Stochastic PiCLES pull request by tomprotin · Pull Request #42 · mochell/PiCLES.jl

tomprotin · 2025-10-07T15:09:10Z

This works in PiCLES 1.10.0. You can try to run tests/test_case_1.jl (but I advise decreasing the number of particles in this test case, at line 114, you can set it to 150 or 1500 to see if it works).
I don't know how your test cases work, but I changed the behavior of some critical functions like :

run()
advance() (when the actual stepping is computed for an individual particle, at lines 173 -175 in mapping_2D.jl)
time_step!()

…comming...

…re left to do

…g types and threshold structure

…each node when using the "nonparametric" spreading type

…uctures for the validation of the 1/r decay of energy density

… versions

mochell

move SIR_analysis maybe into analysis/SIR_analysis
remove the listed files in .gignore from the branch and then remove them from .gitignore
adjust Abstract model defition and the timestepping methods (as discussed)
time_step!(model::Abstract2DModel, ... ):
- Look at the version of the function in this branch (my version), I refactors it with time_step!_advance, you could do the same here.
core_2D_spread.jl It maybe the easierst to rename it back to core_2D.jl to see the differences better. I have not looked at this file in more detail
Mapping_2d.jl the ParticleToNode! routines changed. I deleted the definitions that are not used any more.. I hope it still works..
Do you actually use particle_waves_v6.jl ? I think you can just use v5 and and then we can get rid of v6

mochell · 2025-11-05T18:16:31Z

src/Models/GeometricalOpticsModels.jl

+                        nPart,
+                        wnds,
+                        cur,
+                        Mstat} <: Abstract2DModel where {Mstat<:Union{Nothing,stat}, PCollection<:Union{Vector,Array}}


Change Abstract2DModel to something like Abstract2DStochasticModel. Define the model Type in Architectures.jl, and then make a timestepper for that model type

Suggested change

Mstat} <: Abstract2DModel where {Mstat<:Union{Nothing,stat}, PCollection<:Union{Vector,Array}}

Mstat} <: Abstract2DStochasticModel where {Mstat<:Union{Nothing,stat}, PCollection<:Union{Vector,Array}}

src/Models/GeometricalOpticsModels.jl

mochell · 2025-11-05T18:25:45Z

src/Operators/TimeSteppers.jl

+function write_particles_to_csv(wave_model, Δt)
+    sec=string(Int64(floor((wave_model.clock.time+Δt)/60)))
+    dec=string(Int64(floor(10*((wave_model.clock.time+Δt)/60-floor((wave_model.clock.time+Δt)/60)))))
+    save_path = wave_model.plot_savepath
+
+    nParticles = wave_model.n_particles_launch
+    @info wave_model.clock.time/60
+
+    parts = wave_model.ParticleCollection[(end-nParticles+1):end]
+
+    logE = zeros(nParticles)
+    cx = zeros(nParticles)
+    cy = zeros(nParticles)
+    x = zeros(nParticles)
+    y = zeros(nParticles)
+    for i in 1:nParticles
+            logE[i] = parts[i].ODEIntegrator[1]
+            cx[i] = parts[i].ODEIntegrator[2]
+            cy[i] = parts[i].ODEIntegrator[3]
+            x[i] = parts[i].ODEIntegrator[4]
+            y[i] = parts[i].ODEIntegrator[5]
+    end
+
+    data = DataFrame(id=1:nParticles, logE = logE, cx = cx, cy = cy, x = x, y = y)
+    data2 = Tables.table(transpose(wave_model.State[:, :, 1]))
+    CSV.write(save_path*"/data/particles_"*sec*","*dec*".csv", data)
+    CSV.write(save_path*"/data/mesh_values_"*sec*","*dec*".csv", data2)
+end


this should move to Utils/InputOutput.jl

mochell · 2025-11-05T18:36:26Z

src/Operators/TimeSteppers.jl

 """
 time_step!(model, Δt; callbacks=nothing)


delete this method since you don't do 1D.

