Physical realization of the extremal 2D S-matrix $S=-S_{\mathrm{SG}}$ for a bound state with $m_1<\sqrt{2}\,m$, and optimal coupling bounds away from integrable points

The two-dimensional S-matrix bootstrap bounds the cubic coupling $g_1$ of a bound state of the lightest particle using only unitarity, real analyticity, and crossing. For a bound-state mass $m_1>\sqrt{2}\,m$ the bound is optimal, being saturated by the first sine-Gordon breather; for $m_1<\sqrt{2}\,m$ the extremal amplitude is $-S_{\mathrm{SG}}$, the sine-Gordon amplitude with an overall minus sign, and the source paper states that no theory with this S-matrix is known. Two of the open directions named in the source form the subject of this task: (i) the physical status of $-S_{\mathrm{SG}}$ (existence of a consistent QFT versus a strictly stronger bound once amplitudes involving the bound-state particle and inelastic channels are included), and (ii) the improvement of coupling bounds away from integrable points by inelastic channels, including the conjectured ridge that would track the non-integrable scaling Ising model through its $E_8$ magnetic point.

Background

In a gapped, unitary, Lorentz-invariant quantum field theory, the 2-to-2 scattering amplitude of the lightest particle is strongly constrained by three structural axioms: unitarity ($|S(s+i0)|=1$ above the two-particle cut), real analyticity (the amplitude is analytic in the physical strip $0<s<4m^2$ with $S(s^*)=S(s)^*$), and crossing symmetry (in $1+1$ dimensions, where the Mandelstam variable $s$, the squared center-of-mass energy, is the only independent Lorentz invariant, this reads $S(s)=S(4m^2-s)$, exchanging $s$- and $t$-channels). Equivalently one uses the rapidity $\theta$ with $s=4m^2\cosh^2(\theta/2)$, in which crossing is $S(\theta)=S(i\pi-\theta)$. A bound state of two lightest particles of mass $m$ appears as a pole at $s=m_1^2<4m^2$ whose residue fixes the effective cubic coupling $g_1$; requiring the correct (positive) residue sign is the physical pole condition. The S-matrix bootstrap asks for the maximal coupling $g_1^{\max}(m_1/m)$ compatible with the axioms, computed numerically as a semidefinite program (a convex optimization over a discretized amplitude ansatz, paired with a dual functional that rigorously certifies the resulting bound). A 1+1-dimensional precursor is Creutz's 1972 bound on two-dimensional coupling constants; the modern program was launched by the source paper and its companions.

The source's central result: $g_1^{\max}$ is invariant under $m_1^2\leftrightarrow 4m^2-m_1^2$ (a reflection about $m_1=\sqrt{2}\,m$), because any solution maps to another upon multiplying the S-matrix by $-1$ and re-interpreting the $s$- and $t$-channel poles. For $m_1>\sqrt{2}\,m$ the bound is optimal: it is saturated by the sine-Gordon model (the integrable relativistic field theory of a periodic cosine potential, whose soliton-antisoliton attraction supports a tower of neutral bound states called breathers), identifying $m$ with the first breather and $m_1$ with the second. The first-breather amplitude is

$S_{\mathrm{SG}}(s)=\frac{\sqrt{s(4m^2-s)}+m_1\sqrt{4m^2-m_1^2}}{\sqrt{s(4m^2-s)}-m_1\sqrt{4m^2-m_1^2}},$

with poles at $s=m_1^2$ (the bound state) and at $s=4m^2-m_1^2$ (its crossing image). For $m_1<\sqrt{2}\,m$ the numerical extremal is exactly $-S_{\mathrm{SG}}(s)$: unitary, crossing-symmetric, and elastic, with the pole signs correct after the $s\leftrightarrow t$ re-interpretation; the sign flip also exchanges the high-energy limit ($S_{\mathrm{SG}}(\infty)=1$, so $-S_{\mathrm{SG}}(\infty)=-1$, the same value as in the free Majorana fermion, whose exact amplitude is $S=-1$). No QFT with this S-matrix is known. The question is sharp because saturating two-particle unitarity forces all particle production to vanish; in $1+1$ dimensions elastic, factorized scattering is the hallmark of integrability, and a consistent factorized theory must in addition satisfy the bound-state fusion bootstrap for every amplitude involving the bound-state particle as an external state. The source's own three-particle analysis shows that generic unitarity-saturating extremal amplitudes fail this fusion self-consistency, which is why 'not all extremal amplitudes correspond to physical theories'.

