2.5D Gaussian Solver #1166

qianglbl · 2025-10-06T22:37:39Z

added 3 input parameters for the 2.5D Gaussian solver, modified Gauss2p5dPush.cpp, added examples in fodo_space_charge.

Enhances the implementation from #1127

Add the new Gauss 3D space charge pusher.

- replace with new file name in doc - move into own section in CMake

- avoid temporary arrays - integrate in one go

Only used for particles. Ref orbit asserts well, too.

… 2.5D sC

src/particles/spacecharge/Gauss2p5dPush.cpp

+        int tp5d_bins = 129;
+        pp_algo.queryAddWithParser("gauss2p5_bins", tp5d_bins);
+
+        // Measure beam size, extract the min, max of particle positions


src/particles/spacecharge/Gauss2p5dPush.cpp

+        int tp5d_bins = 129;
+        pp_algo.queryAddWithParser("gauss2p5_bins", tp5d_bins);
+
+        // Measure beam size, extract the min, max of particle positions


src/particles/spacecharge/Gauss2p5dPush.cpp

+        int tp5d_bins = 129;
+        pp_algo.queryAddWithParser("gauss2p5_bins", tp5d_bins);
+
+        // Measure beam size, extract the min, max of particle positions


src/particles/spacecharge/Gauss2p5dPush.cpp

+        int tp5d_bins = 129;
+        pp_algo.queryAddWithParser("gauss2p5_bins", tp5d_bins);
+
+        // Measure beam size, extract the min, max of particle positions


src/particles/spacecharge/Gauss2p5dPush.cpp

src/particles/spacecharge/Gauss3dPush.cpp

examples/fodo_space_charge/analysis_fodo_Gauss3D_sc.py

cemitch99 · 2025-10-09T00:26:55Z

Great! Changing the tolerance now allows the test of the initial beam moments to pass in all cases. For the final beam moments, the Gauss3d test still fails in a few cases, and this does not appear to be due to noise or tolerance. (For example, the initial and final beam moments output by the test on macOS are identical to each other.)

cemitch99 · 2025-10-16T19:48:42Z

Summary of failed tests (on macOS / AppleClang, OpenMP / GCC w/MPI w/Python, Tooling / Clang ASAN UBSAN): FODO.Gauss3d.sc fails, and the initial and final beam moments are identical.

qianglbl · 2025-10-16T20:07:21Z

I will look into that. Thanks, Ji

…

On Thu, Oct 16, 2025 at 12:49 PM Chad Mitchell ***@***.***> wrote: *cemitch99* left a comment (BLAST-ImpactX/impactx#1166) <#1166 (comment)> Summary of failed tests (on macOS / AppleClang, OpenMP / GCC w/MPI w/Python, Tooling / Clang ASAN UBSAN): FODO.Gauss3d.sc fails, and the initial and final beam moments are identical. — Reply to this email directly, view it on GitHub <#1166 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AJHRSUNNKUT65B6UJKZ344L3X7ZDDAVCNFSM6AAAAACIOQIZ7WVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMZTIMJSGYYDKOBQGU> . You are receiving this because you authored the thread.Message ID: ***@***.***>

ax3l · 2025-10-16T20:33:41Z

src/particles/spacecharge/Gauss2p5dPush.cpp

+    // Input particle locations (x,y) and RMS sizes (sigx,sigy) and return the integrals for SC fields.
+    //
+    AMREX_GPU_DEVICE
+    void potInt(amrex::ParticleReal delta, int nint, amrex::ParticleReal xin, amrex::ParticleReal yin, amrex::ParticleReal sigx, amrex::ParticleReal sigy,


Suggested change

void potInt(amrex::ParticleReal delta, int nint, amrex::ParticleReal xin, amrex::ParticleReal yin, amrex::ParticleReal sigx, amrex::ParticleReal sigy,

void potInt (amrex::ParticleReal delta, int nint, amrex::ParticleReal xin, amrex::ParticleReal yin, amrex::ParticleReal sigx, amrex::ParticleReal sigy,

examples/fodo_space_charge/analysis_fodo_Gauss2p5D_sc.py

cemitch99 · 2025-10-17T01:31:24Z

I think I see why the remaining tests are failing now. I can help you resolve those tomorrow if you have time.

examples/fodo_space_charge/analysis_fodo_Gauss2p5D_sc.py

cemitch99 · 2025-10-17T22:17:14Z

examples/fodo_space_charge/analysis_fodo_Gauss3D_sc.py


 print("Initial Beam:")
-sig_xi, sig_yi, sig_ti, emittance_xi, emittance_yi, emittance_ti = get_moments(initial)
+sig_xi, sig_yi, sig_ti, emittance_xi, emittance_yi, emittance_ti = get_moments(final)


Suggested change

sig_xi, sig_yi, sig_ti, emittance_xi, emittance_yi, emittance_ti = get_moments(final)

sig_xi, sig_yi, sig_ti, emittance_xi, emittance_yi, emittance_ti = get_moments(initial)

