On the spectrum of Hamiltonians in finite dimensions Roberto Oliveira Paraty, August 14 th 2007. Joint with David DiVincenzo and Barbara IBM Watson.

Slides:



Advertisements
Similar presentations
Subspace Embeddings for the L1 norm with Applications Christian Sohler David Woodruff TU Dortmund IBM Almaden.
Advertisements

SADC Course in Statistics Importance of the normal distribution (Session 09)
Metric Embeddings with Relaxed Guarantees Hubert Chan Joint work with Kedar Dhamdhere, Anupam Gupta, Jon Kleinberg, Aleksandrs Slivkins.
1 Methods of Experimental Particle Physics Alexei Safonov Lecture #21.
Introduction to Molecular Orbitals
Week 1: Basics Reading: Jensen 1.6,1.8,1.9. Two things we focus on DFT = quick way to do QM – 200 atoms, gases and solids, hard matter, treats electrons.
Boris Altshuler Columbia University Anderson Localization against Adiabatic Quantum Computation Hari Krovi, Jérémie Roland NEC Laboratories America.
No friction. No air resistance. Perfect Spring Two normal modes. Coupled Pendulums Weak spring Time Dependent Two State Problem Copyright – Michael D.
Complexity of simulating quantum systems on classical computers Barbara Terhal IBM Research.
From PET to SPLIT Yuri Kifer. PET: Polynomial Ergodic Theorem (Bergelson) preserving and weakly mixing is bounded measurable functions polynomials,integer.
SUMS OF RANDOM VARIABLES Changfei Chen. Sums of Random Variables Let be a sequence of random variables, and let be their sum:
Lecture 24: Thurs. Dec. 4 Extra sum of squares F-tests (10.3) R-squared statistic (10.4.1) Residual plots (11.2) Influential observations (11.3,
Modern Navigation Thomas Herring
Ch 9 pages ; Lecture 21 – Schrodinger’s equation.
Preperiodic Points and Unlikely Intersections joint work with Laura DeMarco Matthew Baker Georgia Institute of Technology AMS Southeastern Section Meeting.
1 CE 530 Molecular Simulation Lecture 7 David A. Kofke Department of Chemical Engineering SUNY Buffalo
McGraw-Hill/IrwinCopyright © 2009 by The McGraw-Hill Companies, Inc. All Rights Reserved. Chapter 7 Sampling Distributions.
2/6/2014PHY 770 Spring Lectures 7 & 81 PHY Statistical Mechanics 11 AM-12:15 PM & 12:30-1:45 PM TR Olin 107 Instructor: Natalie Holzwarth.
F.F. Assaad. MPI-Stuttgart. Universität-Stuttgart Numerical approaches to the correlated electron problem: Quantum Monte Carlo.  The Monte.
Quantum Two 1. 2 Time Independent Approximation Methods 3.
Merlin-Arthur Games and Stoquastic Hamiltonians B.M. Terhal, IBM Research based on work with Bravyi, Bessen, DiVincenzo & Oliveira. quant-ph/ &
Constraining theories with higher spin symmetry Juan Maldacena Institute for Advanced Study Based on & to appearhttp://arxiv.org/abs/
Central Limit Theorem Example: (NOTE THAT THE ANSWER IS CORRECTED COMPARED TO NOTES5.PPT) –5 chemists independently synthesize a compound 1 time each.
Lab 3b: Distribution of the mean
Chapter 7: Sample Variability Empirical Distribution of Sample Means.
SUPA Advanced Data Analysis Course, Jan 6th – 7th 2009 Advanced Data Analysis for the Physical Sciences Dr Martin Hendry Dept of Physics and Astronomy.
Barriers in Hamiltonian Complexity Umesh V. Vazirani U.C. Berkeley.