The second setting is the scaling limit of the Ising model: the relevant deformation of the Ising conformal field theory by a thermal (even) and a magnetic (odd) perturbation. At critical temperature with magnetic field turned on it is the integrable $E_8$ theory of Zamolodchikov, an exactly solvable scattering theory whose eight stable particles have mass ratios fixed by the $E_8$ root system ($m_2/m\simeq 1.618$, $m_3/m\simeq 1.989$); turning on temperature destroys integrability and switches on particle production. In the source's analysis of theories whose three lightest masses obey $\sqrt{2}\,m<m_2<\sqrt{3}\,m<m_3<2m$, the elastic bound is saturated only at this 'magnetic point', and near it the exact form-factor residue of the scaling Ising model approaches the bound. The stated improvement strategy is to include the inelastic channels $m+m\to m+m_2$ and $m+m\to m_2+m_2$ (which vanish at the magnetic point but not away from it) in the coupled unitarity relations, and the conjecture is that a 'ridge' of the allowed region would then pass through the magnetic point and track the non-integrable thermal deformation; multi-particle processes such as $2\to 3$ were noted as kinematically more complicated and left unexplored. A 'ridge' here denotes a curve of extremal points of the maximal coupling $g_1^{\max}(m_2/m,\,m_3/m)$ in the $(m_2/m,\,m_3/m)$ plane that passes through the magnetic point, and it 'tracks the thermal deformation' when the direction along that curve agrees with the thermal-deformation line — the trajectory of the scaled Ising model's mass ratios under a temperature shift, whose slope near the magnetic point is fixed by form-factor perturbation theory and continued farther away by the Hamiltonian-truncation and matrix-product-state data — within the accuracy of that data. The ridge question is then answered affirmatively if the inelastic allowed region develops such a curve and negatively if it does not.

Problem Statement

The common objective is to make the two-dimensional S-matrix bootstrap bounds on the lightest particle's cubic coupling optimal and physically interpretable in the regimes where they are currently neither: at bound-state masses below the crossing-symmetric value $m_1=\sqrt{2}\,m$, and away from the integrable points. Two linked deliverables, stated with the conventions of the source paper (external particles of mass $m$ are the lightest; $S(s)$ denotes their elastic $2\to2$ amplitude; the single $s$-channel bound-state pole at $s=m_1^2$ has residue defining $g_1$; the axioms are unitarity, real analyticity in the physical strip $0<s<4m^2$, crossing $S(s)=S(4m^2-s)$, and polynomial boundedness with constant $S_\infty\in[-1,1]$).

Part 1 (the minus-sign extremal amplitude). For $m_1<\sqrt{2}\,m$ the maximal coupling is attained by $S(s)=-S_{\mathrm{SG}}(s)$, where

$S_{\mathrm{SG}}(s)=\frac{\sqrt{s(4m^2-s)}+m_1\sqrt{4m^2-m_1^2}}{\sqrt{s(4m^2-s)}-m_1\sqrt{4m^2-m_1^2}},$

