Modeling and Classification of Topological Phases of Matter Zhenghan Wang

advertisement
Modeling and Classification of
Topological Phases of Matter
Zhenghan Wang
Microsoft Station Q & UC Santa Barbara
Texas, March 24, 2015
Microsoft Station Q
Search for non-abelian anyons in topological phases
of matter, and build a topological quantum computer
Theory: Station Q,…
Experiment: Delft, Copenhagen,…
Computer Science: QuArc
Mike Freedman (director, math-Fields medalist), Chetan Nayak (physics),
Kevin Walker (math), Matt Hastings (physics), Parsa Bonderson (physics),
Roman Lutchyn (physics), Bela Bauer (physics)
+ collaborators
+ postdocs
+graduate students
+visitors
http://stationq.ucsb.edu/
As a mathematician working also in physics…
For his contributions to the theory of the atomic nucleus
and the elementary particles, particularly through the discovery
and application of fundamental symmetry principles
Convergence of Physics and Mathematics
Physics
Mathematics
Newtonian Mechanics
Calculus (arranged marriage)
General Relativity and Gauge theory
Differential Geometry
Quantum Mechanics
Linear Algebra and Functional Analysis
Many-body Entanglement Physics
(second revolution in quantum mechanics?)
???
Universal Properties of 2D Topological Phases
Topological Quantum Field Theory (TQFT)
and Modular Tensor Category (MTC)
2D Topological Phases in Nature
•
Quantum Hall States
1980 Integral Quantum Hall Effect --von Klitzing (1985 Nobel)
1982 Fractional QHE---Stormer, Tsui, Gossard at ๐œˆ =
1
3
(1998 Nobel for Stormer, Tsui, and Laughlin)
1987 Non-abelian FQHE???---R. Willett et al at ๐œˆ =
• T
โ€‹ opological superconductors
• Topological insulators
•
โ‹ฏ
5
2
๐‘…๐ป = ๐œˆ −1
โ„Ž
๐‘’2
๏พ100 Fractional Quantum Hall States
๐‘
๏ฎ=๐‘ ๐‘’
๏ฆ
filling factor or fraction
๐‘๐‘’ = # of electrons
๐‘๏ฆ =# of flux quanta
Topological phases at a filling
fraction are modeled by
TQFTs/UMTCs/CFTs/…
(or pattern of zeros)
1/3
2/3
4/3
5/3
7/3
8/3
1/5 1/7 1/9 2/11 2/13 2/15
2/5 2/7 2/9 3/11 3/13 4/15
3/5 3/7 4/9 4/11 4/13 7/15
4/5 4/7 5/9 5/11 5/13 8/15
6/5 5/7 7/9 6/11 6/13 11/15
7/5 9/7 11/9 7/11 7/13 22/15
8/5 10/7 13/9 8/11 10/13 23/15
11/5 12/7 25/9 16/11 20/13
12/5 16/7
17/11
19/7
m/5, m=14,16, 19
2/17
3/17
4/17
5/17
6/17
8/17
9/17
3/19 5/21 6/23 6/25
4/19 10/21
5/19
9/19
10/19
5/2
7/2
19/8
Pan et al (2008)
Topological Phases of Quantum Matter
Local Hilbert Space
๐ฟ =⊗๐‘– ๐ฟ๐‘–
Local, Gapped Hamiltonian
E
Egap
๐ป: ๐ฟ → ๐ฟ
Two gapped Hamiltonians ๐ป1 , ๐ป2 realize the
same phase of matter if there exists a
continuous path connecting them without
closing the gap/a phase transition.
A topological phase, to first approximation, is a class of gapped
Hamiltonians that realize the same phase. Topological order in
a 2D topological phase is encoded by a TQFT or anyon
model=unitary modular tensor category (MTC) or CFT.
