Determining eigenstates and thermal states on a quantum computer using quantum imaginary time evolution
read more
Citations
Quantum computational chemistry
Quantum Algorithms for Quantum Chemistry and Quantum Materials Science.
Real- and Imaginary-Time Evolution with Compressed Quantum Circuits
Near-optimal ground state preparation
Practical quantum computation of chemical and nuclear energy levels using quantum imaginary time evolution and Lanczos algorithms
References
Simulating physics with computers
An iteration method for the solution of the eigenvalue problem of linear differential and integral operators
A variational eigenvalue solver on a photonic quantum processor
The density-matrix renormalization group in the age of matrix product states
The density-matrix renormalization group in the age of matrix product states
Related Papers (5)
Quantum supremacy using a programmable superconducting processor
Frequently Asked Questions (15)
Q2. What is the minimum error for real a[m]?
Minimizing for real a[m] corresponds to minimizing the quadratic function f(a[m])f(a[m]) = f0 + ∑IbIa[m]I + ∑IJa[m]ISIJa[m]J (8)wheref0 = 〈∆0|∆0〉 , (9) SIJ = 〈Ψ|σ†IσJ |Ψ〉 , (10) bI = i 〈Ψ|σ†I |∆0〉 − i 〈∆0|σI |Ψ〉 , (11)whose minimum obtains at the solution of the linear equation( S+ ST ) a[m] = −b (12)In general, S+ST may have a non-zero null-space.
Q3. What are the main methods of generating thermal averages?
Several procedures have been proposed for quantum thermal averaging, ranging from generating the finite-temperature state explicitly by equilibration with a bath [34], to a quantum analog of Metropolis sampling [35] that relies on phase estimation, as well as methods based on ancilla based Hamiltonian simulation with postselection [36] and approaches based on recovery maps [37].
Q4. How did the authors converge the two density matrices?
Using the 21 × 31 lattice and χ = 60, the authors were able to converge entries of single-site density matrices ρ(i) to a precision of ±10−6 (two site density matrices ρ(i, j) had higher precision).
Q5. How did the authors limit the bond dimension to D = 5?
To control the computational cost, the authors limited their bond dimension to D = 5 and used an optimized contraction scheme [50], with maximum allowed bond dimension of χ = 60 during the contraction.
Q6. How can one generate thermal averages of observables without ancillae or?
given a method for imaginary time evolution, one can generate thermal averages of observables without any ancillae or deep circuits.
Q7. What was the funding for the Rigetti computations?
The Rigetti computations were made possible by a generous grant through Rigetti Quantum Cloud services supported by the CQIA-Rigetti Partnership Program.
Q8. What is the way to measure the correlation length?
The authors chose a magnetic field value h = 3.5 which is detuned from the critical field (h ≈ 3.044) but still maintains a correlation length long enough to see interesting behaviour.
Q9. What is the cost of solving the linear system?
This is also the cost to solve classically the linear system which gives the associated Hamiltonian A[s] and of finding a circuit decomposition of Us = e iA[s]/n in terms of two qubit gates.
Q10. How many contractions did the authors need to perform for the tensor update?
Even though the simple update procedure was used for the tensor update, the authors still needed to contract the 21 × 31 PEPS at at every imaginary time step β for a range of correlation functions, amounting to a large number of contractions.
Q11. What is the simplest way to compute the METTS steps?
For instance, for the odd METTS steps, |φi〉 is collapsed onto the X-basis (assuming a Z computational basis, tensor products of |+〉 and |−〉), and for the even METTS steps, |φi〉 is collapsed onto the Z-basis (tensor products of |0〉 and |1〉).
Q12. how many vectors can be used to regularize the problem?
To regularize the problem, out of the set of time-evolved states the authors extract a better-behaved sequence as follows (i) start from |Φlast〉 = |Φ0〉 (ii) add the next |Φl〉 in the set of timeevolved states s.t. |〈Φl|Φlast〉| < s, where s is a regularization parameter 0 < s < 1 (iii) repeat, setting the |Φlast〉 = Φl (obtained from (ii)), until the desired number of vectors is reached.
Q13. What is the smallest eigenvalue of (i)?
For β = 0.001 − 0.012, the smallest eigenvalue of ρ(i) fell below this precision threshold, leading to significant noise in I(i, j).
Q14. What is the standard Metropolis sampling for thermal states?
In standard Metropolis sampling for thermal states, one starts from |φi〉 and obtains the next state |φj〉 from randomly proposing and accepting based an acceptance probability.
Q15. What is the simplest way to find the unitaries?
Us can be written as eiA[s]/n.Total Run Time: Theorem 1 gives an upper bound on the maximum support of the unitaries needed for a Trotter update, while tomography of local reduced density matrices gives a way to find the unitaries.