Skip to content
Draft
Show file tree
Hide file tree
Changes from all commits
Commits
File filter

Filter by extension

Filter by extension

Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
111 changes: 108 additions & 3 deletions src/entanglement.jl
Original file line number Diff line number Diff line change
Expand Up @@ -135,15 +135,35 @@ canonicalize_clip!(ps::Base.AbstractVecOrTuple{PauliOperator}, args...; kwargs..
"""
$TYPEDSIGNATURES

Get the bigram of a tableau.
The Bigram `B` of stabilizer endpoints represents the "span" of each stabilizer within a set of Pauli operators `𝒢 = {g₁,…,gₙ}`.

It is the list of endpoints of a tableau in the clipped gauge.
For each stabilizer `g`, the left endpoint `𝓁(g)` is defined as the minimum site `x` where `g` acts non-trivially, while the
right endpoint `𝓇(g)` is the maximum site where `g` acts non-trivially.

The site `x` represent the position within the system, taking values from `{1,2,…,n}` where `n` is the number of qubits.

The bigram set `B(𝒢)` encodes these endpoints as pairs:

`B(𝒢) ≡ {(𝓁(g₁),𝓇(g₁)),…,(𝓁(gₙ),𝓇(gₙ))}`

If `clip=true` (the default) the tableau is converted to the clipped gauge in-place before calculating the bigram.
Otherwise, the clip gauge conversion is skipped (for cases where the input is already known to be in the correct gauge).

Introduced in [nahum2017quantum](@cite), with a more detailed explanation of the algorithm in [li2019measurement](@cite) and [gullans2021quantum](@cite).

```jldoctest
julia> s = ghz(3)
+ XXX
+ ZZ_
+ _ZZ

julia> bigram(s)
3×2 Matrix{Int64}:
1 3
1 2
2 3
```

See also: [`canonicalize_clip!`](@ref)
"""
function bigram(state::AbstractStabilizer; clip::Bool=true)::Matrix{Int} # JET-XXX The ::Matrix{Int} should not be necessary, but they help with inference
Expand Down Expand Up @@ -174,16 +194,41 @@ the most performant one depending on the particular case.

Currently implemented are the `:clip` (clipped gauge), `:graph` (graph state), and `:rref` (Gaussian elimination) algorithms.
Benchmark your particular case to choose the best one.

See Appendix C of [nahum2017quantum](@cite).
"""
function entanglement_entropy end


"""
$TYPEDSIGNATURES

Get bipartite entanglement entropy of a contiguous subsystem by passing through the clipped gauge.

If `clip=false` is set the canonicalization step is skipped, useful if the input state is already in the clipped gauge.

See also: [`bigram`](@ref), [`canonicalize_clip!`](@ref)
```jldoctest
julia> using Graphs # hide

julia> s = ghz(3)
+ XXX
+ ZZ_
+ _ZZ

julia> entanglement_entropy(s, 1:3, Val(:clip))
0

julia> s = Stabilizer(Graph(ghz(4)))
+ XZZZ
+ ZX__
+ Z_X_
+ Z__X

julia> entanglement_entropy(s, [1,4], Val(:graph))
1
```

See also: [`bigram`](@ref), [`canonicalize_clip!`](@ref).
"""
function entanglement_entropy(state::AbstractStabilizer, subsystem_range::UnitRange, algorithm::Val{:clip}; clip::Bool=true)
# JET-XXX The ::Matrix{Int} should not be necessary, but they help with inference
Expand All @@ -196,6 +241,8 @@ end


"""
$TYPEDSIGNATURES

Get bipartite entanglement entropy by first converting the state to a graph and computing the rank of the adjacency matrix.

Based on "Entanglement in graph states and its applications".
Expand All @@ -210,6 +257,8 @@ end


"""
$TYPEDSIGNATURES

Get bipartite entanglement entropy by converting to RREF form (i.e., partial trace form).

The state will be partially canonicalized in an RREF form.
Expand All @@ -231,3 +280,59 @@ function entanglement_entropy(state::AbstractStabilizer, subsystem::AbstractVect
end

entanglement_entropy(state::MixedDestabilizer, subsystem::AbstractVector, a::Val{:rref}) = entanglement_entropy(state, subsystem, a; pure=nqubits(state)==rank(state))

"""
$TYPEDSIGNATURES

The mutual information between subsystems `𝒶` and `𝒷` in a stabilizer state is given by `I(𝒶, 𝒷) = S𝒶 + S𝒷 - S𝒶𝒷`.

```jldoctest
julia> using QuantumClifford

julia> using Graphs; using QuantumClifford: mutual_information # hide

julia> mutual_information(ghz(3), 1:2, 3:4, Val(:clip))
2

julia> s = Stabilizer(Graph(ghz(4)))
+ XZZZ
+ ZX__
+ Z_X_
+ Z__X

julia> mutual_information(s, [1,2], [3, 4], Val(:graph))
2
```