Atiyah-Segal Type (2+1)-TQFT: Codim=1
A symmetric monoidal “functor” (๐‘‰, ๐‘):
category of 2-3-mfds ๏ƒ  Vec
2-mfd ๐‘Œ ๏ƒ  vector space ๐‘‰(๐‘Œ)
3-bord ๐‘‹ from ๐‘Œ1 to ๐‘Œ2 ๏ƒ  Z(X): ๐‘‰ ๐‘Œ1 → ๐‘‰ ๐‘Œ2
•
•
•
•
•
๐‘‰
๐‘‰
๐‘‰
๐‘
๐‘
∅ =โ„‚
๐‘Œ1 โŠ” ๐‘Œ2 ≅ ๐‘‰ ๐‘Œ1 ⊗ ๐‘‰(๐‘Œ2 )
−๐‘Œ ≅ ๐‘‰ ∗ (๐‘Œ)
๐‘Œ × ๐ผ = Id๐‘‰ ๐‘Œ
๐‘‹1 ∪ ๐‘‹2 = ๐œ… ๐‘š ⋅ ๐‘ ๐‘‹1 ⋅ ๐‘(๐‘‹2 ) (anomaly)
๐‘‹1
๐‘Œ1
๐‘‹2
๐‘Œ2
๐‘Œ3
TQFTs and Higher Category Theories
Basic Principle:
Physics is local, so realistic TQFTs are determined by local data.
(n+1)-Topological Quantum Field Theories
(2+1)-TQFTs
(n+1)-Categories
Modular Tensor Categories
Quantum Finite Group Algebras
1. Not fully extended. Not covered by Lurie’s cobordism hypothesis.
2. Frontiers are in (3+1)D both mathematically and physically:
(2+1)-TQFTs are unemployed---no major topological problems to solve,
(3+1)-TQFTs that can detect smooth structures are highly desired.
Generalization of Two Theorems
• Landau’s Theorem:
Given ๐‘Ÿ, there are only finitely many groups
with exactly ๐‘Ÿ irreducible representations.
• Cauchy Theorem:
Given a finite group G, the prime factors of
the order of G and the exponent of G are the
same set.
A modular (tensor) category is a spherical fusion
category with a non-degenerate braiding
A fusion category is a categorification of
a based ring โ„ค[๐’™๐ŸŽ , … , ๐’™๐’“−๐Ÿ ]
finite rigid โ„‚-linear semisimple monoidal category with simple unit
monoidal: (⊗, ๐Ÿ),
semisimple: ๐‘‹ ≅ ๐‘– ๐‘š๐‘– ๐‘‹๐‘– ,
linear: Hom ๐‘‹, ๐‘Œ ∈ Vecโ„‚,
rigid: ๐‘‹ ∗ ⊗ ๐‘‹ โ†ฆ ๐Ÿ โ†ฆ ๐‘‹ ⊗ ๐‘‹ ∗
finite rank: Irr ๐’ž = {๐Ÿ = ๐‘‹0 , … , ๐‘‹๐‘Ÿ−1 }
๐‘ฟ simple if ๐‡๐จ๐ฆ ๐‘ฟ, ๐‘ฟ = โ„‚
Rank of ๐’ž: r ๐’ž = ๐‘Ÿ = dim ๐‘‰(๐‘‡ 2 )
Spherical Fusion Category
• Rigidity defines a functor ∗∗ : ๐’ž → ๐’ž. A pivotal structure is a natural
isomorphism between the identity functor IdC and ∗∗ .
Define a left trace and a right trace for any morphism ๐‘“: ๐‘ฅ → ๐‘ฅ:
• A pivotal structure is spherical if the two traces are equal.
• A fundamental open question:
Is every fusion category pivotal/spherical?
Modular Category
A fusion category is braided if there exist braidings
๐‘๐‘Ž,๐‘ : ๐‘Ž ⊗ ๐‘
๐‘⊗๐‘Ž
satisfy hexagons.
A simple object a is
transparent if for any simple ๐‘,
๐‘๐‘,๐‘Ž ⋅ ๐‘๐‘Ž,๐‘ = ida⊗๐‘ .
A braiding is non-degenerate
if the only transparent simple
is the tensor unit.
Examples
• Pointed: ๐’ž(๐ด, ๐‘ž), ๐ด finite abelian group, ๐‘ž non-degenerate
quadratic form on ๐ด.
