Update for latest Observers.jl syntax

Project.toml

@@ -5,6 +5,7 @@ version = "0.0.25"
 
 [deps]
 ChainRulesCore = "d360d2e6-b24c-11e9-a2a3-2a2ae2dbcce4"
+DataFrames = "a93c6f00-e57d-5684-b7b6-d8193f3e46c0"
 HDF5 = "f67ccb44-e63f-5c2f-98bd-6dc0ccc4ba2f"
 ITensors = "9136182c-28ba-11e9-034c-db9fb085ebd5"
 JLD2 = "033835bb-8acc-5ee8-8aae-3f567f8a3819"
@@ -18,6 +19,7 @@ StatsBase = "2913bbd2-ae8a-5f71-8c99-4fb6c76f3a91"
 
 [compat]
 ChainRulesCore = "1.10"
+DataFrames = "1.6"
 HDF5 = "0.13.1, 0.14, 0.15, 0.16"
 ITensors = "0.3.20"
 JLD2 = "0.4.14"

examples/04_circuitobserver.jl

@@ -28,7 +28,7 @@ end
 σz(ψ::MPS) = [measure_pauli(ψ, j, "Z") for j in 1:length(ψ)]
 
 # define the Circuit observer
-obs = Observer([
+obs = observer([
   "χs" => linkdims,      # bond dimension at each bond
   "χmax" => maxlinkdim,  # maximum bond dimension
   "σˣ(2)" => σx2,        # pauli X on site 2

examples/07_qst_circuit.jl

@@ -35,7 +35,7 @@ opt = Optimisers.Descent(0.01)
 
 # Initialize the observer for the fidelity
 F(ψ::MPS; kwargs...) = fidelity(ψ, Ψ)
-obs = Observer(["F" => F])
+obs = observer(["F" => F])
 
 # Run quantum state tomography, where a variational MPS `|ψ(θ)⟩`
 # is optimized to mimimize the cross entropy between the data and 
@@ -84,7 +84,7 @@ opt = Optimisers.ADAM()
 # Initialize the observer
 F(ρ::LPDO; kwargs...) = fidelity(ρ, ϱ)
 
-obs = Observer(["F" => F])
+obs = observer(["F" => F])
 
 # Run quantum state tomography, where a variational LPDO `ρ(θ)`
 # is optimized to mimimize the cross entropy between the data and 

examples/08_qpt_circuit.jl

@@ -36,7 +36,7 @@ U0 = randomprocess(Û; χ=χ)
 opt = Optimisers.Descent(0.01)
 
 F(U::MPO; kwargs...) = fidelity(U, Û; process=true)
-obs = Observer(["F" => F])
+obs = observer(["F" => F])
 
 # Initialize stochastic gradient descent optimizer
 @show maxlinkdim(U0)
@@ -86,7 +86,7 @@ N = length(Φ)
 opt = Optimisers.ADAM()
 
 F(Λ::LPDO; kwargs...) = fidelity(Λ, Φ; process=true)
-obs = Observer(["F" => F])
+obs = observer(["F" => F])
 
 # Run process tomography
 println("Run process tomography to learn noisy process Λ")

examples/09_qst_ising.jl

@@ -59,8 +59,8 @@ F(ψ::MPS) = fidelity(ψ, Ψ)
 ZZ(ψ::MPS) = correlation_matrix(Ψ, "Z", "Z")
 
 # Initialize observer
-obs = Observer(["fidelity" => F, "energy" => Energy, "correlations" => ZZ])
-#obs = Observer(["energy" => Energy])
+obs = observer(["fidelity" => F, "energy" => Energy, "correlations" => ZZ])
+#obs = observer(["energy" => Energy])
 