and no QFT with this S-matrix is known. Settle its status, in either of two admissible ways: (a) exhibit a unitary, gapped, $1+1$-dimensional QFT whose particle spectrum contains at least the lightest particle of mass $m$ and the bound state of mass $m_1<\sqrt{2}\,m$, with any further stable particles included in the amplitude set, and whose lightest-particle elastic amplitude is exactly $-S_{\mathrm{SG}}(s)$ — since this amplitude saturates two-particle unitarity, any such theory has no particle production, i.e., is a factorized-scattering (integrable-type) theory, so the construction must specify the full amplitude set ($m+m$, $m+m_1$, $m_1+m_1$) and satisfy the bound-state fusion bootstrap (the recursive closure of the spectrum under pole fusion) with no spurious poles; its consistency may be established by any rigorous means, for example the thermodynamic Bethe ansatz (the finite-volume ground-state equation that any factorized S-matrix must satisfy), the form-factor bootstrap (the recursive on-shell construction of matrix elements of local operators), Hamiltonian truncation (diagonalization on a truncated free-particle basis), matrix product states (variational one-dimensional tensor-network states), or an explicit Lagrangian construction; or (b) show that no consistent QFT realizes it for $m_1<\sqrt{2}\,m$, for example by proving that the single-amplitude bound is not optimal once amplitudes involving the bound-state particle as an external state and the inelastic channels are included (a dual functional certifying $g_1^{\max}$ strictly below the $-S_{\mathrm{SG}}$ value), or by an analytic no-go argument from the axioms. The source poses both the existence question and the possible downward movement of the bound for the entire left region $m_1<\sqrt{2}\,m$; a strictly stronger certified bound or a no-go argument at a single pinned mass ratio would be substantial partial progress but would not by itself settle the region, just as a proof that the bound remains optimal under strictly larger axiom sets would be a complementary partial result.

Part 2 (inelastic channels and the Ising ridge). In the three-lightest-particle problem (in which the exchanged bound state is the lightest particle itself, $m_1=m$) with mass window $\sqrt{2}\,m<m_2<\sqrt{3}\,m<m_3<2m$ (which contains the $E_8$ magnetic point at $m_2/m\simeq1.618$, $m_3/m\simeq1.989$), the elastic bound is known to be non-optimal away from the magnetic point. Incorporate the inelastic channels $m+m\to m+m_2$ and $m+m\to m_2+m_2$ into the bootstrap with the coupled unitarity relations, and determine quantitatively: how the maximal coupling and the allowed $(m_2/m,\,m_3/m)$ region strengthen away from the magnetic point, with saturation preserved at the magnetic point; and whether the allowed region develops a ridge passing through the magnetic point that tracks the thermal deformation of the scaling Ising model, compared against form-factor perturbation theory (the expansion of observables in matrix elements of the perturbing thermal operator) near the magnetic point and against Hamiltonian-truncation and matrix-product-state scattering data farther away.

The problem is complete when Part 1 is settled on one of its admissible branches — branch (a) by an explicit consistent theory at some bound-state mass $m_1<\sqrt{2}\,m$, branch (b) by a resolution covering the full left region $m_1<\sqrt{2}\,m$ — and the Part-2 bounds with inelastic channels are established and the ridge question answered in the affirmative or negative.

