The random singlet phase (RSP) is the infinite-randomness fixed point governing the ground states of strongly disordered antiferromagnetic spin-1/2 chains, where the strong-disorder renormalization group builds the state as a product of singlets joining arbitrarily distant sites. For a tripartition A1∪A2∪BA_1\cup A_2\cup B, the mixed-state entanglement between A1A_1 and A2A_2 is encoded in the eigenvalue distribution of the partially transposed density matrix ρAT2\rho_A^{T_2} (the negativity spectrum). Turkeshi, Ruggiero and Calabrese [Phys. Rev. B 101, 064207 (2020)] derived, via a renewal equation for the statistics of singlets crossing the cuts, the leading logarithmic asymptotics of the disorder-averaged negativity moments tr(ρAT2)α‾\overline{\mathrm{tr}(\rho_A^{T_2})^\alpha} — but the full disorder-averaged spectral density P(λ)‾\overline{\mathbb{P}(\lambda)}, including its λ\lambda-dependence, finite-size features, and the corrections generated by the multiple-decimation processes neglected in the derivation, has not been derived. This problem asks for that complete derivation, or an equivalent full characterization of the joint singlet-count statistics from which the spectrum follows. The extensions raised in the same source — negativity moments at other infinite-randomness fixed points and quench dynamics of the negativity spectrum — are recorded as separate follow-on open problems.

Background

For a pure state of a spin chain partitioned into three parts A1∪A2∪BA_1\cup A_2\cup B, tracing out BB leaves the blocks A1A_1 and A2A_2 in a mixed state whose quantum correlations cannot be separated from classical ones by Rényi entropies. The standard diagnostic is the partial transpose: the operator ρAT2\rho_A^{T_2} obtained from the reduced density matrix ρA=trB∣Ψ⟩⟨Ψ∣\rho_A=\mathrm{tr}_B|\Psi\rangle\langle\Psi| (A=A1∪A2A=A_1\cup A_2) by transposing only the indices belonging to A2A_2. In one dimension, ρAT2\rho_A^{T_2} has a negative eigenvalue if and only if A1A_1 and A2A_2 are entangled. The entanglement negativity N=(∥ρAT2∥1−1)/2\mathcal{N}=(\|\rho_A^{T_2}\|_1-1)/2 and the logarithmic negativity E=log⁡∥ρAT2∥1\mathcal{E}=\log\|\rho_A^{T_2}\|_1, with ∥⋅∥1\|\cdot\|_1 the trace norm, quantify this entanglement. The full spectral density P(λ)=∑iδ(λ−λi)\mathbb{P}(\lambda)=\sum_i\delta(\lambda-\lambda_i) over the eigenvalues λi\lambda_i of ρAT2\rho_A^{T_2} — the negativity spectrum — contains strictly more information than either scalar; it is fully characterized by the negativity moments MαT2=tr(ρAT2)αM^{T_2}_\alpha=\mathrm{tr}(\rho_A^{T_2})^\alpha, since all moments determine λP(λ)\lambda\mathbb{P}(\lambda) through an inverse Stieltjes transform.

Strong quenched disorder in antiferromagnetic spin-1/2 chains (random-bond Heisenberg and XX models) drives the ground state into the random singlet phase (RSP): the strong-disorder renormalization group (SDRG) introduced by Ma, Dasgupta and Hu and developed by Fisher iteratively decimates the strongest bond into a singlet and generates weaker effective couplings, producing singlets between arbitrarily distant sites and flowing to an infinite-randomness fixed point (IRFP) — a critical point at which the ratio of disorder to energy scale diverges. Refael and Moore showed that the disorder-averaged entanglement entropy of a block grows logarithmically, S‾=(ln⁡2/3)ln⁡ℓ+const\overline{S}=(\ln 2/3)\ln\ell+\mathrm{const}. Ruggiero, Alba and Calabrese subsequently proved, within the SDRG, that the disorder-averaged logarithmic negativity between adjacent intervals A1,A2A_1,A_2 also grows logarithmically, E‾=n‾A1:A2ln⁡2\overline{\mathcal{E}}=\overline{n}_{A_1:A_2}\ln 2 with n‾A1:A2=(1/6)ln⁡(ℓ1ℓ2/(ℓ1+ℓ2))\overline{n}_{A_1:A_2}=(1/6)\ln(\ell_1\ell_2/(\ell_1+\ell_2)): the negativity is proportional to the mean number of singlets shared by the two blocks. Before 2020 the negativity spectrum itself was known only for clean systems — one-dimensional conformal field theories, gapped phases, and free fermions (via a partial time-reversal operation instead of partial transpose).