• Rep(๐ท๐œ” ๐บ), ๐œ” a 3-cocycle on ๐บ a finite group.
• Quantum groups/Kac-Moody algebras: subquotients of
Rep(๐‘ˆ๐‘ž ๐”ค) at ๐‘ž = ๐‘’ ๐œ‹๐‘– ๐‘™ or level ๐‘˜ integrable ๐”ค-modules, e.g.
– SU ๐‘ ๐‘˜ = ๐’ž(๐”ฐ๐”ฉ๐‘ , ๐‘ + ๐‘˜),
– SO ๐‘ ๐‘˜ ,
– Sp ๐‘ ๐‘˜ ,
– for gcd(๐‘, ๐‘˜) = 1, PSU N k ⊂ SU ๐‘ ๐‘˜ “even half”
• Drinfeld center: ๐’ต(๐’Ÿ) for spherical fusion category ๐’Ÿ.
Invariants of Modular Tensor Category
MTC ๐’ž
(2+1)-TQFT (๐‘‰, ๐‘)
• Pairing ๐‘Œ 2 , ๐’ž = ๐‘‰ ๐‘Œ 2 ; ๐’ž ∈ Rep(โ„ณ ๐‘Œ 2 ) for a
surface ๐‘Œ 2 , โ„ณ ๐‘Œ 2 = mapping class group
• Pairing ๐‘๐‘‹,๐ฟ,๐’ž = ๐‘‹ 3 , ๐ฟ๐ถ , ๐’ž ∈ โ„‚ for colored framed
oriented links ๐ฟ๐‘ in 3-mfd ๐‘‹ 3
fix ๐’ž, ๐‘๐‘‹,๐ฟ,๐’ž invariant of (๐‘‹ 3 , ๐ฟ๐‘ )
fix (๐‘‹ 3 , ๐ฟ๐‘ ), ๐‘๐‘‹,๐ฟ,๐’ž invariant of ๐’ž
fix ๐‘Œ 2 , ๐‘‰(๐‘Œ 2 ; ๐’ž) invariant of ๐’ž
Quantum Dimensions and Twists: Unknot
• Label set ๐ฟ = isomorphism classes of simple objects
• Quantum dimension of a simple/label ๐‘Ž ∈ ๐ฟ:
• Topological twist/spin of ๐‘Ž: finite order by Vafa’s thm
• Dimension ๐ท2 of a modular category:
dim(๐’ž) = ๐‘ซ๐Ÿ = ∑๐’‚∈๐‘ณ ๐’…๐Ÿ๐’‚
Modular S-Matrix: Hopf Link
• Modular S-matrix:
๐‘†๐‘Ž๐‘ =
• Modular T-matrix:
๐‘‡๐‘Ž๐‘ = ๐›ฟ๐‘Ž๐‘ ๐œƒ๐‘Ž -diagonal
• (๐‘†, ๐‘‡)-form a projective rep. of ๐‘†๐ฟ(2, โ„ค):
0
1
1
๐‘ก=
0
๐‘ =
−1
0
1
1
S
T
Modular Data
1.
2.
3.
4.
5.
6.
๐‘† = ๐‘† ๐‘ก , ๐‘†๐‘† ๐‘ก = ๐ท 2 Id, modular ๐‘‡ diagonal, ord ๐‘‡ = ๐‘ < ∞
๐‘†๐‘‡
3
= ๐‘+
๐‘๐‘–๐‘—๐‘˜ = ∑๐‘Ž
๐‘† 2,
๐‘+ ๐‘− =
๐‘†๐‘–๐‘Ž ๐‘†๐‘—๐‘Ž ๐‘†๐‘˜๐‘Ž
๐ท2 ,
๐‘+ ๐‘
๐‘−
=1
∈ โ„•, Verlinde formulas for fusion rules
๐ท 2 ๐‘‘๐‘Ž
๐œƒ๐‘– ๐œƒ๐‘— ๐‘†๐‘–๐‘— = ∑๐‘˜ ๐‘๐‘–๐‘˜∗ ๐‘— ๐‘‘๐‘˜ ๐œƒ๐‘˜ ,where ๐‘๐‘–๐‘–0∗ uniquely defines ๐‘– ∗ .
