195
Part Five · Convergence

Simulating photosynthetic energy transfer as an open quantum system

Quantum hardware has already been built to imitate a leaf. Between 2018 and 2019, groups at ETH Zurich, Berkeley and Innsbruck used superconducting circuits and trapped ions as analog simulators of excitation transfer through a light-harvesting network, with the environment’s noise injected deliberately. They watched noise help transport, as environment-assisted quantum transport predicts (Chapter 161). This chapter rebuilds the model and asks what such simulators can tell biology that a laptop cannot.

On the plate: a three-site exciton network with tunable noise.

The model

Represent each pigment by a site that can hold one excitation, with energy εj\varepsilon_j and couplings JjkJ_{jk} between sites: a tight-binding Hamiltonian in the single-excitation space. The protein environment enters as dephasing at rate γ\gamma on each site (Lindblad form), with slow loss everywhere and a “sink” from the last site, standing for the reaction centre. Transfer efficiency is the fraction of excitations that reach the sink. Every ingredient maps onto qubits or oscillators: sites to qubits, couplings to tunable interactions, dephasing to injected noise. Established

Tune the noise

Three sites in a chain, couplings 1, site energies in a staircase of step 3.0 (site 1 highest, site 3 lowest, a mismatched energy landscape), sink rate 1 from site 3, loss 0.05 everywhere; dephasing γ=10x\gamma = 10^x, x=x = 0.5. With disorder, efficiency peaks at intermediate noise, as in Chapter 161. In 2018 Anton Potočnik, Andreas Wallraff and colleagues built a three-qubit superconducting version and observed noise-assisted transport; Christine Maier, Christian Roos and colleagues used a chain of ten trapped ions in 2019, with transport that was largely diffusive; Dylan Gorman, Hartmut Häffner and colleagues showed vibrationally assisted transfer between two ions in 2018. Established for the demonstrations.

The critique

Real light-harvesting complexes such as FMO have seven or eight sites and structured, non-Markovian environments. Classical HEOM and tensor-network methods simulate them accurately (Chapter 161), so the analog experiments are demonstrations of control and of physics, not of computational advantage. Where analog simulators might eventually help is in much larger networks with strong, structured system–bath coupling, beyond exact classical treatment, or as testbeds for theories of how noise and structure shape efficiency. The biological lesson they illustrate, that a noisy environment can assist transport, was already established theoretically. Speculative for advantage.

What you now have

A tight-binding exciton network with dephasing, loss and a sink, a three-site simulation showing noise-assisted transport, the 2018–19 superconducting and trapped-ion analog experiments, and why classical methods still suffice for natural complexes.

· · ·

Three questions to carry with you.

What does the sink represent?tap
The reaction centre that captures the excitation.
Why does noise help a disordered chain?tap
Dephasing broadens mismatched site energies so excitations can hop.
Why are these experiments not advantages?tap
Classical HEOM and tensor networks simulate systems of this size exactly.

And one to work out. A coherence that decays in 50 fs corresponds to a dephasing rate γ=1/τ\gamma = 1/\tau of how many inverse picoseconds?

γ =

Sources: A. Potočnik et al., Nat. Commun. 9, 904 (2018). C. Maier et al., Phys. Rev. Lett. 122, 050501 (2019). D. J. Gorman et al., Phys. Rev. X 8, 011038 (2018). M. B. Plenio and S. F. Huelga, New J. Phys. 10, 113019 (2008). A. Ishizaki and G. R. Fleming, Proc. Natl Acad. Sci. USA 106, 17255 (2009). Checked by the plate: RK4 integration of the three-site Lindblad equation.

Next · Chapter 196
Simulating radical-pair dynamics for magnetoreception

A robin's compass may involve about 20 nuclear spins. That alone is 10⁶ amplitudes per electron-spin state.

Turn the page