Turkeshi, Ruggiero and Calabrese gave the first treatment for a disordered critical system. At the SDRG fixed point, a fixed disorder realization gives a factorized ρAT2\rho_A^{T_2} whose spectrum is exactly discrete: λ±=±2−(nA:B+nA1:A2)\lambda_\pm=\pm 2^{-(n_{A:B}+n_{A_1:A_2})} with degeneracies d−=2nA:B+nA1:A2−1(2nA1:A2−1)d_-=2^{n_{A:B}+n_{A_1:A_2}-1}(2^{n_{A_1:A_2}}-1) and d+=2nA:B+2nA1:A2−d−d_+=2^{n_{A:B}+2n_{A_1:A_2}}-d_-, where nA:Bn_{A:B} and nA1:A2n_{A_1:A_2} count singlets crossing the corresponding cuts; the spectrum depends on the realization only through these two integers. The disorder-averaged spectrum is therefore a functional of the joint distribution P(nA:B,nA1:A2)P(n_{A:B},n_{A_1:A_2}), equivalently of its generating function g(t,s)=log⁡⟨etnA:B+snA1:A2⟩g(t,s)=\log\langle e^{t n_{A:B}+s n_{A_1:A_2}}\rangle. The authors write a renewal equation for g(t,s)g(t,s), keeping only the dominant process of isolated single decimations (formation of an in–out singlet at one bond at a time, with waiting-time distribution inherited from the SDRG flow) and treating the singlet-emission probabilities p,qp,q (p+q=1p+q=1) as constants equal to their asymptotic values — p=2/3p=2/3, q=1/3q=1/3 for adjacent intervals with ℓ1∝ℓ2≫1\ell_1\propto\ell_2\gg 1. The equation is explicitly an "educated conjecture" whose assumptions, in the authors' words, "can be tested only a posteriori". Solving it at leading order in the RG time μ∝ln⁡ℓ\mu\propto\ln\ell yields the universal leading scaling of the disorder-averaged moments, log⁡MαT2‾=Eα\log\overline{M^{T_2}_\alpha}=\mathcal{E}_\alpha with

Eαe=5+(1+q) 23−α−32 μ+…(α even),Eαo=5+23−α−32 μ+…(α odd),\mathcal{E}^{e}_\alpha=\frac{\sqrt{5+(1+q)\,2^{3-\alpha}}-3}{2}\,\mu+\dots \quad(\alpha\ \mathrm{even}), \qquad \mathcal{E}^{o}_\alpha=\frac{\sqrt{5+2^{3-\alpha}}-3}{2}\,\mu+\dots \quad(\alpha\ \mathrm{odd}),

and exposes a structural feature: the disorder average does not commute with the logarithm, so the averaged negativity and averaged logarithmic negativity are not trivially related. The parallel program for the simpler entanglement spectrum (moments of ρA\rho_A rather than ρAT2\rho_A^{T_2}), carried out earlier by Fagotti, Calabrese and Moore, stopped at the same stage — leading moment asymptotics — and its full spectrum is likewise still missing. What remains unknown is everything beyond this leading asymptotics: the complete λ\lambda-dependent distribution P(λ)‾\overline{\mathbb{P}(\lambda)}, its nonleading finite-size features, and the corrections induced by the processes (multiple simultaneous decimations, RG-time dependence of p,qp,q, memory effects) discarded by the renewal equation.

Problem Statement

Derive analytically the full disorder-averaged negativity spectrum P(λ)‾\overline{\mathbb{P}(\lambda)} of the partial transpose ρAT2\rho_A^{T_2} for a tripartition of an infinite random spin-1/2 chain in the random singlet phase — canonically, two adjacent intervals A1,A2A_1,A_2 of lengths ℓ1,ℓ2\ell_1,\ell_2 embedded in an infinite chain. Equivalently, determine the complete generating function g(t,s)g(t,s) of the joint distribution P(nA:B,nA1:A2)P(n_{A:B},n_{A_1:A_2}) of singlets crossing the cuts at all RG scales, from which P(λ)‾\overline{\mathbb{P}(\lambda)} follows through the fixed-realization discrete spectrum λ±=±2−(nA:B+nA1:A2)\lambda_\pm=\pm 2^{-(n_{A:B}+n_{A_1:A_2})} with degeneracies d±d_\pm. A complete answer must: (i) go beyond the leading logarithmic asymptotics of the disorder-averaged moments and reconstruct the entire λ\lambda-dependent distribution, including nonleading finite-size features and subleading terms in the RG time μ∝ln⁡ℓ\mu\propto\ln\ell; and (ii) either quantify the corrections generated by the processes neglected in the single-decimation renewal equation — multiple simultaneous decimations, the RG-time dependence and activation times of the emission probabilities pp and qq, and memory effects — or replace that conjectural equation by a controlled derivation. The result must reproduce the known leading even/odd moment coefficients (with q=1/3q=1/3 for adjacent intervals), the normalization tr ρAT2=1\mathrm{tr}\,\rho_A^{T_2}=1, and be consistent with numerical strong-disorder renormalization group and exact free-fermion (random XX chain) computations of the partial transpose. The further directions raised in the same source — negativity moments at other infinite-randomness fixed points, and the quench dynamics of the negativity spectrum — are distinct follow-on problems, not part of this one.