Initial and final moments are switched here.

cemitch99 · 2025-10-17T23:59:44Z

@qianglbl It looks like the Gauss3d tests are all passing! The failing Gauss2p5d tests are now due to errors in the final sigma_x and sigma_y of around 10% and 5%, respectively. (The tolerance is 3%.) Those errors seems too large...

examples/fodo_space_charge/analysis_fodo_Gauss2p5D_sc.py

cemitch99 · 2025-11-04T18:18:08Z

@qianglbl @ax3l I tried running this test locally on my Mac with 12 OMP threads, and it worked fine. It is notable that the three failing checks above all return the same wrong answer.

ax3l · 2025-11-06T19:32:36Z

src/particles/spacecharge/Gauss2p5dPush.cpp

+        // Call charge deposition function
+        impactx::particles::wakefields::DepositCharge1D(pc, charge_distribution, bin_min, bin_size, is_unity_particle_weight);
+
+        // Call charge density derivative function
+        amrex::Gpu::DeviceVector<amrex::Real> slopes(charge_distribution.size() - 1, 0.0);
+        impactx::particles::wakefields::DerivativeCharge1D(charge_distribution, slopes, bin_size,GetNumberDensity);


Oh no, we forget to MPI-AllReduce the deposited charge to all ranks.

Will help fix that.

impactx/src/particles/wakefields/HandleWakefield.H

Lines 81 to 92 in 10feefc

// Call charge deposition function

impactx::particles::wakefields::DepositCharge1D(particle_container, charge_distribution, bin_min, bin_size, is_unity_particle_weight);

// Sum up all partial charge histograms to one MPI process to calculate the global wakefield.

// Once calculated, we will distribute convolved_wakefield back to every MPI process.

amrex::ParallelReduce::Sum(

charge_distribution.data(),

charge_distribution.size(),

amrex::ParallelDescriptor::IOProcessorNumber(),

amrex::ParallelDescriptor::Communicator()

);

Before we calc the derivatives.

ax3l · 2025-11-07T05:48:33Z

examples/fodo_space_charge/README.rst

+.. _examples-fodo-2p5dGaussian-sc:
+
+FODO Cell with 2.5D Gaussian Space Charge Using Particle Tracking


@qianglbl this section of the README still needs to be finished, please.

qianglbl and others added 22 commits September 4, 2025 16:59

Gauss 3D SC Pusher

8739136

Add the new Gauss 3D space charge pusher.

add reference for 3D Gaussian SC solver

5e185f2

Fix Compilation Together

e019105

fixed a defect in 3D Gaussian SC

c35918d

added FODO test example with SC from a 3D Gaussian distribution

e953591

added reference for the 3D SC Gaussian distribution solver.

ef44a50

User-Facing Docs (Manual)

42e59e7

Example: README & CMake Update

cdd97d9

- replace with new file name in doc - move into own section in CMake

Source: Cleaning, Formatting, TODOs

d1bee59

Update analysis

75d25a9

GPU Support, Performance Opt

c30c8a4

- avoid temporary arrays - integrate in one go

Envelope: Abort on Gauss3D

823a199

Only used for particles. Ref orbit asserts well, too.

Python Bindings: New Allowed Value

8cb841f

Python Example

b63339a

Cleanup

b9024b9

Merge remote-tracking branch 'mainline/development' into topic-gauss3d

a01d416

GPU Kernel: AMREX_GPU_DEVICE

ae8b780

Apply suggestions from code review

b7c297a

fixed the 3D Gaussian distribution pusher

bbc6cb7

added 2.5D SC kicks with transverse Gaussian distribution

33b30be

updated Gauss2.5D SC on GPUs

b865c75

added nint,bins, and delta input parameters and examples for Gaussian…

af76457

… 2.5D sC

qianglbl requested review from ax3l and cemitch99 October 6, 2025 22:37

Merge branch 'development' into topic-gauss2p5dv2

7ba4b2c

github-advanced-security bot found potential problems Oct 6, 2025

View reviewed changes

ax3l reviewed Oct 7, 2025

View reviewed changes

src/particles/spacecharge/Gauss3dPush.cpp Outdated Show resolved Hide resolved

ax3l reviewed Oct 7, 2025

View reviewed changes

src/particles/spacecharge/Gauss3dPush.cpp Outdated Show resolved Hide resolved

This was referenced Oct 7, 2025

Gauss 3D: Fix Push Constants #1168

Merged

Topic gauss2p5d #1163

Closed

cemitch99 reviewed Oct 8, 2025

View reviewed changes

examples/fodo_space_charge/analysis_fodo_Gauss3D_sc.py Show resolved Hide resolved

qianglbl added 3 commits October 8, 2025 15:48

Update analysis_fodo_Gauss3D_sc.py

3182852

Update parameters.rst

3129814

Update python.rst

d552588

Update analysis_fodo_Gauss3D_sc.py

3875a9c

Fix analysis_fodo_Gauss3D_sc.py (final moments)

