Add FEL simulation by mbkkunze · Pull Request #555 · IPPL-framework/ippl

mbkkunze · 2026-06-25T08:27:53Z

Summary

Adds a free-electron laser simulation under fel/, built with -DIPPL_ENABLE_FEL=ON.
It runs an FDTD/PIC simulation of a relativistic electron bunch through an
undulator in a co-moving Lorentz frame and writes the forward radiated power to CSV
(with an optional Poynting-flux video).

Build & run

cmake -S . -B build -DIPPL_ENABLE_FEL=ON -DCMAKE_CXX_STANDARD=20
cmake --build build --target FreeElectronLaser
cd build && mkdir -p ../renderdata
./fel/FreeElectronLaser ../fel/config.json --info 5

Notes

Bunch initialization and the resonance-power diagnostic are ported from
MITHRA (A. Fallahi, GPL).
More detail in fel/README.md.

…o, extended tests to include multiple particles.

…orm.

…cted simulation config parameters.

…el/, fix default config path

mbkkunze · 2026-06-25T08:30:14Z

Please add @s-mayani as reviewer

s-mayani · 2026-06-25T10:53:02Z

+    /* Type of the bunch which is one of the manual, ellipsoid, cylinder, cube, and 3D-crystal. If
+     * it is manual the charge at points of the position vector will be produced.
+     */
+    // std::string     			bunchType_;


why is this commented? remove if not used

s-mayani · 2026-06-25T10:53:27Z

+    FieldVector<scalar> initialDirection_;
+
+    /* Position of the center of the bunch in the unit of length scale. */
+    // std::vector<FieldVector<scalar> >	position_;


remove commented line

s-mayani · 2026-06-25T10:53:48Z

+    FieldVector<scalar> betaVector_;
+
+    /* Initialize the parameters for the bunch initialization to some first values. */
+    // BunchInitialize ();


s-mayani · 2026-06-25T10:54:03Z

+     */
+    Double e;
+
+    // Charge();


remove commented line

s-mayani · 2026-06-25T10:54:38Z

+        // printmessage(std::string(__FILE__), __LINE__, std::string("Warning: The number of
+        // particles in the bunch is not a multiple of four. ") +
+        //     std::string("It is corrected to ") +  std::to_string(bunchInit.numberOfParticles_) );


you could also print this in an IPPL inform message, and uncomment it

s-mayani · 2026-06-25T10:55:02Z

+        // printmessage(std::string(__FILE__), __LINE__, std::string("The bunching factor can not be
+        // larger than one or a negative value !!!") ); exit(1);


can uncomment this and turn it into an IpplException

s-mayani · 2026-06-25T10:55:27Z

+        // if 	( bunchInit.generator_ == "random" )
+        return (randomNumbers.at(n * 2 * Np / ng + m));
+        // else
+        //   return  ( randomNumbers[ n * 2 * Np/ng + m ] );
+        // TODO: Return halton properly
+        // return 	( halton(n,m) );


remove all the commented code here (do we really need halton rng? )

s-mayani · 2026-06-25T10:57:13Z

+    }
+
+    /* If the longitudinal type of the bunch is uniform a tapered part needs to be added to remove
+     * the CSE from the tail of the bunch.


