13 CHCl 3 Ensemble NMR Implementation of the Deutsch- Jozsa - - PowerPoint PPT Presentation

13 chcl 3 ensemble nmr implementation of the deutsch
SMART_READER_LITE
LIVE PREVIEW

13 CHCl 3 Ensemble NMR Implementation of the Deutsch- Jozsa - - PowerPoint PPT Presentation

13 CHCl 3 Ensemble NMR Implementation of the Deutsch- Jozsa Algorithm Dennis V. Perepelitsa with Brian J. Pepper 27 April 2007 8.14 Junior Lab MIT Department of Physics Outline of Talk 1. Introduction 2. Quantum Computation 3. Initial


slide-1
SLIDE 1

13CHCl3 Ensemble NMR

Implementation of the Deutsch- Jozsa Algorithm

Dennis V. Perepelitsa with Brian J. Pepper 27 April 2007 8.14 Junior Lab MIT Department of Physics

slide-2
SLIDE 2

Outline of Talk

  • 1. Introduction
  • 2. Quantum Computation
  • 3. Initial State Preparation
  • 4. Deutsch-Jozsa Algorithm
  • 5. Experimental Setup
  • 6. Calibration
  • 7. Experimental Results
  • 8. Error Analysis
  • 9. Conclusion

... Introduction DVP 4/27/2007

slide-3
SLIDE 3

Introduction

  • 1985: Feynman proposes quantum computation,

characterizes quantum logic gates.

  • 1985: Deutsch characterizes the universal quantum

computer.

  • 1992: Deutsch and Jozsa present an algorithm that

requires less operations on a quantum computer.

  • 1997: Gershenfeld and Chuang propose an

implementation using bulk-ensemble spin systems.

  • 1997: Cory, et. al. propose using NMR spectroscopy.

... Quantum Computation DVP 4/27/2007

slide-4
SLIDE 4

Quantum Computation

  • Key Idea #1: Every pair of spin eigenstates is a qubit.
  • Key Idea #3: Implement a function f with a unitary operator Uf.
  • Key Idea #2: Operations are implemented with RF pulses.
  • Important example: controlled-not. Implements f(x) = x.
  • Ensemble NMR --> measure expectation values.

... Initial State Preparation DVP 4/27/2007

slide-5
SLIDE 5

Initial State Preparation

The Boltzmann distribution of the energy eigenspectrum is Problem: We need the pure state |00> ! Solution: Time-average to obtain a pseudopure state.

  • Take three trials, cyclically permuting the non-|00> elements.
  • The average of the three is the action on the pure state |00> !
  • Finally, fire a “read out” pulse and observe the FID signal.

... Deutsch-Jozsa Algorithm

Density matrix leads to spectrum

DVP 4/27/2007

slide-6
SLIDE 6

Deutsch-Jozsa Algorithm (1992)

  • Classically: Look at both sides to decide if a coin is fair. (2 evaluations)
  • Quantum computation can do better!
  • Enumerate (and implement) all possible functions:
  • Apply pulse sequence to |00>
  • If f(0) = f(1): We see |00>
  • If f(0) != f(1): We see |10>

... Experimental Setup DVP 4/27/2007

slide-7
SLIDE 7

Experimental Setup

(Heavily modified from 8.14 Quantum Information Processing lab guide )

... Calibration DVP 4/27/2007

slide-8
SLIDE 8

Calibration

... Calibration (cont)

( xkcd.com – used under the Creative Commons License 2.5 )

DVP 4/27/2007

slide-9
SLIDE 9

Calibration

DVP 4/27/2007 ... Calibration (cont)

Fourier transform into frequency domain.

slide-10
SLIDE 10

Calibration

... Experimental Results (cont)

180-T-90 pulse sequence to obtain T1

DVP 4/27/2007

Vary pulse width and measure net magnetization to find pw-90

slide-11
SLIDE 11

Experimental Results

... Experimental Results (cont) DVP 4/27/2007

Calibration from the thermal state proton carbon Theoretical Prediction: Uf1, Uf2 (6,0) (6,0) Uf3, Uf4 (-6,0) (0,6)

slide-12
SLIDE 12

Experimental Results

... Experimental Results (cont)

Uf1 Uf2

DVP 4/27/2007

slide-13
SLIDE 13

Experimental Results

... Error Analysis

Uf4 Uf3

DVP 4/27/2007

slide-14
SLIDE 14

Error Analysis

  • Integrate absolute peaks numerically.
  • Random error calculated from 20 repeated “baseline” spectra ~ 1%
  • Uncertainty from fluctuations in the background ~ .5%
  • Calibrate from thermal state, propagate error.

But:

  • Deviation from theoretical prediction typically ~10-20%.
  • Our error does not cover this!

Significant systematics:

  • Asymmetric lineshapes due to improper shimming.
  • Imperfect pulse widths – deterioration over time.
  • Inhomogeneous magnetic field.
  • Inhomogeneous RF field.

... Conclusion DVP 4/27/2007

slide-15
SLIDE 15

Conclusion

We have:

  • Presented an implementation of Deutsch-Jozsa using ensemble NMR.
  • Calibrated computing system.
  • Successfully performed DJ on all four functions.
  • Discussed sources of error.

Stay tuned:

  • Brian Pepper on Grover's Algorithm.

Questions? DVP 4/27/2007