6b5caef

ax3l reviewed Oct 16, 2025

View reviewed changes

ax3l mentioned this pull request Oct 16, 2025

Gauss3D: Expose Integration Steps #1153

Closed

ax3l reviewed Oct 16, 2025

View reviewed changes

examples/fodo_space_charge/analysis_fodo_Gauss2p5D_sc.py Outdated Show resolved Hide resolved

Beam Moments (Final)

b9bfb5c

ax3l reviewed Oct 16, 2025

View reviewed changes

examples/fodo_space_charge/analysis_fodo_Gauss2p5D_sc.py Show resolved Hide resolved

2.5D Gaussian SC example

5dca30c

fixed an error related to 2.5D Gaussian SC

d49eca4

cemitch99 reviewed Oct 17, 2025

View reviewed changes

examples/fodo_space_charge/analysis_fodo_Gauss2p5D_sc.py Outdated Show resolved Hide resolved

cemitch99 reviewed Oct 17, 2025

View reviewed changes

examples/fodo_space_charge/analysis_fodo_Gauss2p5D_sc.py Outdated Show resolved Hide resolved

cemitch99 reviewed Oct 17, 2025

View reviewed changes

cemitch99 added 2 commits October 17, 2025 15:18

Fix switched initial/final moments in analysis_fodo_Gauss3D_sc.py

fb0e9bb

Relax tolerance slightly.

12c2e9c

ax3l reviewed Oct 23, 2025

View reviewed changes

examples/fodo_space_charge/analysis_fodo_Gauss2p5D_sc.py Show resolved Hide resolved

ax3l reviewed Nov 6, 2025

View reviewed changes

ax3l added 3 commits November 6, 2025 21:13

Fix MPI: AllReduce ChargeDistribution

9f4eff6

Constants, Formatting, Docs

9f91794

Doc: User-facing Option

8a4a6c2

ax3l reviewed Nov 7, 2025

View reviewed changes

	void potInt(amrex::ParticleReal delta, int nint, amrex::ParticleReal xin, amrex::ParticleReal yin, amrex::ParticleReal sigx, amrex::ParticleReal sigy,
	void potInt (amrex::ParticleReal delta, int nint, amrex::ParticleReal xin, amrex::ParticleReal yin, amrex::ParticleReal sigx, amrex::ParticleReal sigy,

	sig_xi, sig_yi, sig_ti, emittance_xi, emittance_yi, emittance_ti = get_moments(final)
	sig_xi, sig_yi, sig_ti, emittance_xi, emittance_yi, emittance_ti = get_moments(initial)

	// Call charge deposition function
	impactx::particles::wakefields::DepositCharge1D(particle_container, charge_distribution, bin_min, bin_size, is_unity_particle_weight);

	// Sum up all partial charge histograms to one MPI process to calculate the global wakefield.
	// Once calculated, we will distribute convolved_wakefield back to every MPI process.
	amrex::ParallelReduce::Sum(
	charge_distribution.data(),
	charge_distribution.size(),
	amrex::ParallelDescriptor::IOProcessorNumber(),
	amrex::ParallelDescriptor::Communicator()
	);

		.. _examples-fodo-2p5dGaussian-sc:

		FODO Cell with 2.5D Gaussian Space Charge Using Particle Tracking

2.5D Gaussian Solver #1166

Are you sure you want to change the base?

2.5D Gaussian Solver #1166

Uh oh!

Conversation

qianglbl commented Oct 6, 2025 • edited by ax3l Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

Check notice

Check notice

Check notice

Check notice

Uh oh!

Uh oh!

Uh oh!

Uh oh!

cemitch99 commented Oct 9, 2025

Uh oh!

cemitch99 commented Oct 16, 2025

Uh oh!

qianglbl commented Oct 16, 2025 via email

Uh oh!

ax3l Oct 16, 2025

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Uh oh!

cemitch99 commented Oct 17, 2025

Uh oh!

Uh oh!

Uh oh!

cemitch99 Oct 17, 2025

Choose a reason for hiding this comment

Uh oh!

cemitch99 Oct 17, 2025

Choose a reason for hiding this comment

Uh oh!

cemitch99 commented Oct 17, 2025

Uh oh!

Uh oh!

cemitch99 commented Nov 4, 2025

Uh oh!

ax3l Nov 6, 2025

Choose a reason for hiding this comment

Uh oh!

ax3l Nov 6, 2025

Choose a reason for hiding this comment

Uh oh!

ax3l Nov 6, 2025

Choose a reason for hiding this comment

Uh oh!

ax3l Nov 7, 2025

Choose a reason for hiding this comment

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

3 participants

qianglbl commented Oct 6, 2025 •

edited by ax3l

Loading