mochell · 2025-11-05T19:07:56Z

src/Simulations/run.jl

+        if sim.model.save_particles
+                save_path = sim.model.plot_savepath
+                filename = save_path*"/data/simu_infos.csv"
+
+                Nx = sim.model.grid.Nx
+                Ny = sim.model.grid.Ny
+                xmin = sim.model.grid.xmin
+                xmax = sim.model.grid.xmax
+                ymin = sim.model.grid.ymin
+                ymax = sim.model.grid.ymax
+                lne_source = sim.model.ODEdefaults.lne
+                c_x_source = sim.model.ODEdefaults.c̄_x
+                c_y_source = sim.model.ODEdefaults.c̄_y
+                x_source = sim.model.ODEdefaults.x
+                y_source = sim.model.ODEdefaults.y
+                angular_spread_source = sim.model.ODEdefaults.angular_σ
+                Δt = sim.Δt
+                stop_time = sim.stop_time
+
+                data = DataFrame(Nx=Nx, Ny=Ny, xmin=xmin,
+                                xmax=xmax, ymin=ymin, ymax=ymax,
+                                lne_source=lne_source, c_x_source=c_x_source,
+                                c_y_source=c_y_source, x_source=x_source,y_source=y_source,
+                                angular_spread_source=angular_spread_source,
+                                Δt=Δt, stop_time=stop_time)
+                CSV.write(filename, data)
+
+                filename2 = save_path*"/data/sigma.csv"
+
+                covariance_init = sim.model.proba_covariance_init
+                data2 = DataFrame(covariance_init, :auto)
+                CSV.write(filename2, data2)
+        end


Write this in a function and put it in the same place were write_particle_to_cvs is. then just call the function here.

mochell · 2025-11-05T19:09:43Z

src/Simulations/run.jl

+                if sim.model.plot_steps
+                        plot_state_and_error_points(sim, gridnotes)
+                        sec=string(Int64(floor((sim.model.clock.time)/60)))
+                        dec=string(Int64(floor(10*(sim.model.clock.time/60-floor((sim.model.clock.time)/60)))))
+                        plt.savefig(joinpath([save_path*"/plots/", "energy_plot_no_spread_"*sec*","*dec*".png"]))
+                end


write this in a function as well.

mochell · 2025-11-05T19:11:10Z

src/Simulations/run.jl

 usually the initilization uses wind constitions to seed the particles.
 """
-function init_particles!(model::Abstract2DModel; defaults::PP=nothing, verbose::Bool=false) where {PP<:Union{ParticleDefaults1D,ParticleDefaults2D,Nothing}}
+function init_particles!(model::Abstract2DModel; defaults::PP=nothing, verbose::Bool=false) where {PP<:Union{ParticleDefaults,Array{Any,1},Nothing}}


change model type here so this is then a new function type and doesn't conflict with the other code

mochell · 2025-11-05T19:12:00Z

src/Simulations/run.jl

        # @info ParticleCollection
        model.ParticleCollection = ParticleCollection
+
+        if defaults isa ParticleDefaults


maybe clean this up and comment your code a bit so one has an idea what you do

mochell · 2025-11-05T19:18:34Z

src/Simulations/run.jl

+        gridnotes = TwoDGridNotes(model.grid)

-                        SeedParticle2D(
-                                model.State, ij,
-                                model.ODEsystem, defaults, model.ODEsettings,
-                                model.grid.stats, model.grid.ProjetionKernel, model.grid.PropagationCorrection,
-                                ij_mesh, ij_wind,
-                                model.ODEsettings.timestep,
-                                model.boundary, model.periodic_boundary)
+        ParticleCollection = []
+        SeedParticle_i = SeedParticle_mapper(SeedParticle2D!,
+                ParticleCollection, model.State,
+                model.ODEsystem, nothing, model.ODEsettings,
+                gridnotes, model.winds, model.ODEsettings.timestep,
+                model.boundary, model.periodic_boundary)


This may not work anymore since I changed the type of the ParticleCollection.. we have to see ..

…ucture

…/PiCLES-fork into main-fork-pull-request

…e type, and new model type

…tic) work alongside each other

tomprotin added 27 commits February 16, 2024 16:50

small changes to make PiCLES work on my local

df1fa24

created new model for geometrical optics management, lots of changes …

713c64a

…comming...

