Fig. 5: Initialization of RVB eigenstates.

a Evolution of RVB oscillations as a function of the time to set all exchanges equal (J_ij ≃ 25 MHz) (see Fig. 3a). For t_ramp ≳ 140 ns, the ground state with s-wave symmetry is adiabatically prepared. b Evolution of the singlet-singlet probability \({P}_{{S}_{12}{S}_{34}}\) with \(\delta {V}_{x}^{{\prime} }\) after adiabatic initialization of the ground state. c Evolution of the mean singlet-singlet probability measured after adiabatic initialization of the ground state with \(\delta {V}_{x}^{{\prime} }\) for both readout directions. The experiments are compared to theoretical expectations using exchange coupling values extracted from four-spin singlet-triplet oscillations (Supplementary Notes 4 and 5). The shaded areas correspond to one standard deviation for the experimental data, and for the theoretical data they correspond to the uncertainties on the expected probabilities computed using the uncertainties on the exchange couplings values. Rescaled data (dark red and blue triangles) are obtained by multiplying each raw dataset (red and blue triangles) by a constant factor corresponding to the mean ratio of the predicted probabilities over the measured probabilities. d Experimental sequence used to investigate the formation of the d-wave state. Before the free evolution step, one exchange pulse on vB₂₃ is applied for a time t_J. e, f Evolution of singlet-singlet oscillations measured for different exchange pulse durations t_J. The vanishing of oscillations at t_J ≃ 25 ns marks the formation of a d-wave state. g, h Linecuts of (e) and (f) for t_J = 25 ns.