๐‘›
2๐œ‹๐‘–
1
๐œƒ๐‘–
๐‘˜
๐œˆ๐‘› ๐‘˜ โ‰” 2 ∑๐‘–,๐‘— ๐‘๐‘–๐‘— ๐‘‘๐‘– ๐‘‘๐‘—
∈ โ„ค[๐‘’ ๐‘ ] satisfies:
๐ท
๐œƒ๐‘—
๐œˆ2 ๐‘˜ ∈ {0, ±1}
โ„š ๐‘† ⊂ โ„š(๐‘‡), Aut โ„š โ„š ๐‘† ⊂ ๐”–๐‘Ÿ , Aut โ„š
๐‘± โ‰” ∑๐‘— ๐‘‘๐‘—2 ๐œƒ๐‘—±1
๐‘†
โ„š ๐‘‡ ≅ โ„ค2
๐‘˜
๐‘๐‘–๐‘—๐‘˜ = dim Hom(๐‘‹๐‘– ⊗ ๐‘‹๐‘— , ๐‘‹๐‘˜ )
Rank-Finiteness
Theorem (Bruillard-Ng-Rowell-W., 2013):
For a fixed rank, there are only finitely many
equivalence classes of modular categories.
Remarks
.
1. Refinement of Ocneanu rigidity: fix the fusion rule, finite
2. Rank-finiteness for fusion/spherical fusion categories open.
3. An explicit bound and effective algorithm.
Finite Group Analogue
Theorem (E. Landau 1903)
For any ๐‘Ÿ ∈ โ„•, there are finitely many groups ๐บ with Irr ๐บ
= ๐‘Ÿ.
Proof.
Use class equation:
๐บ = ∑๐‘Ÿ๐‘–=1 |๐‘”๐‘– |,
๐‘”๐‘– distinct conjugacy classes. Set ๐‘ฅ๐‘– = [๐บ: ๐ถ ๐‘”๐‘– ] (index of
centralizers) to get
1
1 = ∑๐‘Ÿ๐‘–=1 .
๐‘ฅ๐‘–
๐‘ฅ๐‘– ≤ ๐‘Ž ๐‘Ÿ where ๐‘Ž 1 = 2, ๐‘Ž 2 = 3, ๐‘Ž ๐‘› = ๐‘Ž ๐‘› − 1 ๐‘Ž ๐‘› − 2 + 1
is Sylvester’s sequence. Therefore ๐บ = max ๐‘ฅ๐‘– is bounded.
โ–ก
๐‘–
Dimension Equation
2
Fix rank ๐‘Ÿ, dim(๐’ž) = ๐ท2 = ๐‘‘02 + ๐‘‘12 + โ‹ฏ + ๐‘‘๐‘Ÿ−1
, ๐‘‘0 = 1
Rewrite: 1 − ๐‘ซ๐Ÿ + ๐’…๐Ÿ๐Ÿ + โ‹ฏ + ๐’…๐Ÿ๐’“−๐Ÿ = 0
Quantum dimensions ๐’…๐Ÿ๐’‚ and ๐‘ซ๐Ÿ are special algebraic integers:
๐“ข-units
Let ๐•‚ be a number field and ๐’ฎ ∈ Spec ๐’ช๐•‚ be finite. The ๐’ฎ-units:
×
๐’ช๐•‚,๐’ฎ
= {๐‘ฅ ∈ ๐•‚ โˆฃ ๐‘ฅ =
๐”ญ๐›ผ๐”ญ }
๐”ญ∈๐’ฎ
where ๐›ผ๐”ญ ∈ โ„ค.
×
๐’ช๐•‚,๐’ฎ
= {๐‘ฅ ∈ ๐•‚× โˆฃ ๐‘ฅ
๐‘ฃ
= 1 for all ๐‘ฃ ∉ ๐’ฎ}
Reduction to Evertse’s Theorem
Theorem (Evertse 1984)
There are finitely many solutions to 0 = 1 + ๐‘ฅ0 + โ‹ฏ + ๐‘ฅ๐‘Ÿ−1 with ๐‘ฅ๐‘– ∈
×
๐’ช๐•‚,๐’ฎ
such that no sub-sum of 1 + ๐‘ฅ0 + โ‹ฏ + ๐‘ฅ๐‘Ÿ−1 vanishes.
