+ All Categories
Home > Documents > Perturbative gadgets with constant-bounded interactions

Perturbative gadgets with constant-bounded interactions

Date post: 31-Dec-2015
Category:
Upload: ashely-mullins
View: 31 times
Download: 1 times
Share this document with a friend
Description:
Perturbative gadgets with constant-bounded interactions. Sergey Bravyi 1 David DiVincenzo 1 Daniel Loss 2 Barbara Terhal 1. 1 IBM Watson Research Center 2 University of Basel. Classical and Quantum Information Theory Santa Fe, March 27, 2008. arXiv:0803.2686. Outline:. Physics. - PowerPoint PPT Presentation
Popular Tags:
30
Perturbative gadgets with constant-bounded interactions Sergey Bravyi 1 David DiVincenzo 1 Daniel Loss 2 Barbara Terhal 1 Classical and Quantum Information Theory Santa Fe, March 27, 2008 1 IBM Watson Research Center 2 University of Basel arXiv:0803.2686
Transcript

Perturbative gadgets with constant-bounded interactions

Sergey Bravyi1

David DiVincenzo1

Daniel Loss2 Barbara Terhal1

Classical and Quantum Information TheorySanta Fe, March 27, 2008

1 IBM Watson Research Center2 University of Basel

arXiv:0803.2686

High-energy fundamentaltheory, full Hamiltonian H (simple)

Effective low-energy Hamiltonian Heff

(complex)

High-energy simulator Hamiltonian H (simple)

Low-energy target Hamiltonian Htarget

(complex)

Outline:Physics Perturbative gadgets

Goal: develop a rigorous formalism for constructinga simulator Hamiltonian within a physical range of parameters

Rigorous, but unphysical scaling of interactions

Rigorous, but unphysical scaling of interactions

Non-rigorous territory

Non-rigorous territory

Motivation:

1. Htarget is chosen for some interesting ground-state properties

Toric code model

Quantum loop models

Briegel-Raussendorf cluster state

2. Htarget is chosen for some computational hardness properties

Quantum NP-hard Hamiltonians

Adiabatic quantum computation

What is realistic simulator Hamiltonian ?

1. Only two-qubit interactions

2. Norm of the interactions is bounded by a constant (independent of the system size)

3. Each qubit can interact with a constant number of otherqubits (bounded degree)

4. Nearest-neighbor interactions on a regular lattice(desirable but not necessary)

Main question: given a target Hamiltonian, what groundstate properties can be reproduced by a realistic simulator Hamiltonian?

Main question: given a target Hamiltonian, what groundstate properties can be reproduced by a realistic simulator Hamiltonian?

Wish list:

1. Ground-state energy; small ``extensive’’ error

2. Expectation values of extensive observables (e.g. average magnetization) small extensive error

3. Expectation values of local observables

4. Spectral gap

5. Topological Quantum Order

Gap

ped

Ham

ilton

ians

?

Today’s talk

Some Terminology:

Example: 2D Heisenberg model

2-local

Pauli degree =12+1=13

Interaction strength

Main result

*Can be improved to Pauli degree = 3

[Kempe, Kitaev, Regev 05] 3-local to 2-local

[Oliveira, Terhal 05] k-local to 2-local on 2D lattice

[Bravyi, DiVincenzo, Oliveira, Terhal 06] k-local to 2-local for stoquastic Hamiltonians

[Biamonte, Love 07] simulator with XZ,X,Z only

[Jordan, Farhi 08] k-th order perturbative gadgets

[Schuch, Verstraete 07] simulator with Heisenberg interactions

Idea of perturbation gadget

Main improvement: interaction strength of the simulator is reduced from J poly(n) to O(J)

Shortcoming: can not go beyond a small extensive error

If the only purpose of the simulator H is to reproduce the ground state energy, why don’t we “simulate” Htarget simply by computing its groundstate energy with a small extensive error ?

1. We hope that H reproduces more than just the ground state energy (for example, expectation values of extensive observables)

2. Computing the ground state energy of Htarget with a small extensive error is NP-hard problem even for classical Hamiltonians

[see Vijay Vazirani, “Approximation Algorithms”, Chapter 29]

Hardness of Approximation = PCP theorem

Our result:

Hamiltonians on a regular lattice

The Simulation Theorem: plan of the proof

1. Add ancillary high-energy “mediator” qubits to the“logical” qubits acted on by Htarget

2. Choose appropriate couplings between the mediatorand the logical qubits

3. Construct a unitary operator generating an effectivelow-energy Hamiltonian acting on the system qubits

4. Apply Lieb-Robinson type arguments to bound the error

new

Fol

low

s ol

d id

eas

unpertubed Hamiltonian perturbation

Toy model: why interesting ? Perturbation gadgets

2nd order Lemma: the lower bound

2nd order Lemma: the upper bound

Local block-diagonalization: Schrieffer-Wolff transformation

Basic properties:

Global block-diagonalization

Generalization: combining SW-formalism with the coupled cluster method

Coupled cluster method [F. Coester 1958]: heuristic simulation algorithm formany-body quantum systems. One of the most powerful techniques in themodern quantum chemistry. Main idea: use variational states

Where C is so called creation operator

It is expected that ground states of realistic Hamiltonians can be approximated by taking into account only subsets Γ of small size (C is a local operator)

Generalization: combining SW-formalism with the coupled cluster method

Generalization: combining SW-formalism with the coupled cluster method

Main question: given a target Hamiltonian, what groundstate properties can be reproduced by a realistic simulator Hamiltonian? Remains largely open…

Wish list:

1. Ground-state energy; small ``extensive’’ error

2. Expectation values of extensive observables;small extensive error

3. Expectation values of local observables

4. Spectral gap

5. Topological Quantum Order

Gap

ped

Ham

ilton

ians

?

Today’s talk


Recommended