[BUG] Problem with rydberg_h function and juliacall

[vrige]

Hi, I am new to Julia and Bloqade, so I may be wrong, but I think I have found an issue in using the function rydberg_h using juliacall from a python program. I have tested the same code in julia (without using Juliacall) and this problem doesn't arise.
More precisely, the function jl.Bloqade.rydberg_h(atoms, Ω=Ω, Δ=Δ, ϕ =ϕ) doesn't work in my python code, while jl.Bloqade.rydberg_h(atoms, Ω=Ω, Δ=Δ) works perfectly.

from juliacall import Main as jl, convert as jlconvert
from juliacall import Pkg as jlPkg
import numpy as np
import matplotlib.pyplot as plt
from time import time
from math import pi as pi

jlPkg.activate(".")
jlPkg.status()

jl.seval("using Graphs")
#jl.seval("using BloqadeExpr")
jl.seval("using Bloqade")
jl.seval("using Random")
jl.seval("using GenericTensorNetworks")
jl.seval("using Optim")
jl.seval("using PythonCall")
jl.seval("using Plots")

jl.Random.seed_b(3)
atoms = jl.Bloqade.generate_sites(jl.Bloqade.SquareLattice(), 3, 4, scale=4.5)
atoms = jl.Bloqade.random_dropout(atoms, 0.3)

jl.Bloqade.plot(atoms, blockade_radius=7.5)

graph = jl.Bloqade.BloqadeMIS.unit_disk_graph(atoms, 7.5)
mis_size_and_counting = jl.getindex(jl.GenericTensorNetworks.solve(jl.GenericTensorNetworks.IndependentSet(graph), jl.GenericTensorNetworks.CountingMax()))
print(jl.getindex(mis_size_and_counting))

total_time = 3.0
Ω_max = 2 * pi * 4
ϕ_max = 47.0 / (2 * pi * 1e6)
time_points = np.array([0.0, 0.1, 2.1, 2.2, total_time])
amplitude_value = np.array([0.0, Ω_max, Ω_max, 0, 0])
phase_values = np.array([0.0, ϕ_max, ϕ_max, 0, 0])
detuning_values = np.array([0.0, Ω_max, Ω_max, 0, 0])

time_points_jl = jlconvert(jl.Array, time_points)
amplitude_value_jl = jlconvert(jl.Array, amplitude_value)
phase_values_jl = jlconvert(jl.Array, phase_values)
detuning_values_jl = jlconvert(jl.Array, detuning_values)

Omega = jl.Bloqade.piecewise_linear(clocks=time_points_jl, values=amplitude_value_jl)
Delta = jl.Bloqade.piecewise_linear(clocks=time_points_jl, values=phase_values_jl)
Phi = jl.Bloqade.piecewise_linear(clocks=time_points_jl, values=detuning_values_jl)

T_max = time_points[-1]

number_of_shots = 100

start = time()
hamiltonian = jl.Bloqade.rydberg_h(atoms,  Ω=Omega, Δ=Delta, ϕ=Phi)
reg = jl.Bloqade.zero_state(jl.Bloqade.nqubits(hamiltonian))
prob = jl.Bloqade.SchrodingerProblem(reg, T_max, hamiltonian)
jl.Bloqade.emulate_b(prob)
jl.println(jl.Bloqade.measure(reg, nshots=number_of_shots))
end = time()

print("time: ", end - start)

This is the error that I receive:

