Quantum
Computing
for Modeling
of Molecules
and Materials
Sparse-rank factorization
methods in physics
Thomas E. Baker
Canada Research Chair in Quantum Computing for Modeling of Molecules and Materials
Department of Physics & Astronomy
Department of Chemistry
Centre for Advanced Materials and Related Technology
University of Victoria
Outline
Quantum Chemistry
● Electron interactions
● Density functionals
Quantum Computing
● Special rules
● Near-term computers
https://en.wikipedia.org/wiki/Disaccharide
Quantum Algorithms
● Sparse-rank factorization: Lanczos
● Ground states and excitations
● Density functionals
Quantum chemistry
Electron-electron interactions
● Born-Oppenheimer approximation
1 2 1 1
∑ ∑ ∑
H=− ∇i + + v(ri)
2 i 2 i≠j | ri − rj | i
● Second quantization:
∑( )
tijc†iσ
̂ cjσ
̂ + Vijkℓc†iσ
̂ c†ĵ σ̄ cℓ̂ σ̄ ckσ
̂
∑
H=
ijσ kℓσ̄
Hardness of many-body problem
QMA-hard
Density functional theory
Walter Kohn:
• Nobel Prize in Chemistry (1998)
A. Hohenberg & Kohn Theorem
(1964)
• Compact representation
• Provably equivalent
B. Kohn & Sham (1965)
• Also John Pople
Density functional theory
∬ ∫
n(r) = … | Ψ(r, r2, …, rNe |2 drdr2…drNe
Density functional theory
E = min (F[n] + V[n])
n
∫
V[n] = v(r)n(r)dr
F[n] = …?
Quantum Computing
Lanyon, et. al., Nature Chemistry 2, 106–111 (2010)
Constraints
Quantum computing
● Uncertainty
● No copying (no cloning)
● Less memory
● Simple operations are harder
○ Addition has a high gate count
● Reverse Copenhagen interpretation
Algorithms
L. K. Grover, “Quantum Mechanics Helps in Searching for a
Grover’s search algorithm Needle in a Haystack,” Phys. Rev. Lett. 79, 325 (1997)
● Classical analogy with coupled oscillators
L. K. Grover, “From Schrödinger’s equation to the quantum
search algorithm,” Pramana 56, 333 (2001)
3 G 200 12 …
● Classical: O(N); Quantum: O( N)
● A few other notable algorithms (Shor’s algorithm, Deutsch-Josza)
Grover’s search for the density functional?
What is N?
● Superposition of all states
H N
(⨂ )
⊗N 1
∑
H | ψ⟩ = H | 0⟩ = N
| σ1σ2…σN⟩
H i=1 2 σ1σ2…σN
Grover’s search for the density functional?
What is N?
● N is exponentially sized in the quantum case
○ Uninformed search
○ Not just square integrable functions
E = min (F[n] + V[n])
n
● Not feasible…
Near-term quantum algorithms
Variational Quantum Eigensolver (VQE)
● Minimize energy with classical parameters
E= min ⟨ψ(θ1, θ2, …, θN ) | H | ψ(θ1, θ2, …, θN )⟩
θ1,θ2,…,θN
● NP-hard in general
● Noisy
Real-time evolution
Trotter-Suzuki decomposition
● Decompose the time evolution operator
−iHt −iHAt −iHBt 2
e =e e + O(t ) Lanyon, et. al., Nature Chemistry 2, 106–111 (2010)
● How long to run? Months on small molecules
Quantum Lanczos Recursion
T.E. Baker, Phys. Rev. A 103, 032404 (2021)
Ground-states
● Run a Lanczos recursion
● How to actually do this?
∑
○ Linear combination of unitaries H = αiUi
○ Oblivious Amplitude Amplification
■ Guarantees operator applied is correct i
Quantum Lanczos Recursion
T.E. Baker, Phys. Rev. A 103, 032404 (2021)
Minimal wavefunction preparation
● Sampling trick
● State-preserving quantum counting (QAM-sampling/Quantum
Amplitude Estimation)
Tangent: Quantum Metropolis sampling
● Temme, et. al. Quantum Metropolis Sampling Nature 471,
87 (2011)
● Metropolis algorithm
Tangent: Quantum Metropolis sampling
| 0⟩ | 0⟩, | 1⟩
| 0⟩ | E′⟩
QPE
| Ψ⟩ Ĉ
| E⟩

Tangent: Quantum Metropolis sampling
● Quantum counting
○ “Counts” the number of transitions between a wavefunction
○ QAM-sampling
○ Marriott and Watrous, Quantum Arthur- Merlin games, Comput.
Complex 14, 122 (2005)
Ĉ | Ψ⟩ = α | Ψ⟩ + α ⊥ | Ψ⊥⟩
● Generates expectation values of any operator!
○ No wavefunction destruction!
○ Reversible
Quantum Lanczos Recursion
T.E. Baker, Phys. Rev. A 103, 032404 (2021)
Green’s functions
Quantum computing: Excited States
T.E. Baker, arxiv: 2109.14114
Excitations at polynomial cost
● Block Lanczos method
○ Block matrices replace Lanczos coefficients
■ O(d2) for d excitations
Quantum computing: Excited States
T.E. Baker, arxiv: 2109.14114
Fast convergence
Quantum computing: Excited States
T.E. Baker, arxiv: 2109.14114
Fast convergence
Quantum computing: Excited States
T.E. Baker, arxiv: 2109.14114
Fast convergence
Quantum computing: Excited States
T.E. Baker, arxiv: 2109.14114
Linear error relationship
Density Functional Theory & Quantum Computing
Density Functional Theory from the quantum computer
● Back to density functionals…
∑
n(r) = ρijϕi(r)ϕj(r)
ij
ρij = ⟨Ψ | c†î cĵ | Ψ⟩
● Coefficients of the density matrix are just expectation values!
● Energy and density is sufficient
Exact Kohn-Sham potentials
● Always v-representable
● Cost function with Quantum Gradient Algorithm
Density Functional Theory from the quantum computer
● Type of density functional
○ Many varieties
● Technically, potential functionals so far
○ There is a Potential Functional Theory that has a Hohenberg-Kohn
proof attached.
○ Proper labelling of training data
● Can get gradients from Quantum Gradient Algorithm
Density Functional Theory from the quantum computer
● Potential functional theory
Density Functional Theory from the quantum computer
● Time-dependent density functional theory
Density Functional Theory from the quantum computer
● Finite-temperature density functional theory
Density Functional Theory from the quantum computer
● More!
○ Ensemble DFT
○ Non-Born-Oppenheimer DFT
○ Relativistic DFT
○ Molecular dynamics simulations that are accurate with DFT
● Summary:
○ Two algorithms: Quantum counting and quantum gradient
algorithms
○ Any type of density functional
○ Minimal measurement
Conclusion DMRjulia
● New algorithms
○ DFT from the quantum computer
○ Lanczos-based wavefunction preparation (rapid convergence!)
● From old algorithms
○ Quantum Counting
○ Quantum Gradient Algorithm
● Continued Fraction Green’s function
● Excitations
● Needs error correction
● Might have near-term application
