Constructing and programming quantum hardware for robust quantum annealing processes

Patent metadata

Field	Value
Publication number	US-10346760-B2
Application number	US-201415109614-A
Country	US
Kind code	B2
Filing date	Dec 31, 2014
Priority date	Jan 6, 2014
Publication date	Jul 9, 2019
Grant date	Jul 9, 2019

Abstract

Methods, systems and apparatus for constructing and programming quantum annealing hardware. In one aspect, a method includes deriving data characterizing an energy spectrum of a Hamiltonian Htotal that characterizes quantum states of a quantum processor, wherein the quantum processor is controllable such that the Hamiltonian Htotal evolves from an initial Hamiltonian Hi to a problem Hamiltonian Hp comprising an energy spectrum that encodes a solution to an optimization problem, the deriving being based on Hi and Hp at a time that Htotal has the energy spectrum; estimating an average phonon energy of a bath in which the quantum processor is located; and determining, based on the derived data and the calculated average phonon energy, an additional Hamiltonian HQG that when combined with Htotal limits evolution of the quantum state of the quantum processor into the ground state of Hp when Htotal evolves to Hp without diagonalizing Hp.

First claim

What is claimed is:

1. A method comprising:
deriving, by one or more processors, data characterizing an energy spectrum of a Hamiltonian H total that characterizes controllable quantum states of a quantum processor, wherein the quantum processor is controllable such that
the Hamiltonian H total evolves from an initial Hamiltonian H i to a problem Hamiltonian H p comprising an energy spectrum that encodes a solution to a machine learning optimization problem, and
a quantum state of the quantum processor evolves from a ground state of H i towards a ground state of H p as H total evolves from H i to H p ,

the deriving being based on a combination of H i and H p at a time that H total has the energy spectrum;
estimating, based on reorganization energy and frequency of a bath in which the quantum processor is located and by the one or more processors, an average phonon energy of the bath; and
determining, by the one or more processors, an additional quantum Hamiltonian H QG based on the derived data characterizing the energy spectrum of H total and the estimated average phonon energy, such that when combined with H total , H QG limits evolution of the quantum state into the ground state of H p when H total evolves to H p without diagonalizing H p .

2. The method of claim 1 , comprising recording a selected probability mass function for a probability of the quantum state being in the ground state of H p when H total evolves to H p , and H QG is determined also based on the selected probability mass function and using the selected probability mass function in determining the additional quantum Hamiltonian.

3. The method of claim 1 , wherein the energy spectrum is obtained at approximately half time of a total time for H total to evolve from H i to H p .

4. The method of claim 1 , wherein deriving the data characterizing the energy spectrum of H total comprises evaluating ground state energy of H total using quantum Monte-Carlo techniques, mean-field theories, or Markus theory.

5. The method of claim 4 , wherein deriving the data characterizing the energy spectrum information about an energy spectrum of H total comprises deriving data characterizing the energy spectrum information about an energy spectrum of H total the data at a time of a quantum phase transition.

6. The method of claim 1 , wherein the information about an energy spectrum comprises actual energy levels, spacing among the actual energy levels, distribution of the spacing among the actual energy levels, or an average spacing between adjacent average energy levels.

7. The method of claim 1 , wherein calculating the average phonon energy of the bath comprises calculating the average phonon energy using an open quantum systems model.

8. The method of claim 7 , wherein the average phonon energy satisfies the following equation:

ω
_

=

∑
0
∞

⁢

ω
⁢

⁢

J
⁡

(
ω
)

⁢
d
⁢

⁢

ω
/

(

e

ω
/
kT

-
1

)

∑
0
∞

⁢

J
⁡

(
ω
)

⁢
d
⁢

⁢

ω
/

(

e

ω
/
kT

-
1

)

,

where J(ω) is one of an Omhic spectral density:

J
⁡

(
ω
)

=

λω
⁢

⁢

e

-

ω
γ

,

a super-Omhic spectral density:

J
⁡

(
ω
)

=

λω
⁢

3

⁢

e

-

ω
γ

,

a Drude-Lorentz spectral density:

J
⁡

(
ω
)

Assignees

• Google Llc

Inventors

• Mohseni Masoud
• Neven Hartmut

Classifications

• G06N7/01

  Probabilistic graphical models, e.g. probabilistic networks · CPC title

• G06F15/82

  data or demand driven · CPC title

• G06F17/11

  for solving equations {, e.g. nonlinear equations, general mathematical optimization problems (optimization specially adapted for a specific administrative, business or logistic context G06Q10/04)} · CPC title

• G06N20/00

  Machine learning · CPC title

• G06N7/005

  Physics · mapped topic