Figure 1

(a) Fluctuation of the transfer efficiency $p_{\mathrm{out}}^{(\mathcal{T})}$ from input to output, for $500$ different random conformations of $N=7$ sites (•). Horizontal lines indicate the transfer efficiency of the experimentally inferred [13] FMO Hamiltonian, as well as that of the optimal configuration FMO* compatible with the experimental error margin. (b) Probability densities $P(p_{\mathrm{out}}^{(\mathcal{T})})$ of the transfer efficiency $p_{\mathrm{out}}^{(\mathcal{T})}$ for $2.5\times {10}^8$ different conformations. For fully coherent dynamics (black line) the mean value of $p_{\mathrm{out}}^{(\mathcal{T})}$ amounts to $4.9\%$, and only $4.5$ out of ${10}^6$ configurations provide efficiencies larger than $90\%$. Under local dephasing (gray line), the mean efficiency drops to $3.9\%$. (c) Gains (•) and losses (○) of the transfer efficiency with dephasing.Reuse & Permissions

Figure 2

(a) Optimal spatial configuration of $N=7$ sites offering fast, robust, and complete transport from input to output. (b) Time evolution of the on-site probabilities $|\langle i|\mathrm{<ruby><rb>\psi </rb><rt>ぷさい</rt></ruby>}(t)\rangle |{}^2$ generated by the Hamiltonian defined by (a). $i$ is either the input site (green solid line), the output node (red dashed), or an intermediate site (black thin). At time ${\mathcal{T}}^{\prime }$, only the output site is populated. At intermediate times $t<{\mathcal{T}}^{\prime }$, the excitation is spread over several sites, leading to high values of the bipartite (dash-dotted) and quadripartite (dash-dash-dotted) entanglement (see text).Reuse & Permissions

Figure 3

(a) Contour plot of the conditional probability density $P(p_{\mathrm{out}}^{(\mathcal{T})}|c_2^{({\mathcal{T}}^{\prime })})$, i.e., the probability distribution of the transport efficiency $p_{\mathrm{out}}^{(\mathcal{T})}$ across seven sites, given the generation of a certain maximal level $c_2^{({\mathcal{T}}^{\prime })}$ of at least bipartite entanglement, during the transfer time ${\mathcal{T}}^{\prime }$. (b) Same as (a), but for quadripartite entanglement $c_4^{({\mathcal{T}}^{\prime })}$. (c) Same conditional probability distribution as in (a), when all sites are locally coupled to a dephasing environment, with dephasing rate $\mathrm{<ruby><rb>\gamma </rb><rt>がんま</rt></ruby>}=2/\mathcal{T}$. In all three cases, large transport efficiencies require a minimum amount of entanglement.Reuse & Permissions