oxygen evolving complex (photosynthesis)
Cluster models of metalloenzymes
$$[\text{Mn}_4\text{CaO}_5]$$
$$[\text{Cu}_2\text{O}_2]$$
tyrosinase
(melanin production)
Example: \( [\text{Cu}_2\text{O}_2]^{2+} \) isomerization
DFT: large differences in relative energies depending on the functional
Active space wave functions: (28e, 32o) includes Cu 3d, 4d, and O 2p, 3p orbitals, Hilbert space dimension: \(10^{17}\)!
Cramer et al. J. Phys. Chem. A 2006, 110, 1991-2004
arXiv:2008.00220
Nitrogenase FeMo cofactor
7 iron atoms, 1 molybdenum atom
Hubbard:
hopping (kinetic energy)
onsite e-e repulsion
Ground state of half-filled 1-D 6 site model: U = 1
\( x\)
\(E\)
Efficient description using delocalized canonical orbital configurations:
Hartree-Fock reference
particle hole excitations
Ground state of half-filled 1-D 6 site model: U = 10
Efficient description with a Jastrow factor using local orbitals:
Jastrow
HF ref.
Correlation by penalizing local charge fluctuations
Combining Jastrow with multi-Slater: Jastrow encodes strong interactions, while multi-Slater captures weak and more general interactions, like exchange
Energy of any "valid" wave function is an upper bound for the ground state energy:
To find an approximation to the ground state wave function:
local energy of walker \(|n\rangle\)
The local energy is given by
Sum of \(N_c\) such terms for each \(|m\rangle\). Since there are \(O(N^4)\) \(|m\rangle\), this is expensive.
What is the overlap between a local configuration and a canonical one?
\(\hat{L}\): local, \(\hat{C}\): canonical
\(\langle L_1L_2|C_1C_2\rangle =\) Amplitude of \(|L_1L_2\rangle\rightarrow|C_1C_1\rangle\)
Two ways to accomplish this:
or
\(L_1\rightarrow C_1\) and \(L_2\rightarrow C_2\): \( \langle L_1|C_1\rangle \times \langle L_2|C_2\rangle\)
\(L_1\rightarrow C_2\) and \(L_2\rightarrow C_1\): \( \langle L_1|C_2\rangle \times \langle L_2|C_1\rangle\)
fermion antisymmetry
Cost of calculating \(N\times N\) determinant: \(O(N^3)\)
Green's function,
involves calculating inverse (\(O(N^3)\))
\(2\times2\) determinant
ratio of \(N\times N\) determinants
By storing the Green's functions, calculation of large determinants can be avoided.
general formula
With these Green's functions, local energy can be calculated at cost \(O(N^4N_c)\), which is still very expensive when \(N_c > 10^3\).
There's a pattern in the contribution of a configuration to the local energy:
the bottom right block of the determinants is the same
Using this observation, local energy calculation can be performed at cost \(O(N^5+N_c)\), by storing some intermediates.
(This concept is analogous to when performing the product of three matrices \(A\times B\times C\), one stores \(A\times B\) or \(B\times C\) as an intermediate)
\(n = 28\): active space of \(\pi\) orbitals (28e, 28o)
arXiv:2008.06477
[\(\text{Cu}_2\text{O}_2\)]
arXiv:2008.00220
Both of these could be potentially tackled by using projection Monte Carlo
Imaginary time evolution (projection)
Exponential of one body part can be calculated in \(O(N^3)\) time, but that of the two-body part very expensive to do exactly.
Approximations:
Sign problem \(\rightarrow\) trial function \(\rightarrow\) multi-Slater
\(\exp(-\tau\hat{H})|\phi\rangle \rightarrow\) ground state, for \(\tau\rightarrow\infty\)