Set ๐‘š = lcm(ord ๐‘‡ ) for all rank = ๐‘Ÿ modular ๐‘‡, ๐•‚ = โ„š(๐‘’
๐’ฎ = {๐‘ ๐‘– ∈ Spec ๐’ช๐•‚ โˆฃ ๐‘ ๐‘– |๐“… ∈ โ„ณ๐‘Ÿ ? ? ? }
2๐œ‹๐‘–
๐‘š
).
Evertse’s Theorem implies:
− dim ๐’ž , ๐‘‘1 2 , … , ๐‘‘๐‘Ÿ−1 2 < ∞.
Hence dim(๐’ž) is bounded. By Verlinde formulas, only finitely many
fusion rules. Rank-finiteness follows from Ocneanu rigidity.
Need to show: โ„ณ๐‘Ÿ is finite, and ๐’…๐Ÿ๐’‚ and ๐๐ข๐ฆ(๐“’) are ๐“ข-units
Prime Factorization
• For a rank = ๐‘Ÿ modular category ๐’ž, N=ord(T), modular T:
๐’ฎ๐’ž = {๐“… ∈ Spec โ„ค ๐œ๐‘ โˆฃ ๐“…| dim ๐’ž }, and
โ„ณ๐’ž = {๐“… ∈ Spec โ„ค ๐œ๐‘ โˆฃ ๐“… ord ๐‘‡ .
• For a fixed rank ๐‘Ÿ,
๐’ฎ๐‘Ÿ โ‰”
๐’ฎ๐’ž
rank ๐’ž =๐‘Ÿ
and
โ„ณ๐‘Ÿ โ‰”
โ„ณ๐’ž
rank ๐’ž =๐‘Ÿ
Cauchy Theorem for Modular Categories
Theorem (Bruillard-Ng-Rowell-W.)
๐“œ๐“’ = ๐“ข๐“’ , i.e. prime divisors of ๐๐ข๐ฆ(๐“’) and ๐ =
๐จ๐ซ๐(๐‘ป) in โ„ค(๐œ๐‘ ) form the same set.
• โ„ณ๐‘Ÿ = ๐’ฎ๐‘Ÿ for any rank ๐‘Ÿ, ๐‘‘๐‘Ž2 and dim ๐’ž are ๐“ข-units.
• Finiteness of โ„ณ๐‘Ÿ follows from Ng-Schauenburg
congruence subgroup theorem for modular rep. of
SL(2, โ„ค).
Classification of Unitary Modular Categories
rank = ๐Ÿ, ๐Ÿ‘, ๐Ÿ’ with Rowell and Stong, rank = ๐Ÿ“ with Bruillard, Ng, Rowell
A
1
Trivial
A
2
NA
Semion
2
Fib
BU
A
2
โ„ค3
NA
8
NA
2
(SO(3), 5)
Ising
BU
A
5
Toric Code
A
4
โ„ค4
NA
4
Fib × Semion
BU
NA
2
NA
(SO(3), 7)
BU
3
DFib
BU
The ๐‘–th-row lists all rank = ๐‘– unitary modular tensor categories.
Middle symbol: the fusion rule.
Upper left corner: A = abelian theory, NA = non-abelian.
Upper right corner number = the number of distinct theories.
Lower left corner BU = there is a universal braiding anyon.