1 Chapter 7 Sampling Distributions. 2 Chapter Outline  Selecting A Sample  Point Estimation  Introduction to Sampling Distributions  Sampling Distribution.
Exam 1 next Thursday (March 7 th ) in class 15% of your grade Covers chapters 1-6 and the central limit theorem I will put practice problems, old exams,
Quantum Two 1. 2 Evolution of Many Particle Systems 3.
ECE 8443 – Pattern Recognition ECE 8527 – Introduction to Machine Learning and Pattern Recognition LECTURE 07: BAYESIAN ESTIMATION (Cont.) Objectives:
A simple nearest-neighbour two-body Hamiltonian system for which the ground state is a universal resource for quantum computation Stephen Bartlett Terry.
13. Extended Ensemble Methods. Slow Dynamics at First- Order Phase Transition At first-order phase transition, the longest time scale is controlled by.
Lecture 21: On to Finite Nuclei! 20/11/2003 Review: 1. Nuclear isotope chart: (lecture 1) 304 isotopes with t ½ > 10 9 yrs (age of the earth) 177.
Two-dimensional SYM theory with fundamental mass and Chern-Simons terms * Uwe Trittmann Otterbein College OSAPS Spring Meeting at ONU, Ada April 25, 2009.
Vibrational Motion Harmonic motion occurs when a particle experiences a restoring force that is proportional to its displacement. F=-kx Where k is the.
Lecture 8: Measurement Errors 1. Objectives List some sources of measurement errors. Classify measurement errors into systematic and random errors. Study.
STA302/1001 week 11 Regression Models - Introduction In regression models, two types of variables that are studied:  A dependent variable, Y, also called.
STATISTICAL MECHANICS PD Dr. Christian Holm PART 5-6 Some special topics, Thermal Radiation, and Plank distribution.
Charles University Charles University STAKAN III
Oliver Schulte Machine Learning 726
Density Matrix Density Operator State of a system at time t:
Time Dependent Two State Problem
CHAPTER 5 The Schrodinger Eqn.
Programming assignment #1. Solutions and Discussion
Stat 31, Section 1, Last Time Sampling Distributions
The units of g(): (energy)-1
4. Numerical Integration
Chapter 7: Sampling Distributions
CHAPTER 5 The Schrodinger Eqn.
Quantum mechanics from classical statistics
Lecture 10: Sketching S3: Nearest Neighbor Search
4. The Postulates of Quantum Mechanics 4A. Revisiting Representations
Ising Model of a Ferromagnet
CSCI B609: “Foundations of Data Science”
cse 521: design and analysis of algorithms
Adaptive Perturbation Theory: QM and Field Theory
QM2 Concept test 8.1 We have three non-interacting particles in a one-dimensional infinite square well. The energy of the three particle system is
Quantum Two Body Problem, Hydrogen Atom
The Pseudopotential Method Builds on all of this.
Lecture 15: Least Square Regression Metric Embeddings
Nonparametric density estimation and classification
QM2 Concept test 3.1 Choose all of the following statements that are correct about bosons. (1) The spin of a boson is an integer. (2) The overall wavefunction.
Accelerator Physics Statistical Effects
QM2 Concept test 8.1 We have three non-interacting particles in a one-dimensional infinite square well. The energy of the three particle system is
Introduction to topological superconductivity and Majorana fermions
Regression Models - Introduction
Quantum One.
Presentation transcript:

