arXiv:2610.23096
Nonequilibrium Dynamical Quantum Phase Transitions in the Extended Agassi Model
Ritesh Kumar Palai, Rajiv Kumar, Venktesh Singh, Sunil K. Mishra
Abstract
We investigate dynamical quantum phase transitions (DQPTs) in the extended Agassi model following sudden quenches of same level pairing ($Σ$), monopole-monopole ($χ$) and cross level pairing ($Λ$) strengths. Using exact diagonalization, we characterize the non-equilibrium dynamics through the Loschmidt echo and intensive return-rate function. Quenches across equilibrium phase boundaries produce cusp-like structures in the return rate that shows non-analyticities with increasing system size. Equilibrium finite size to large scaling identifies the $χ$-driven transition as a second order phase transition similar to mean-field, Dicke, and Lipkin-Meshkov-Glick (LMG) models critical behaviour with the exponent of the order parameter $β\simeq 1/2$, while the $Σ$- and $Λ$-driven transitions are first order. We further characterize the associated dynamical singularities through a local scaling exponent $α$, with $α_χ\simeq 2/3$, $α_Λ\simeq 1/3$, and $α_Σ\simeq 3/2$ of all three parameters. The distinct scaling behaviours across the parameter channels show that dynamical criticality is not determined solely by the equilibrium order of the transition, revealing multiple dynamical scaling regimes. These non-analyticities are associated with the Fisher zeros of the Loschmidt amplitude, which cross the real time axis at critical times and the time derivatives of the return-rate function to characterize the order of DQPTs within the extended Agassi model.
AI Summary
The paper investigates whether dynamical quantum phase transitions (DQPTs) in the extended Agassi model depend on the interaction channel and the order of the equilibrium transition. This collective two-level fermionic model combines same-level pairing $\Sigma$, monopole–monopole coupling $\chi$, and cross-level pairing $\Lambda$. Exact diagonalization in the seniority-zero collective-pair sector is used at half filling for $j=8,100,150,300$, with $N=2j$. Ground-state quenches from $(\Sigma_i,\chi_i,\Lambda_i)=(0.5,0.5,0)$ are analyzed through Loschmidt echoes, return-rate functions, complex-time Fisher zeros, and time derivatives. Equilibrium order-parameter fits give $\beta_\chi=0.5267\pm0.0035$ at $j=300$, near the mean-field value $1/2$; the selected $\Sigma$- and $\Lambda$-driven transitions are identified as first order. The return rates develop cusp-like finite-size structures. Local fits to $|\lambda(t)-\lambda(t_c)|\propto|t-t_c|^\alpha$ at $j=300$ yield $\alpha_\Sigma=1.5373\pm0.1243$, $\alpha_\chi=0.6225\pm0.0285$, and $\alpha_\Lambda=0.3191\pm0.0124$, at critical times approximately $12.52$, $16.06$, and $13.26$. These are interpreted as channel-dependent scaling estimates near $3/2$, $2/3$, and $1/3$. Fisher-zero proximity and derivative profiles provide complementary diagnostics, with first-order-like DQPTs reported for $\Sigma,\Lambda$ and second-order-like behavior for $\chi$. The paper explicitly notes that quench parameters change across system sizes, preventing controlled finite-size extrapolation. Fisher-zero pinching, transition orders, and asymptotic fractional exponents therefore remain unestablished; fixed-protocol scaling of zero trajectories and derivative peaks is needed to determine thermodynamic-limit behavior and universality.
Central claim may not hold: The advertised thermodynamic fractional exponents and DQPT singularities are not established by local finite-size fits with size-dependent quench parameters or by Fisher-zero plots lacking controlled real-axis-pinch scaling. Interpreted as leading asymptotic powers, αχ≈0.622 implies a divergent first derivative and αΣ≈1.537 implies a continuous first derivative, contradicting the paper’s respective second-order and first-order derivative classifications. The protocols called Σ- and Λ-driven change multiple interactions between initial and final Hamiltonians, so their comparison does not isolate the named interaction as the cause of the different fitted exponents. Figure R4 and Table tab:dqpt_exponents specify different final Hamiltonians for the quoted Λ-exponent calculation, leaving its physical protocol ambiguous. The appendix’s first terms in the J+ and J− actions change the stated J0 eigenvalue in the wrong direction and therefore violate the stated commutation relations, although an implementation error is not established.
Written by gpt-6.1-sol from the paper's full text. Automated: it can be wrong.
- Comments
- 21 pages , 26 figures, to be submitted to journal soon. Comments will be appreciated