Realization as Topological Phase
Kitaev’s Toric Code: H= -๏ƒฅv Av -๏ƒฅpBp
=T2
v
p
L=๏ƒ„๐’†๐’…๐’ˆ๐’†๐’” โ„‚2
Av=๏ƒ„๐’†๏ฅ๐’— ๏ณz ๏ƒ„๐’๐’•๐’‰๐’†๐’“๐’” Ide,
Bp=๏ƒ„๐’†๏ฅ๐’‘ ๏ณx ๏ƒ„๐’๐’•๐’‰๐’†๐’“๐’” Ide,
0 1
1 0
1 0
๐œŽ๐‘ง =
0 −1
๐œŽ๐‘ฅ =
Toric Code Exactly Solvable
• Av, Bp all commute with each other
• Ground states are ๏€ โ„‚4, i.e. 4-fold degenerate
• Gapped in the thermodynamic limit: ๐œ†1 − ๐œ†0 >c>0
• Excitations are mutual anyons
Unitary Modular Category Realized by Toric Code
• 4 types of simple objects=anyons {1, ๐‘’, ๐‘š, ๐œ“}:
1=ground state or vaccum, ๐‘’, ๐‘š=bosons, ๐œ“=fermion,
๐‘’ ⊗ ๐‘’ = 1, ๐‘š ⊗ ๐‘š = 1, ๐‘’ ⊗ ๐‘š = ๐œ“
The fusion rule same as โ„ค2 ⊕โ„ค2 .
•
The anyons form a Drinfeld center ๐ท ๐‘2 :
Herbertsmithite
Physical Theorem (Jiang-W.-Balents):
The spin=12 Heisenberg anti-ferromagnetic Kagome model
๐’š
๐’š
๐’š
๐’š
H=๐‘ฑ๐Ÿ ∑<๐’Š๐’‹> (๐ˆ๐’™๐’Š ๐ˆ๐’™๐’‹ +๐ˆ๐’Š ๐ˆ๐’‹ +๐ˆ๐’›๐’Š ๐ˆ๐’›๐’‹ )+๐‘ฑ๐Ÿ ∑โ‰ช๐’Š๐’‹โ‰ซ (๐ˆ๐’™๐’Š ๐ˆ๐’™๐’‹ +๐ˆ๐’Š ๐ˆ๐’‹ +๐ˆ๐’›๐’Š ๐ˆ๐’›๐’‹ )
represents a topological phase of matter which is in the same universality class
of the toric code when 0<๐ฝ2 ๐ฝ1 <0.15, where <ij> means summation over the
nearest neighborhood and <<ij>> the next nearest neighborhood.
How to identify the MTC/TQFT?
Entanglement and classification of MTCs.
Entanglement
• Relative to locality:
Hilbert space of states ๐ฟ =⊗๐‘–∈๐‘  ๐ฟ๐‘– decomposed into parts ๐ฟ๐‘– with dim ๐ฟ๐‘– > 1,
a state ๐œ“ is a product if ๐œ“ = ๐œ“๐›ผ ⊗ ๐œ“๐›ฝ for some states ๐œ“๐›ผ , ๐œ“๐›ฝ , ๐›ผ ∪ ๐›ฝ = ๐‘ .
Otherwise, a state is entangled.
In quantum computation, ๐ฟ๐‘– = โ„‚2 -called a qubit.
Spin-singlet ๐œ“ = (|01 > − |10 >)/√2 entangled.
• Whole is more definite than parts:
Spin-singlet ๐œ“ pure, but each qubit in a mixed state.
• Spooky action at a distance:
Measuring one results a definite state of the other.
von Neumann Entropy
Topological Entanglement Entropy (TEE)
Preskill-Kitaev, Levin-Wen
2006
1
๐‘บ๐’—๐‘ต = ๐œถ๐‘ณ − ๐œธ + ๐’ช( ),
๐ฟ
๐ฟ = length of the smooth boundary.
TEE γ quantifies long-range entanglement.
Identify MTC/TQFT
Compute topological entanglement entropy:
๐›พ = 0, trivial.
๐›พ ≠ 0, some non-trivial MTC, ๐œธ = ๐ฅ๐ง ๐‘ซ.
But which one?
Use mathematical classification to identify:
Fix ๐›พ, only finitely many MTCs.
For simple cases, completely classified. In some cases,
there are extra information to identify the MTC/TQFT,
e.g. spin=12 Heisenberg anti-ferromagnetic Kagome model
Topological Quantum Computation
Topological nanowires
34
Download