On the spectrum of Hamiltonians in finite dimensions Roberto Oliveira Paraty, August 14 th Joint with David DiVincenzo and Barbara IBM Watson.

In one slide: Ground state energy is hard. Bulk of the spectrum is Gaussian and universal.

The setup H = Hamiltonian on a set V of N spin ½ particles. Spec(H) = { 0 · 1 · 2 …}. We will assume Tr(H)=0.  2 (H) = “variance” = Tr(H 2 )/2 N Write H =  X ½ V H X, where H X acts on the spins in X. Can assume Tr(H X H Y )=0 if X  Y.

site term H {k} = h  z [k] E.g.: Ising with transverse field H = (bond terms) + (site terms) bond term H {i,j} = J  x  x [i,j] ij k H=  k H {k} +  i~j H {i,j}

site term H {k} = h  z [k] E.g.: Ising with transverse field H = (bond terms) + (site terms) bond term H {i,j} = J  x  x [i,j] ij k H=  k H {k} +  i~j H {i,j} +  F H F face term H F

Dimensionality assumption 9 metric  and d, C, c,  such that: Radius R center x B(x,R) = {v 2 V :  (v,x) · R} cR d ·  X ½ B(x,R)  2 (H X ) · CR d  X ½ B(x,R) ||H X || · CR d  X ½  B(x,R)  2 (H X ) · CR d- 

E.g. nearest neighbor in Z d L 1 norm ¼ R d terms inside ¼ R d-1 at the boundary Radius R

Gaussian spectrum Plot a histogram of Spec(H/  (H)). with small but fixed bin size b>0. That will approach a Gaussian as N adiverges, for fixed parameters.

A bit more formally There is a probability measure on the line given by H:  H = 2 -N  2 Spec(H)  We show that this measure is approximately normal in the sense that for all a<b, as N grows:  H (a  (H),b  (H)) ! (2  ) -1/2 s a b exp(-t 2 /2)dt

Even more formally Recall: strength inside ball ¼ R d, strength across boundary ¼ R d- , with C,c extra parameteres. Then for all a<b, |  H [a  (H),b  (H)]-(2  ) -1/2 s a b exp(-t 2 /2)dt| · D(C,c,d,  ) (Diam(  )) -  d/8

A bit less formally again Inside ¼ R d, boundary ¼ R d-   Radius R center x B(x,R) = {v 2 V :  (v,x) · R}  H/  (H) [x,y] ¼ (2  ) -1/2 s x y exp(-t 2 /2)dt

A simple case We will now explain a special case of the Theorem. Nearest neighbor interactions on a n x n patch of the planar square lattice (N=n 2 spins). Also assume that all terms in the Hamiltonian have norm of constant order.

site term H {k} = h  z [k] Back to that old slide H = (bond terms) + (site terms) bond term H {i,j} = J  x  x [i,j] ij k H=  k H {k} +  i~j H {i,j}

What do we do? Main idea: ignore terms acting on red lines (i.e. treat non-interacting systems). Then put them back in via a perturbation argument.

Omitting the interactions G =  s G s, subsystems might have high dimension. How does one compute the global spectrum? Answer: convolution of the individual spectral distributions. By the usual Central Limit Theorem,  G  G) is approximately Gaussian as long as some conditions are satisfied.

Which conditions? Many terms in the sum. The influence of any given term is small. m ¿ n 1/2 suffices.

Putting red lines back in

Next step Recall that we have.  H/  (H) = 2 -N  2 Spec(H)   We know that  G/  (G) is approx. gaussian.  G/  (G) = 2 -N  2 Spec(G)   G  We will show that  (H-G) is small. By a perturbation theory argument, this implies that  H/  (H) ¼  G/  (G).

 H-G) is small Variance ¼ # of qubits Total # of qubits ¼ n 2 Qubits on red lines ¼ m(n/m) 2 = n 2 /m )  (H-G) 2 ·  (H) 2 /m, small if m À 1

To conclude Take some 1 ¿ m ¿ n 1/2 (e.g. m=n 1/3 ). Then  G/  (G) is approx. Gaussian. Also  H/  (H) ¼  G/  (G). So we are done. The key step: m x m boxes have m 2 vertices but only ¼ m vertices on their boundaries.

General case Inside ¼ R d, boundary ¼ R d-   Radius R center x B(x,R) = {v 2 V :  (v,x) · R}  H/  (H) [a,b] ¼ (2  ) -1/2 s a b exp(-t 2 /2)dt

General result: proof sketch Main idea: break the system into subparts with small total boundary strength. Treat isolated systems via standard CLT. Put the boundary back in via perturbation theory.

Conclusions Spectral distribution is approximately Gaussian with std. deviation ¼ N 1/2. This is universal for quantum spin systems in finite dimensional structures when long-range interactions decay fast enough.

Further work Bounds are actually weak for many problems. Are there better bounds for specific systems? Fermions? Bosons? Applications?