Traceback (most recent call last):
  File "/home/path/PycharmProjects/aquila_tesi/testing_julia.py", line 56, in <module>
    hamiltonian = jl.Bloqade.rydberg_h(atoms, Ω=Ω, Δ=Δ, ϕ=ϕ)
  File "/home/path/.julia/packages/PythonCall/1f5yE/src/jlwrap/any.jl", line 208, in __call__
    return self._jl_callmethod($(pyjl_methodnum(pyjlany_call)), args, kwargs)
juliacall.JuliaError: MethodError: no method matching rydberg_h(::AtomList{2, Float64}; φ::Waveform{BloqadeWaveforms.PiecewiseLinear{Float64, Interpolations.Extrapolation{Float64, 1, Interpolations.GriddedInterpolation{Float64, 1, Float64, Interpolations.Gridded{Interpolations.Linear{Interpolations.Throw{Interpolations.OnGrid}}}, Tuple{Vector{Float64}}}, Interpolations.Gridded{Interpolations.Linear{Interpolations.Throw{Interpolations.OnGrid}}}, Interpolations.Throw{Nothing}}}, Float64}, Ω::Waveform{BloqadeWaveforms.PiecewiseLinear{Float64, Interpolations.Extrapolation{Float64, 1, Interpolations.GriddedInterpolation{Float64, 1, Float64, Interpolations.Gridded{Interpolations.Linear{Interpolations.Throw{Interpolations.OnGrid}}}, Tuple{Vector{Float64}}}, Interpolations.Gridded{Interpolations.Linear{Interpolations.Throw{Interpolations.OnGrid}}}, Interpolations.Throw{Nothing}}}, Float64}, Δ::Waveform{BloqadeWaveforms.PiecewiseLinear{Float64, Interpolations.Extrapolation{Float64, 1, Interpolations.GriddedInterpolation{Float64, 1, Float64, Interpolations.Gridded{Interpolations.Linear{Interpolations.Throw{Interpolations.OnGrid}}}, Tuple{Vector{Float64}}}, Interpolations.Gridded{Interpolations.Linear{Interpolations.Throw{Interpolations.OnGrid}}}, Interpolations.Throw{Nothing}}}, Float64})

Closest candidates are:
  rydberg_h(::Any; C, Ω, ϕ, Δ) got unsupported keyword argument "φ"
   @ BloqadeExpr ~/.julia/packages/BloqadeExpr/1c235/src/interface.jl:77
  rydberg_h(::Any, !Matched::Any, !Matched::Any, !Matched::Any, !Matched::Any) got unsupported keyword arguments "φ", "Ω", "Δ"
   @ BloqadeExpr ~/.julia/packages/BloqadeExpr/1c235/src/interface.jl:81

Stacktrace:
 [1] kwerr(::NamedTuple{(:φ, :Ω, :Δ), Tuple{Waveform{BloqadeWaveforms.PiecewiseLinear{Float64, Interpolations.Extrapolation{Float64, 1, Interpolations.GriddedInterpolation{Float64, 1, Float64, Interpolations.Gridded{Interpolations.Linear{Interpolations.Throw{Interpolations.OnGrid}}}, Tuple{Vector{Float64}}}, Interpolations.Gridded{Interpolations.Linear{Interpolations.Throw{Interpolations.OnGrid}}}, Interpolations.Throw{Nothing}}}, Float64}, Waveform{BloqadeWaveforms.PiecewiseLinear{Float64, Interpolations.Extrapolation{Float64, 1, Interpolations.GriddedInterpolation{Float64, 1, Float64, Interpolations.Gridded{Interpolations.Linear{Interpolations.Throw{Interpolations.OnGrid}}}, Tuple{Vector{Float64}}}, Interpolations.Gridded{Interpolations.Linear{Interpolations.Throw{Interpolations.OnGrid}}}, Interpolations.Throw{Nothing}}}, Float64}, Waveform{BloqadeWaveforms.PiecewiseLinear{Float64, Interpolations.Extrapolation{Float64, 1, Interpolations.GriddedInterpolation{Float64, 1, Float64, Interpolations.Gridded{Interpolations.Linear{Interpolations.Throw{Interpolations.OnGrid}}}, Tuple{Vector{Float64}}}, Interpolations.Gridded{Interpolations.Linear{Interpolations.Throw{Interpolations.OnGrid}}}, Interpolations.Throw{Nothing}}}, Float64}}}, ::Function, ::AtomList{2, Float64})
   @ Base ./error.jl:165
 [2] pyjlany_call(self::typeof(rydberg_h), args_::Py, kwargs_::Py)
   @ PythonCall ~/.julia/packages/PythonCall/1f5yE/src/jlwrap/any.jl:34
 [3] _pyjl_callmethod(f::Any, self_::Ptr{PythonCall.C.PyObject}, args_::Ptr{PythonCall.C.PyObject}, nargs::Int64)
   @ PythonCall ~/.julia/packages/PythonCall/1f5yE/src/jlwrap/base.jl:69
 [4] _pyjl_callmethod(o::Ptr{PythonCall.C.PyObject}, args::Ptr{PythonCall.C.PyObject})
   @ PythonCall.C ~/.julia/packages/PythonCall/1f5yE/src/cpython/jlwrap.jl:47

Process finished with exit code 1

I feel like the function mistakes the symbol ϕ for the symbol φ, so it is not able to find the right function to call because the parameters are different.

