IQI Seminar Apr' 9, 2002

1
Quantum search by measurement
Andrew J. Landahl
Andrew Childs, MITEnrico Deotto, MITEdward
Farhi, MITJeffrey Goldstone, MITSam Gutmann,
Northeastern
Based on quant-ph/0204013
IQI SeminarApr. 9, 2002
2
How well can Nature compute?
The Church-Turing Thesis
The Church-Turing-Deutsch Thesis
All physical systems can be efficiently simulated
by a Turing machine.
All physical systems can be efficiently simulated
by a quantum Turing machine.
Important to challenge!!
3
Models of quantum computation
Measurement circuit
Unitary circuit
Adiabatic
Topological field theory
4
Adiabatic Theorem Born, Fock 28
Qualitative
System stays in ground state of if
changed slowly enough.
Quantitative
If , then the probability of
atransition
5
Adiabatic algorithms Farhi et al. 00
Search problem
Find the minimum of the function
Adiabatic solution

• Encode in the Hamiltonian .

• Prepare the ground state of
  .

• Evolve adiabatically from to .

• Measure the final energy .

N.B. By Lloyd 96 this can be simulated by a
quantum circuit.
6
Adiabatic algorithm for 3SAT
Instance
Set U of one-bit variables and collection C of
conjunctive clauses of at most three literals,
where a literal is a variable or a negated
variable in U.
Solution
A truth assignment for U.
Example
As a minimization problem
Minimize
, where
Adiabatic solution
, where
7
Quantum measurement algorithms?
Can we simulate adiabatic algorithms by the Zeno
effect? How well?
Two questions
How many measurements? How long does it take to
measure?
Perturbation theory
Phase estimation
8
Quantum Zeno Effect
Point-of-view 1
fixed projective measurements suppress
unitary evolution with probability .
Point-of-view 2
rotating projective measurements evolve
states by with probability .
9
Quantum Zeno Effect
Pictorial analogy
Malus Law (Point-of-view 2)
http//230nsc1.phy-astr.gsu.edu/hbase/phyopt/polcr
oss.html
10
Quantum Zeno Effect
Mathematical example
Unitary
Measurement
measurements every
Point-of-view 2
Point-of-view 1
11
How many measurements?
Perturbation theory
So after M perfectly projective measurements, the
probability of staying in the ground state is
12
The System-Meter Model
Interaction
Evolution
13
The System-Meter Model
Measurement complexity
Suppose we can resolve all displacementsto a
resolution
Then if
We can measure if
14
The good, the bad, and the ugly
The good

• Measurement algorithm polynomially related to
  adiabatic algorithm

(Measurement)
(Adiabatic)
The bad

• Meter is continuous, not digital.

The ugly

• Measurements arent perfectly projective.

15
Digital von Neumann Kitaev
Digitizing the system-meter model is phase
estimation!
Computational basis momentum eigenbasis
Pointer r qubits
Hamiltonian n qubits (digitizable a la Zalka
98)
Algorithm Phase estimation
16
Effects of finite precision
Under the system-meter evolution
The reduced system density matrix becomes
Where is the Hadamard (elementwise) product,
and
17
Effects of finite precision
Define
Then
If measurement were perfect, then we would have
. All we really need is
Which happens if
18
The Grover problem
Standard problem statement
Given a list of elements, all of which are
but one , find the by querying the list.
As a minimization problem
Minimize the function
19
Grover oracles
Unitary
Hamiltonian
Measurement
20
Adiabatic algorithm for Grover search
Features of the minimum gap
Features known independently of problem instance!
Naïve complexity
Actual complexity
Gap is only small for a time
21
Grover in two measurements!
Another feature of the minimum gap
Naïve complexity
Two-measurement algorithm
Actual complexity

• Measure
• Measure

Success probability ½!
Time
22
Summary

• Its important to challenge the
  Church-Turing-Deutsch thesis.
• The Zeno effect can efficiently simulate
  adiabatic evolution.
• Quantifying measurement complexity requires an
  explicit dynamical model.
• The digitized version of von Neumanns
  system-meter model is Kitaevs phase estimation
  algorithm.
• The Grover problem can be solved with two
  measurements, each taking time.