Changed every occurence of WaveGrowth2D to GeometricalOptics

e601668

Added new particle system copied from particle_waves_v5

b7ab9cb

created new test case with no wind

a343bfe

created test case and fixed minor issue associated with it

46e5539

Merge branch 'main' into geometrical_spreading

6c442de

Added support for angular spreading; Now it works, but some changes a…

76a2cbb

…re left to do

added the possibility to plot during run time; added angular spreadin…

8cb38bb

…g types and threshold structure

added full support for stochas and geometrical spreading types

65adacf

Added a model parameter storing the number of particles to launch at …

f527f4f

…each node when using the "nonparametric" spreading type

Non parametric method for wave propagation; also added some basic str…

71ea18b

…uctures for the validation of the 1/r decay of energy density

updating code for SIR

e73a983

Creating the SIR analysis file

e58e246

added config files for the test case 2 of the paper

a77840c

added config files for the test case 1 of the paper

5b2c5a0

corrected the number of particles in test case 1

a581814

Merge remote-tracking branch 'fork/main' into open_code_STUOD-fork

3e63f27

removed unnecessary test case files

00543fb

further modified and removed useless test case files

5d5163b

fixed test_case_1

975cf7b

removed any reference to point source in Simulations.jl and run.jl

e0f9f91

completed previous commit

61d2c68

fixed ParticleSystem.jl and made ParticleInCell.jl compatible in both…

4bcd201

… versions

fixed mapping_2D.jl

15f9ebb

fixed TimeSteppers.jl

1473731

removed references to point sources in GeometricalOpticsModels.jl

70ee4c0

mochell self-assigned this Nov 5, 2025

mochell requested changes Nov 5, 2025

View reviewed changes

mochell self-requested a review November 5, 2025 19:19

mochell assigned tomprotin Nov 5, 2025

tomprotin and others added 14 commits December 9, 2025 14:37

moved SIR_analysis.jl to analysis/SIR_analysis.jl

18087f8

changed .gitignore

2609a1c

some other local changes

244a690

clean up gitignore and stuff

e871174

created a new abstract class for the stochastic model

04d984a

first modifications to adapt the code to the new stochastic model str…

5a78d35

…ucture

Merge branch 'main-fork-pull-request' of https://github.com/tomprotin…

3c95ea7

…/PiCLES-fork into main-fork-pull-request

changed Project.toml and made corrections to make the code work

32c4004

adapted functions in run.jl to the new Abstract2DStochasticModel

811e000

adapted functions in mapping_2D.jl to the new Abstract2DStochasticModel

5d18f6b

changed the stochastic method so it uses new grid system, new particl…

09ff80f

…e type, and new model type

removed useless print instruction

2d85f15

final modifications to make both versions (stochastic and non-stochas…

c943f31

…tic) work alongside each other

testing wind on non parametric method

8786e74

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Stochastic PiCLES pull request#42

Stochastic PiCLES pull request#42
tomprotin wants to merge 41 commits intomochell:stochastic_PiCLESfrom
tomprotin:main-fork-pull-request

tomprotin commented Oct 7, 2025

Uh oh!

mochell left a comment

Uh oh!

mochell Nov 5, 2025

Uh oh!

Uh oh!

Uh oh!

mochell Nov 5, 2025

Uh oh!

mochell Nov 5, 2025

Uh oh!

mochell Nov 5, 2025

Uh oh!

mochell Nov 5, 2025

Uh oh!

mochell Nov 5, 2025

Uh oh!

mochell Nov 5, 2025

Uh oh!

mochell Nov 5, 2025

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

2 participants

	Mstat} <: Abstract2DModel where {Mstat<:Union{Nothing,stat}, PCollection<:Union{Vector,Array}}
	Mstat} <: Abstract2DStochasticModel where {Mstat<:Union{Nothing,stat}, PCollection<:Union{Vector,Array}}

Conversation

tomprotin commented Oct 7, 2025

Uh oh!

mochell left a comment

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

2 participants