying the subtraction point µ. Notice that we have already made a specific choice for the subtractions of the integrals I (q) and I (q) in 2 4 Eq. (3.18). It is useful to recall the key aspects of renormalization in the simple case of pionless EFT corresponding to g = 0, see e.g. Refs. [10, 42], before dealing with the more A complicated pionfull approach. To be specific, consider the NN S-wave scattering amplitude corresponding to the contact interaction potential of Eq. (2.10), 2 C2m2[I(q)q4−2q2I (q)+I (q)]+4C q2 +2C 2 2 4 2 T = − . (4.1) cont C2m(cid:8)4[I(q)I (q)−I (q)2]+4C m2I (q)+2CI(q)m2(cid:9)−4 2 4 2 2 2 To make the following discussion more transparent, we restrict ourselves to the leading nonrelativistic approximation so that the above expression takes the form C2m[q4J(q)−2q2J (q)+J (q)]+2C q2 +C T = − 2 2 4 2 , (4.2) cont mJ(q)[C2mJ (q)+C]−[C mJ (q)−1]2 2 4 2 2 where 2 d3k {1,~k2,(~k2)2} {J(q),J (q),J (q)} = . (4.3) 2 4 m Z (2π)3 q2 −k2 +iǫ We have verified via explicit calculations that the omitted 1/m-corrections are heavily sup- pressed for the problem at hand and of no relevance for the forthcoming discussion. Using dimensional regularization we express J (q2) and J (q2) in terms of J(q2) as 2 4 J (q2) = q2J(q2), J (q2) = q4J(q2) (4.4) 2 4 and subtract J(q2) at q2 = −µ2 obtaining iq +µ JR(q2) = − . (4.5) 2πm 3 It is not possible to disentangle the D term from the fitted value of C˜R. 9 80 80 60 60 Dg Dg de 40 de 40 @ @ ∆ ∆ 20 20 0 0 0 5 10 15 20 25 30 35 0 5 10 15 20 25 30 35 Elab@MeVD Elab@MeVD FIG. 2: Renormalization scale dependence of the phase shift in 1S partial wave NN scattering 0 emerging by non-perturbativeinclusion of theNLO contact interaction in pionless EFT. Leftpanel corresponds to a natural scattering length while right panel shows the case of an unnaturally large scattering length. Circles (color online: red)on bothpanels referto thesynthetic dataas described inthetextwhilethedashedcurveswithincreasingdashlengthcorrespondtoµ = 1,50,100,200,400 MeV. Subtracted integrals JR(q2) and JR(q2) are obtained by replacing J(q2) in Eq. (4.4) by 2 4 JR(q2) specified in Eq. (4.5). The renormalized amplitude is then given by Eq. (4.2) with the divergent integrals being replaced by their subtracted values and the bare LECs C and C being replaced by the renormalized ones CR(µ) and CR(µ) [42]. Using the scattering 2 2 length a and effective range r to determine these two LECs, we obtain the renormalized expression for the effective range function qcotδ in terms of observable quantities −a2rµq2 +2aµ−2 1 rq2 ar2q4 a2r3q6 qcotδ = = − + + + +.... (4.6) a(arq2 −2aµ+2) a 2 4(aµ−1) 8(aµ−1)2 Notice that the resulting expression is explicitly µ-dependent. This is because the UV divergencies emerging from iterations of the LS equation require counter terms beyond the truncated potential unless the C and higher-order interactions are treated in perturbation 2 theory. For the natural case describing a perturbative scenario corresponding to a ∼ Λ−1, r ∼ Λ−1, ..., with Λ being the hard scale of the order of Λ ∼ M , it is appropriate to π choose the subtraction scale µ of the order of the soft scale in the problem, i.e. of the order of external momenta of the nucleons µ ∼ q ≪ Λ. This ensures that the values of the shape parameters v in Eq. (4.6) scale with the corresponding powers of Λ, so that the residual i µ-dependence in the amplitude is beyond the accuracy of the NLO approximation. This is visualized in the left panel of Fig. 2, where the NLO pionless EFT predictions for the phase shift for the case of a = −M−1 fm and r = M−1 fm are shown as a function of laboratory π π energy E for different choices of the subtraction scale µ. The ”data” in Fig. 2 correspond lab to the effective range approximation with all shape coefficients set to zero. Notice that choosing µ of the order of the hard scale also results in a valid low-energy expansion of the effective-range function as visualized in the figure. On the other hand, for the unnatural case describing the non-perturbative situation of a system being close to the unitary limit and corresponding to very large values of the 10