 @printf("⟨Ψ|Ĥ|Ψ⟩ =  %.5f   ", E)
 # Run tomography

examples/Quantum gates via tunable couplers.ipynb

@@ -39221,7 +39221,7 @@
     "# set initial state |ψ⟩ = |0,1⟩\n",
     "ψ₀ = productstate(hilbert, [0,0,1,0,1])\n",
     "\n",
-    "obs = Observer(observables)\n",
+    "obs = observer(observables)\n",
     "\n",
     "# perform TEBD simulation and generate output `MPS`\n",
     "ψ = runcircuit(ψ₀, circuit; \n",
@@ -41782,7 +41782,7 @@
     "# set initial state |ψ⟩ = |0,1⟩\n",
     "ψ₀ = productstate(hilbert, [0,0,1,0,1])\n",
     "\n",
-    "obs = Observer(observables)\n",
+    "obs = observer(observables)\n",
     "\n",
     "# perform TEBD simulation and generate output `MPS`\n",
     "ψ = runcircuit(ψ₀, circuit; \n",

examples/optimal-coherent-control.ipynb

@@ -167,7 +167,7 @@
     "#define a vector of observables and create the `Observer`.\n",
     "observables = [\"n($α)\" => x -> population(x, k)  # actually x -> expect(x, \"a† * a\"; sites = k)\n",
     "               for (k,α) in enumerate(modes)]\n",
-    "obs = Observer(observables)"
+    "obs = observer(observables)"
    ],
    "metadata": {
     "name": "A slide ",
@@ -1628,7 +1628,7 @@
     "\n",
     "H = hamiltonian(ω⃗, g)\n",
     "\n",
-    "obs = Observer(observables)\n",
+    "obs = observer(observables)\n",
     "\n",
     "circuit = trottercircuit(H; ts = ts, layered = true)\n",
     "\n",
@@ -3323,7 +3323,7 @@
     "ψ₀ = productstate(hilbert, [1,0])\n",
     "observables = [\"n($α)\" => x -> population(x, k)\n",
     "               for (k,α) in enumerate(modes)]\n",
-    "obs = Observer(observables)\n",
+    "obs = observer(observables)\n",
     "ψ = runcircuit(ψ₀, circuit; (observer!) = obs,\n",
     "               move_sites_back_before_measurements = true, outputlevel = 0)\n",
     "res = DataFrame(results(obs));\n",

examples/optimal-coherent-control.jl

@@ -45,7 +45,7 @@ end;
 #define a vector of observables and create the `Observer`.
 observables = ["n($α)" => x -> population(x, k)  # actually x -> expect(x, "a† * a"; sites = k)
                for (k, α) in enumerate(modes)]
-obs = Observer(observables)
+obs = observer(observables)
 
 tg = 30                  # final time (in ns)
 trottersteps = 100       # number of Trotter steps
@@ -75,7 +75,7 @@ p
 
 H = hamiltonian(ω⃗, g)
 
-obs = Observer(observables)
+obs = observer(observables)
 
 circuit = trottercircuit(H; ts=ts, layered=true)
 
@@ -167,7 +167,7 @@ Ht = [hamiltonian(θ⃗, t) for t in ts]
 circuit = trottercircuit(Ht; ts=ts, layered=true)
 ψ₀ = productstate(hilbert, [1, 0])
 observables = ["n($α)" => x -> population(x, k) for (k, α) in enumerate(modes)]
-obs = Observer(observables)
+obs = observer(observables)
 ψ = runcircuit(
   ψ₀, circuit; (observer!)=obs, move_sites_back_before_measurements=true, outputlevel=0
 )

examples/src/optimal-coherent-control.jl

@@ -91,7 +91,7 @@ end;
 #define a vector of observables and create the `Observer`.
 observables = ["n($α)" => x -> population(x, k)  # actually x -> expect(x, "a† * a"; sites = k)
                for (k, α) in enumerate(modes)]
-obs = Observer(observables)
+obs = observer(observables)
 
 #nb # %% A slide [markdown] {"slideshow": {"slide_type": "subslide"}}
 # We are not ready to simulate the system dynamics using a Trotter expansion.
@@ -152,7 +152,7 @@ p
 
 H = hamiltonian(ω⃗, g)
 
-obs = Observer(observables)
+obs = observer(observables)
 
 circuit = trottercircuit(H; ts=ts, layered=true)
 
