A numerical linked-cluster algorithm was recently introduced to study quantum quenches in the thermodynamic limit starting from thermal initial states [M. Rigol, Phys. Rev. Lett. 112, 170601 (2014)]. Here, we tailor that algorithm to quenches starting from ground states. In particular, we study quenches from the ground state of the antiferromagnetic Ising model to the XXZ chain. Our results for spin correlations are shown to be in excellent agreement with recent analytical calculations based on the quench action method. We also show that they are different from the correlations in thermal equilibrium, which confirms the expectation that thermalization does not occur in general in integrable models even if they cannot be mapped to noninteracting ones. where M (c) is the multiplicity of c (number of ways per site in which c can be embedded on the lattice) and W O (c) is the weight of a given observableÔ in c.is calculated using the inclusion-exclusion principle:In Eq. (2), the sum runs over all connected sub-clusters of c andis the expectation value ofÔ calculated for the finite cluster c, with the many-body density matrixρ c . In thermal equilibrium, linked-cluster calculations are usually implemented in the grand-canonical ensemble (GE), soρ c ≡ρ.Ĥ c andN c are the Hamiltonian and the total particle number operators in cluster c, µ and T are the chemical potential and the temperature, respectively, and k B is the Boltzmann constant (k B is set to unity in what follows).Within NLCEs, O(c) in Eq. (3) is calculated using exact diagonalization [22][23][24] (for a pedagogical introduction to numerical linked-cluster expansions and their implementation, see Ref. [25]). For various lattice models of interest in thermal equilibrium, NLCEs typically converge at lower temperatures than high-temperature expansions [22][23][24]. In order to use NLCEs to make calculations in the DE after a quench starting from a thermal state [20], the system is assumed to be disconnected from the bath at the time of the quench, at which, in each cluster c,Ĥ