Which version of Julia can run the following codes

New to Julia
1

Hello, I am new to Julia and for research reasons I have to run a simple code for my friends. The code need Devito and JUDI:Installation · slimgroup/JUDI.jl Wiki · GitHub
Here is the source code (Julia was installed on Ubuntu 16.04.7 LTS):

#using ClusterManagers; addprocs_slurm(1)
#using Distributed; addprocs(64)
using JUDI.TimeModeling, HDF5

n, d, o, m = read(h5open("./Brain_3D_true_model.h5", "r"), "n", "d", "o", "m")

# @everywhere using PyPlot
# figure(); imshow(sqrt.(1 ./ m[:, 200, :]))

# Set up model structure
model = Model((n[1], n[2], n[3]), (d[1], d[2], d[3]), (o[1], o[2], o[3]), m)

tn = 0.2f0#200f0 # micro s
dt = 0.1f0
f0 = .25  # MHz

# set up source
nsrc, src_loc = read(h5open("./Sources_location_512.h5"), "nsrc", "src_loc")
nsrc = Integer(nsrc[1])
xsrc = convertToCell(src_loc[1:2, 1] .* d[1])
ysrc = convertToCell(src_loc[1:2, 2] .* d[2])
zsrc = convertToCell(src_loc[1:2, 3] .* d[3])
nsrc = 2

# set up receivers
xrec = Float32.(xsrc)
yrec = Float32.(ysrc)
zrec = Float32.(zsrc)



src_geometry = Geometry(xsrc, ysrc, zsrc; dt = dt, t = tn)
rec_geometry = Geometry(xrec, yrec, zrec; dt = dt, t = tn, nsrc = nsrc)

# Set up source
wavelet = ricker_wavelet(tn, dt, f0)

# Info for modeling operators
ntComp = get_computational_nt(src_geometry, rec_geometry, model)
info = Info(prod(model.n), nsrc, ntComp)

# Set up modeling options
opt = Options(save_data_to_disk = true, file_path = "./records", file_name = "brain_3D_shot")

# Set up operators
Pr = judiProjection(info, rec_geometry)
Ps = judiProjection(info, src_geometry)
A_inv = judiModeling(info, model; options=opt)
q = judiVector(src_geometry, wavelet)

# Generate data and save as individual SEG-Y files to disk
d_obs = Pr * A_inv * adjoint(Ps) * q

I have tried to install Devito4.3, Julia v0.6.4, v0.7.0, v1.4.2, v1.5.0, v1.5.4, v1.6.1 and installed JUDI subsequently, but it seems none of the Julia version above can run the code.
The error output is:
ERROR: LoadError: MethodError: no method matching Float32(::Vector{Float64})
Closest candidates are:
(::Type{T})(!Matched::AbstractChar) where T<:Union{AbstractChar, Number} at char.jl:50
(::Type{T})(!Matched::Base.TwicePrecision) where T<:Number at twiceprecision.jl:243
(::Type{T})(!Matched::Complex) where T<:Real at complex.jl:37
…
Stacktrace:
[1] _broadcast_getindex_evalf
@ ./broadcast.jl:648 [inlined]
[2] _broadcast_getindex
@ ./broadcast.jl:621 [inlined]
[3] getindex
@ ./broadcast.jl:575 [inlined]
[4] copy
@ ./broadcast.jl:922 [inlined]
[5] materialize(bc::Base.Broadcast.Broadcasted{Base.Broadcast.DefaultArrayStyle{1}, Nothing, Type{Float32}, Tuple{Vector{Vector{Float64}}}})
@ Base.Broadcast ./broadcast.jl:883
[6] top-level scope
@ /media/hp/46E0111EE0111631/jktong/brain_3D/Brain_3D_forward_new.jl:26
if I change line 26-line 29 into
xrec = xsrc
yrec = ysrc
zrec = zsrc
it will prompt: MethodError: no method matching Geometry
PS: my python version is Python 3.8.9.

Thanks a lot!

2

Please use three backtics (```) to proceed and follow your code when pasted into the thread.
It will format the code correctly, making it much easier to help.

I suspect that the solution is to replace Float32(::Vector{Float64}) with Float32.(::Vector{Float64}),
which would make sense if for instance xsrc was a vector of vectors.

3

Thanks for your help! As can be seen from line 26 of the code，we have already used Float32. to convert the variables:

xrec = Float32.(xsrc)
yrec = Float32.(ysrc)
zrec = Float32.(zsrc)

But the it did not solve the problem

4

but you should be getting different errors now right? can you post the error again?

5

No，the error messages are the same…
We used Float32. in the beginning，I did not add ``` before the code at first so the code didn’t format correctly.

6

Can you give an output of typeof(xsrc)?

7

Hi,

You might try visiting the GitHub slimgroup repository (https://github.com/slimgroup/JUDI.jl) and read the “Running with Docker”
information. It’s possible to pull the stated docker image and initiate a jupyter notebook to execute your code. For example: docker run -p 8888:8888 mloubout/judi:1.7-latest should execute and list a URL for an interactive jupyter notebook session. You could paste your code in a cell for testing and debugging.

You could also try executing additional demo scripts from the slimgroup and others for learning and reference.

Without access to the Brain model and source geometry files, it would otherwise be difficult to debug or advise.

Good luck - hope it helps.