See Eq. E6 of [li2019measurement](@cite). See also: [`entanglement_entropy`](@ref)
"""
function mutual_information(state::AbstractStabilizer, A::AbstractVector{<:Integer}, B::AbstractVector{<:Integer}, alg::Val{:clip})
if !isempty(intersect(A, B))
throw(ArgumentError("Ranges A and B must not overlap."))
end
union_AB = union(A, B)
min_AB = minimum(union_AB)
max_AB = maximum(union_AB)
if length(union_AB) != max_AB - min_AB + 1
throw(ArgumentError("For the :clip algorithm, the union of A and B must form a contiguous range."))
end
contiguous_union = min_AB:max_AB
S_A = entanglement_entropy(state, A, alg)
S_B = entanglement_entropy(state, B, alg)
S_AB = entanglement_entropy(state, contiguous_union, alg)
return S_A + S_B - S_AB
end

function mutual_information(state::AbstractStabilizer, A::AbstractVector{<:Integer}, B::AbstractVector{<:Integer}, alg::Val{T}) where T
if !isempty(intersect(A, B))
throw(ArgumentError("Ranges A and B must not overlap."))
end
S_A = entanglement_entropy(state, A, alg)
S_B = entanglement_entropy(state, B, alg)
S_AB = entanglement_entropy(state, union(A, B), alg)
return S_A + S_B - S_AB
end

# Default method: use the :rref algorithm
mutual_information(state::AbstractStabilizer, A::AbstractVector{<:Integer}, B::AbstractVector{<:Integer}) =
mutual_information(state, A, B, Val{:rref}())
59 changes: 58 additions & 1 deletion test/test_entanglement.jl
Original file line number Diff line number Diff line change
Expand Up @@ -3,7 +3,8 @@

test_sizes = [1,2,10,63,64,65,127,128,129] # Including sizes that would test off-by-one errors in the bit encoding.

using QuantumClifford: stab_looks_good, destab_looks_good, mixed_stab_looks_good, mixed_destab_looks_good
using QuantumClifford
using QuantumClifford: stab_looks_good, destab_looks_good, mixed_stab_looks_good, mixed_destab_looks_good, mutual_information, entanglement_entropy

@testset "Clipped gauge of stabilizer states" begin
for n in test_sizes
Expand Down Expand Up @@ -50,4 +51,60 @@
@test entanglement_entropy(copy(s), subsystem, Val(:graph))==2
@test entanglement_entropy(copy(s), subsystem, Val(:rref))==2
end

@testset "Mutual information for Clifford circuits with entanglement_entropy check" begin
using QuantumOpticsBase
import QuantumOpticsBase: entanglement_entropy

for n in [4, 5, 6, 7] # exclude larger test sizes to avoid out of memory error
s = random_stabilizer(n)
a_start = rand(1:max(1, n-2))
a_end = rand(a_start:min(n-1, a_start+2))
subsystem_rangeA = a_start:a_end
b_start = min(n, a_end + rand(1:2))
b_end = rand(b_start:n)
subsystem_rangeB = b_start:b_end

if !isempty(intersect(subsystem_rangeA, subsystem_rangeB))
@test_throws ArgumentError mutual_information(copy(s), subsystem_rangeA, subsystem_rangeB, Val(:clip))
@test_throws ArgumentError mutual_information(copy(s), subsystem_rangeA, subsystem_rangeB, Val(:rref))
@test_throws ArgumentError mutual_information(copy(s), subsystem_rangeA, subsystem_rangeB, Val(:graph))
else
# Now we need to check if the union is contiguous for :clip
union_AB = union(subsystem_rangeA, subsystem_rangeB)
min_AB = minimum(union_AB)
max_AB = maximum(union_AB)
is_contiguous = (length(union_AB) == (max_AB - min_AB + 1))
if is_contiguous
mi_clip = mutual_information(copy(s), subsystem_rangeA, subsystem_rangeB, Val(:clip))
mi_rref = mutual_information(copy(s), subsystem_rangeA, subsystem_rangeB, Val(:rref))
mi_graph = mutual_information(copy(s), subsystem_rangeA, subsystem_rangeB, Val(:graph))
@test mi_clip == mi_rref == mi_graph
@test mi_clip ≥ 0
ψ = Ket(s)
ρ = dm(ψ)
S_A = QuantumOpticsBase.entanglement_entropy(ρ, subsystem_rangeA, entropy_vn)
S_B = QuantumOpticsBase.entanglement_entropy(ρ, subsystem_rangeB, entropy_vn)
# If A ∪ B covers the full system (1:n), set S_AB = 0 to avoid an invalid full-system trace in entanglement_entropy
S_AB = union_AB == (1:n) ? 0.0 : QuantumOpticsBase.entanglement_entropy(ρ, union_AB, entropy_vn)
# For a pure state: I(A:B) = [S(A) + S(B) - S(A∪B)] / 2, and convert nats → bits by dividing by log(2).
mi_indep = (S_A + S_B - S_AB) / (2 * log(2))
@test isapprox(mi_clip, mi_indep; atol=1e-6)
else
@test_throws ArgumentError mutual_information(copy(s), subsystem_rangeA, subsystem_rangeB, Val(:clip))
mi_rref = mutual_information(copy(s), subsystem_rangeA, subsystem_rangeB, Val(:rref))
mi_graph = mutual_information(copy(s), subsystem_rangeA, subsystem_rangeB, Val(:graph))
@test mi_rref == mi_graph # We can still test consistency between rref and graph
@test mi_rref ≥ 0
end
end
end
@testset "Explicit non-contiguous ranges with :clip" begin
n = 6
s = random_stabilizer(n)
subsystem_rangeA = 1:2
subsystem_rangeB = 5:6
@test_throws ArgumentError mutual_information(copy(s), subsystem_rangeA, subsystem_rangeB, Val(:clip))
end
end
end
Loading