I am using pycharm with an Anaconda environment.
For the julia dependecies i am suing a .toml file with the following package versions:

Plots = "1.29"
Optim = "1.7.6"
GenericTensorNetworks = "1.3.4"
Graphs = "1.7"
CUDA = "4"
Adapt = "3.6.2"
Bloqade = "0.1.23"
BloqadeExpr = "0.1"
BloqadeKrylov = "0.1"
BloqadeLattices = "0.1"
BloqadeMIS = "0.1"
BloqadeODE = "0.1"
BloqadeWaveforms = "0.1"
ColorSchemes = "3"
Colors = "0.12"
ForwardDiff = "0.10"
Measurements = "2"
PythonCall = "0.8,0.9"
Reexport = "1"
Yao = "0.8"
YaoSubspaceArrayReg = "0.1"
julia = "1.6.3"

While on pkg.status:

Status `~/.julia/environments/v1.9/Project.toml`
  [79e6a3ab] Adapt v3.6.2
  [bd27d05e] Bloqade v0.1.23
  [bd27d05e] BloqadeExpr v0.1.13
  [34da2185] Compat v4.6.1
⌃ [3521c873] GenericTensorNetworks v1.3.0
  [86223c79] Graphs v1.8.0
  [429524aa] Optim v1.7.6
  [91a5bcdd] Plots v1.38.16
  [6099a3de] PythonCall v0.9.13

[johnzl-777]

Hi @vrige ,

Thanks for bringing this to our attention.

Bloqade wasn't really designed with juliacall usage in mind (or tested thoroughly against it). That being said the bug here seems to be an odd behavior where juliacall, for a reason that isn't apparent to me, converts the U+03D5 phi used in Bloqade (and the one you typed) to a U+03C6 phi which are completely different characters, resulting in the error you see.

I'm going to open an issue on the github repo for juliacall to try to get to the bottom of this but in the meantime I also noticed some parts of your code I'm not sure are intentional or not:

• you define ϕ_max = 47.0 / (2 * pi * 1e6) but by default Bloqade automatically assumes units of radians or radians per microsecond. I’m assuming you were shooting for 47 * 2pi ?
• In Δ = jl.Bloqade.piecewise_linear(clocks=time_points_jl, values=phase_values_jl) and ϕ = jl.Bloqade.piecewise_linear(clocks=time_points_jl, values=detuning_values_jl) you seem to have your values flipped here (you assign phase values to detuning and detuning values to your phase

[vrige]

Thanks for the hard work and for opening the issue on juliacall @johnzl-777.
And yes you are right about my mistakes, thanks again for pointing that out!

[johnzl-777]

No problem! I hope to hear back from the juliacall folks soon, curious as to what could be causing this problem.

Feel free to subscribe to the linked issue as well! You can get notified as soon as they have a response.

[johnzl-777]

Hey @vrige ,

It looks like there's a solution to this issue (kudos to @cjdoris for the help!) but it'll require a minor change of how you call rydberg_h. Instead of what you're currently doing you'll need to replace:

hamiltonian = jl.Bloqade.rydberg_h(atoms,  Ω=Omega, Δ=Delta, ϕ=Phi)

with

hamiltonian = jl.Bloqade.rydberg_h(atoms, **{'Ω': Omega, 'Δ': Delta, 'ϕ': Phi})

Note that you can still use unicode to denote the Rabi, Detuning, and Phase waveform variables in Python, you'd just have to call it like:

hamiltonian = jl.Bloqade.rydberg_h(atoms, **{'Ω': Ω, 'Δ': Δ, 'ϕ': ϕ})

(Although I do realize it may be slightly less friendly to type unicode characters in certain python environments over Julia!)

I went ahead and ran your code on my end with these changes and it runs just fine (: If you're still having issues here please feel free to provide more details, otherwise you can close this issue.