References

  1. X. Turkeshi, P. Ruggiero, P. Calabrese, "Negativity spectrum in the random singlet phase", Phys. Rev. B 101, 064207 (2020). DOI: 10.1103/PhysRevB.101.064207; arXiv:1910.09571, https://arxiv.org/abs/1910.09571
  2. P. Ruggiero, V. Alba, P. Calabrese, "Entanglement negativity in random spin chains", Phys. Rev. B 94, 035152 (2016). DOI: 10.1103/PhysRevB.94.035152; arXiv:1605.00674, https://arxiv.org/abs/1605.00674
  3. M. Fagotti, P. Calabrese, J. E. Moore, "Entanglement spectrum of random-singlet quantum critical points", Phys. Rev. B 83, 045110 (2011). DOI: 10.1103/PhysRevB.83.045110; arXiv:1009.1614, https://arxiv.org/abs/1009.1614
  4. G. Refael, J. E. Moore, "Entanglement Entropy of Random Quantum Critical Points in One Dimension", Phys. Rev. Lett. 93, 260602 (2004). DOI: 10.1103/PhysRevLett.93.260602, https://doi.org/10.1103/PhysRevLett.93.260602
  5. D. S. Fisher, "Critical behavior of random transverse-field Ising spin chains", Phys. Rev. B 51, 6411 (1995). DOI: 10.1103/PhysRevB.51.6411, https://doi.org/10.1103/PhysRevB.51.6411
  6. S.-k. Ma, C. Dasgupta, C.-k. Hu, "Random Antiferromagnetic Chain", Phys. Rev. Lett. 43, 1434 (1979). DOI: 10.1103/PhysRevLett.43.1434, https://doi.org/10.1103/PhysRevLett.43.1434
  7. P. Calabrese, J. Cardy, E. Tonni, "Entanglement Negativity in Quantum Field Theory", Phys. Rev. Lett. 109, 130502 (2012). DOI: 10.1103/PhysRevLett.109.130502, https://doi.org/10.1103/PhysRevLett.109.130502
  8. P. Ruggiero, V. Alba, P. Calabrese, "Negativity spectrum of one-dimensional conformal field theories", Phys. Rev. B 94, 195121 (2016). DOI: 10.1103/PhysRevB.94.195121; arXiv:1607.02992, https://arxiv.org/abs/1607.02992
  9. G. B. Mbeng, V. Alba, P. Calabrese, "Negativity spectrum in 1D gapped phases of matter", J. Phys. A: Math. Theor. 50, 194001 (2017). DOI: 10.1088/1751-8121/aa6734; arXiv:1612.05172, https://arxiv.org/abs/1612.05172
  10. H. Shapourian, P. Ruggiero, S. Ryu, P. Calabrese, "Twisted and untwisted negativity spectrum of free fermions", SciPost Phys. 7, 037 (2019). DOI: 10.21468/SciPostPhys.7.3.037; arXiv:1906.04211, https://arxiv.org/abs/1906.04211
  11. J. S. Zou, H. S. Ansell, I. A. Kovács, "Multipartite entanglement in the random Ising chain", Phys. Rev. B 106, 054201 (2022). DOI: 10.1103/PhysRevB.106.054201; arXiv:2205.14306, https://arxiv.org/abs/2205.14306
  12. P. Ruggiero, X. Turkeshi, "Quantum information spreading in random spin chains", Phys. Rev. B 106, 134205 (2022). DOI: 10.1103/PhysRevB.106.134205; arXiv:2206.02934, https://arxiv.org/abs/2206.02934