Minor documentation fixes around phases and a workaround for Makie …

…plotting regression (#163)

CHANGELOG.md

@@ -5,9 +5,10 @@
 
 # News
 
-## v0.8.15 - dev
+## v0.8.15 - 2023-08-16
 
 - Initial support for GPU accelerated circuit simulation (with CUDA).
+- Minor documentation fixes around `phases` and a workaround for Makie plotting regression.
 
 ## v0.8.14 - 2023-07-19
 

Project.toml

@@ -1,7 +1,7 @@
 name = "QuantumClifford"
 uuid = "0525e862-1e90-11e9-3e4d-1b39d7109de1"
 authors = ["Stefan Krastanov <stefan@krastanov.org>"]
-version = "0.8.14"
+version = "0.8.15"
 
 [deps]
 Combinatorics = "861a8166-3701-5b0c-9a16-15d98fcdc6aa"
@@ -43,7 +43,7 @@ Graphs = "1.4.1"
 HostCPUFeatures = "0.1.6"
 ILog2 = "0.2.3"
 MacroTools = "0.5.9"
-Makie = "0.19"
+Makie = "0.19.7"
 Nemo = "0.34, 0.35"
 Plots = "1.38.0"
 PrecompileTools = "1"

docs/src/stab-algebra-manual.md

@@ -312,14 +312,20 @@ operator**. If you want to perform a Monte Carlo simulation, you need to
 randomize the phase of the stabilizer at the anticommuting index yourself. For
 instance, one can do:
 
-```julia
-newstate, anticomindex, result = project!(state, projector)
-if isnothing(result)
- newstate.phases[anticomindex] = rand([0x0,0x2])
-end
+```jldoctest proj
+julia> newstate, anticomindex, result = project!(copy(s), P"XII")
+ if isnothing(result)
+ phases(newstate)[anticomindex] = rand([0x0,0x2])
+ end
+ result, anticomindex
+(nothing, 2)
 ```
 
-Or we can project on a commuting operator, hence no anticommuting terms (the
+Of course, this is a rather cumbersome way to run a simulation, so we also provide
+[`projectrand!`](@ref) which does the necessary randomization automatically,
+for cases where you do not need the fine grained control of `project!`.
+
+We can project on a commuting operator, hence no anticommuting terms (the
 index is zero), and the result is perfectly determined (-1, or in our convention
 to represent the phase, 0x2).
 
@@ -352,6 +358,7 @@ julia> project!(copy(s), P"ZZI", phases=false)
 ## Sparse single-qubit measurements
 
 In many circumstances only a single-qubit operator is being measured. In that case one should use the [`projectX!`](@ref), [`projectY!`](@ref), and [`projectZ!`](@ref) functions as they are much faster thanks to tracking only a single qubit.
+They have versions that randomize the phase as necessary as well: [`projectXrand!`](@ref), [`projectYrand!`](@ref), and [`projectZrand!`](@ref).
 
 ## Gate-like interface
 

ext/QuantumCliffordMakieExt/QuantumCliffordMakieExt.jl

@@ -36,10 +36,13 @@ function Makie.plot!(myplot::StabilizerPlot)
  throw(ErrorException("`xzcomponents` should be `:split` or `:together`"))
  end
  r = r[end:-1:1,:]'
- Makie.heatmap!(myplot, r;
+ hm = Makie.heatmap!(myplot, r;
  colorrange = (0, 3),
  colormap=myplot.colormap
  )
+ for k in [:colorscale, :highclip, :lowclip]
+ myplot[k] = hm[k]
+ end
  myplot
 end
 

src/QuantumClifford.jl

@@ -601,9 +601,9 @@ Base.copy(d::MixedDestabilizer) = MixedDestabilizer(copy(d.tab),d.rank)
 @inline nqubits(s::AbstractStabilizer) = nqubits(tab(s))
 @inline nqubits(t::Tableau) = t.nqubits
 
-"""The phases of a given tableau."""
+"""The phases of a given tableau. It is a view, i.e. if you modify this array, the original tableau caries these changes."""
 @inline phases(t::Tableau) = t.phases
-@inline phases(s::Stabilizer) = phases(tab(s))
+@inline phases(s::AbstractStabilizer) = phases(tab(stabilizerview(s)))
 
 ##############################
 # Pauli Operator Helpers