update

Author: mofeing

perf/vqe/Project.toml

@@ -7,6 +7,7 @@ EinExprs = "b1794770-133b-4de1-afb4-526377e9f4c5"
 Enzyme = "7da242da-08ed-463a-9acd-ee780be4f1d9"
 LinearAlgebra = "37e2e46d-f89d-539d-b4ee-838fcccc9c8e"
 Metis = "2679e427-3c69-5b7f-982b-ece356f1e94b"
+Muscle = "21fe5c4b-a943-414d-bf3e-516f24900631"
 Networks = "634dcf3f-6aa7-47ba-9e85-1fb329c9ebfe"
 PrettyTables = "08abe8d2-0d0c-5749-adfa-8a2ac140af0d"
 Reactant = "3c362404-f566-11ee-1572-e11a4b42c853"
@@ -19,6 +20,7 @@ YaoBlocks = "418bc28f-b43b-5e0b-a6e7-61bbc1a2c1df"
 [compat]
 EinExprs = "0.6.10"
 Enzyme = "0.13.51"
+Muscle = "0.3.14"
 Reactant = "0.2.138"
-Tenet = "0.9.1"
+Tenet = "0.10.0"
 YaoBlocks = "0.13.15"

perf/vqe/README.md

@@ -3,7 +3,7 @@
 A simple VQE training through exact tensor network contraction and backpropagation.
 Because it uses exact tensor network contraction, it cannot scale to large number of layers but it should be able to scale to mid-range number of qubits (around 50) and shallow circuits.
 
-It uses a [Efficient SU(2)](https://) circuit ansatz for the VQE.
+It uses a "Efficient SU(2)" circuit ansatz for the VQE.
 
 ## Setup
 

perf/vqe/main.jl

@@ -107,7 +107,7 @@ results = Vector{Tuple{String,String,T,T,Float64}}()
 
 # primal
 ## only XLA
-f_xla = @compile compile_options = Reactant.DefaultXLACompileOptions(; sync=true) expectation(
+f_xla = @time @compile compile_options = Reactant.DefaultXLACompileOptions(; sync=true) expectation(
     params_re, observable_re, coef_re
 )
 b = @benchmark $f_xla($params_re, $observable_re, $coef_re) setup = (GC.gc(true))