@@ -280,7 +280,7 @@ Ht = [hamiltonian(θ⃗, t) for t in ts]
 circuit = trottercircuit(Ht; ts=ts, layered=true)
 ψ₀ = productstate(hilbert, [1, 0])
 observables = ["n($α)" => x -> population(x, k) for (k, α) in enumerate(modes)]
-obs = Observer(observables)
+obs = observer(observables)
 ψ = runcircuit(
   ψ₀, circuit; (observer!)=obs, move_sites_back_before_measurements=true, outputlevel=0
 )

src/PastaQ.jl

@@ -1,14 +1,15 @@
 module PastaQ
 
-using ITensors
-using Random
-using LinearAlgebra
+using DataFrames
 using HDF5
+using ITensors
 using JLD2
-using Printf
+using LinearAlgebra
 using Observers
-using StatsBase: StatsBase, Weights
 using Optimisers: Optimisers
+using Printf
+using Random
+using StatsBase: StatsBase, Weights
 
 include("imports.jl")
 include("exports.jl")

src/io.jl

@@ -164,7 +164,7 @@ function printmetric(name::String, metric::Complex)
   end
 end
 
-function printobserver(observer::Observer, print_metrics::Union{String,AbstractArray})
+function printobserver(observer::DataFrame, print_metrics::Union{String,AbstractArray})
   if !isempty(print_metrics)
     if print_metrics isa String
       printmetric(print_metrics, results(observer, print_metrics)[end])

src/tomography/quantumtomography.jl

@@ -158,7 +158,7 @@ function tomography(train_data::AbstractMatrix, L::LPDO; (observer!)=nothing, kw
   localnorm = isqpt ? 2.0 : 1.0
 
   # observer is not passed but earlystop is called
-  observer! = (isnothing(observer!) || earlystop) ? Observer() : observer!
+  observer! = (isnothing(observer!) || earlystop) ? observer() : observer!
 
   # observer is defined
   if !isnothing(observer!)

test/test_io.jl

@@ -113,7 +113,7 @@ end
   ψ = runcircuit(sites, circuit)
   Ftest = fidelity(ψ, ϕ)
   f(ψ::MPS) = fidelity(ψ, ϕ)#; kwargs...) = fidelity(ψ, ϕ)
-  obs = Observer(["f" => f])
+  obs = observer(["f" => f])
   ψ = runcircuit(
     sites,
     circuit;
@@ -150,7 +150,7 @@ end
     ρ = runcircuit(sites, circuit; noise=("DEP", (p=0.001,)))
     Ftest = fidelity(ϱ, ρ)
     g(ρ::MPO; kwargs...) = fidelity(ρ, ϱ)#; kwargs...) = fidelity(ψ, ϕ)
-    obs = Observer(["g" => g])
+    obs = observer(["g" => g])
     outputpath = "simulation"
     ρ₀ = projector(productstate(sites))
     ρ = runcircuit(
@@ -194,7 +194,7 @@ end
   opt = Optimisers.Descent(0.01)
 
   F(ψ::MPS; kwargs...) = fidelity(ψ, Ψ)
-  obs = Observer(["F" => F])
+  obs = observer(["F" => F])
   epochs = 18
 
   batchsize = 10
@@ -243,7 +243,7 @@ end
   opt = Optimisers.Descent(0.01)
 
   F(ρ::LPDO; kwargs...) = fidelity(ρ, ϱ)
-  obs = Observer(["F" => F])
+  obs = observer(["F" => F])
   epochs = 9
 
   batchsize = 10
@@ -292,7 +292,7 @@ end
   opt = Optimisers.Descent(0.01)
 
   F(U::MPO; kwargs...) = fidelity(U, V; process=true)
-  obs = Observer(["F" => F])
+  obs = observer(["F" => F])
   epochs = 9
 
   batchsize = 10
@@ -342,7 +342,7 @@ end
   opt = Optimisers.Descent(0.01)
 
   F(Λ::LPDO; kwargs...) = fidelity(Λ, Φ)
-  obs = Observer(["F" => F])
+  obs = observer(["F" => F])
   epochs = 9
 
   batchsize = 10