References

1. Miguel F. Paulos, Joao Penedones, Jonathan Toledo, Balt C. van Rees, Pedro Vieira, 'The S-matrix Bootstrap II: Two Dimensional Amplitudes', JHEP 11 (2017) 143. DOI 10.1007/JHEP11(2017)143, arXiv:1607.06110, https://arxiv.org/abs/1607.06110.
2. Miguel F. Paulos, Joao Penedones, Jonathan Toledo, Balt C. van Rees, Pedro Vieira, 'The S-matrix bootstrap. Part I: QFT in AdS', JHEP 11 (2017) 133. DOI 10.1007/JHEP11(2017)133, arXiv:1607.06109, https://arxiv.org/abs/1607.06109.
3. Miguel F. Paulos, Joao Penedones, Jonathan Toledo, Balt C. van Rees, Pedro Vieira, 'The S-matrix bootstrap. Part III: higher dimensional amplitudes', JHEP 12 (2019) 040. DOI 10.1007/JHEP12(2019)040, arXiv:1708.06765, https://arxiv.org/abs/1708.06765.
4. Alexandre Homrich, Joao Penedones, Jonathan Toledo, Balt C. van Rees, Pedro Vieira, 'The S-matrix Bootstrap IV: Multiple Amplitudes', JHEP 11 (2019) 076. DOI 10.1007/JHEP11(2019)076, arXiv:1905.06905, https://arxiv.org/abs/1905.06905.
5. Carlos Bercini, Matheus Fabri, Alexandre Homrich, Pedro Vieira, 'S-matrix bootstrap: Supersymmetry, $Z_2$, and $Z_4$ symmetry', Phys. Rev. D 101 (2020) 045022. DOI 10.1103/PhysRevD.101.045022, arXiv:1909.06453, https://arxiv.org/abs/1909.06453.
6. Andrea L. Guerrieri, Alexandre Homrich, Pedro Vieira, 'Dual S-matrix bootstrap. Part I. 2D theory', JHEP 11 (2020) 084. DOI 10.1007/JHEP11(2020)084, arXiv:2008.02770, https://arxiv.org/abs/2008.02770.
7. Piotr Tourkine, Alexander Zhiboedov, 'Scattering from production in 2d', JHEP 07 (2021) 228. DOI 10.1007/JHEP07(2021)228, arXiv:2101.05211, https://arxiv.org/abs/2101.05211.
8. António Antunes, Miguel S. Costa, José Pereira, 'Exploring inelasticity in the S-matrix Bootstrap', Phys. Lett. B 846 (2023) 138225. DOI 10.1016/j.physletb.2023.138225, arXiv:2301.13219, https://arxiv.org/abs/2301.13219.
9. Miguel Correia, Joao Penedones, Antoine Vuignier, 'Injecting the UV into the bootstrap: Ising Field Theory', JHEP 08 (2023) 108. DOI 10.1007/JHEP08(2023)108, arXiv:2212.03917, https://arxiv.org/abs/2212.03917.
10. Christian Copetti, Lucia Córdova, Shota Komatsu, 'S-matrix bootstrap and non-invertible symmetries', JHEP 03 (2025) 204. DOI 10.1007/JHEP03(2025)204, arXiv:2408.13132, https://arxiv.org/abs/2408.13132.
11. Martin Kruczenski, Joao Penedones, Balt C. van Rees, 'Snowmass White Paper: S-matrix Bootstrap', arXiv:2203.02421, https://arxiv.org/abs/2203.02421.
12. Barak Gabai, Xi Yin, 'On the S-matrix of Ising field theory in two dimensions', JHEP 10 (2022) 168. DOI 10.1007/JHEP10(2022)168, arXiv:1905.00710, https://arxiv.org/abs/1905.00710.
13. Hongbin Chen, A. Liam Fitzpatrick, Denis Karateev, 'Bootstrapping 2d $\phi^4$ theory with Hamiltonian truncation data', JHEP 02 (2022) 146. DOI 10.1007/JHEP02(2022)146, arXiv:2107.10286, https://arxiv.org/abs/2107.10286.
14. Raghav G. Jha, Ashley Milsted, Dominik Neuenfeld, John Preskill, Pedro Vieira, 'Real-time scattering in Ising field theory using matrix product states', Phys. Rev. Research 7 (2025) 023266. DOI 10.1103/9dxz-k5wb, arXiv:2411.13645, https://arxiv.org/abs/2411.13645.
15. Miguel Correia, Celina Pasiecznik, 'Resonances: Universality and Factorization on Higher Sheets', arXiv:2512.13775, https://arxiv.org/abs/2512.13775.
16. Michael Creutz, 'Rigorous Bounds on Coupling Constants in Two-Dimensional Field Theories', Phys. Rev. D 6 (1972) 2763. DOI 10.1103/PhysRevD.6.2763, https://doi.org/10.1103/PhysRevD.6.2763.
17. A. B. Zamolodchikov, 'Integrable field theory from conformal field theory', Adv. Stud. Pure Math. 19 (1989) 641. No DOI; INSPIRE record https://inspirehep.net/literature/279986.