change the CSE to full form of the acronym

s-mayani

Left some minor comments to be addressed

aaadelmann · 2026-06-30T08:53:38Z

cscs-ci run cscs-ci-gh200, cscs-ci-mi300, cscs-ci-openmp

The deposition kernel applies a half-cell shift before computing CIC weights: gp = (mid - origin) / h - 0.5 This was introduced so current deposition uses the same centering convention as the field gather on the cell-centered UniformCartesian grid. With that shift, a particle segment whose midpoint is at the geometric cell center no longer splits uniformly over the surrounding 2^Dim nodes. Instead, it lands exactly on the shifted node location. The exact-value unit test still used the older unshifted convention: gp = (mid - origin) / h This made the test expect, for example, a 2D x-directed segment from 0.25 to 0.75 at y=0.5 to deposit Jx = 0.125 on each of four neighboring nodes. The current kernel correctly deposits the full Jx = 0.5 on the shifted node and zero on the other three nodes. This mismatch caused TestCurrentDeposition to fail for all 2D and 3D typed exact-value cases. This change updates the expected tables and comments in TestCurrentDeposition.cpp for the half-cell-shifted convention: Verification: - Built TestCurrentDeposition - Ran TestCurrentDeposition via CTest: passed - Ran full OpenMP unit_tests CTest tree: 35/35 passed

aaadelmann · 2026-06-30T09:33:49Z

cscs-ci run cscs-ci-gh200, cscs-ci-mi300, cscs-ci-openmp

aaadelmann · 2026-06-30T09:34:57Z

@s-mayani @mbkkunze please check carefully ccd6227

aaadelmann · 2026-06-30T09:35:40Z

cscs-ci run cscs-ci-gh200, cscs-ci-mi300, cscs-ci-openmp

aaadelmann · 2026-06-30T10:09:14Z

cscs-ci run cscs-ci-gh200, cscs-ci-mi300, cscs-ci-openmp

s-mayani · 2026-06-30T10:14:11Z

@s-mayani @mbkkunze please check carefully ccd6227

I am confused, why do we need this? I thought the tests were already working correctly

aaadelmann · 2026-06-30T10:40:12Z

This test was failing at CSCS and local on my laptop. Lets see if I can find earlier tests at CSCS that passed.

aaadelmann · 2026-06-30T11:13:07Z

At cscs #555 did never pass all tests (https://my.cdash.org/queryTests.php?project=IPPL&date=2026-06-30&filtercount=2&showfilters=1&filtercombine=and&field1=status&compare1=62&value1=passed&field2=buildname&compare2=63&value2=555)

Bit still it could be a side effect of an other commit.

s-mayani · 2026-06-30T11:50:57Z

At cscs #555 did never pass all tests (https://my.cdash.org/queryTests.php?project=IPPL&date=2026-06-30&filtercount=2&showfilters=1&filtercombine=and&field1=status&compare1=62&value1=passed&field2=buildname&compare2=63&value2=555)

Bit still it could be a side effect of an other commit.

Okay I see, but I am still confused since the current projection was merged in a separate PR which did pass tests.

mbkkunze · 2026-07-01T23:43:10Z

Pushed the cleanup from @s-mayani comments.

Thank you very much for fixing the tests numbers @aaadelmann! Looks all correct to me. A bit of explanation: At the time when the current deposition was merged previously, I was still thinking that the solver was using a staggered grid, so I wrote the deposition for that. Then it turned out that this was not the case, so I had to change to the colocated grid. When I later fixed a half-cell off bug, I forgot to update the tests accordingly.

Do we want to keep the staggered version for future use (If yes I will re-add it and its tests), or keep only the colocated one which is what the Maxwell solver uses?

aaadelmann · 2026-07-02T07:52:59Z

Yes push the other test too!

mbkkunze added 10 commits June 7, 2026 04:01

Adjusted CurrentDeposition to colocated grid used by FDTD solver. Als…

c1da98b

…o, extended tests to include multiple particles.

Clean and updated FEL implementation.

eaf5c9f

Repoduced radiation curve from mithra. Imporant fix in lorentz transf…

cc4d18c

…orm.

Radiation curve plotting, simulation for standard solver case.

f3d9cdc

Some cleaning, added undulator exit fringe.

75475aa

Adjusted simulation parameters to mithra.

ffee24a

Replicated resonance frequency filter on forward radiation from MITHRA

c3c1c54

Fixed half-cell-off bug in deposition. Seeded bunch generation. Corre…

06e8651

…cted simulation config parameters.

FEL: drop root scratch/config files, ship cleaned example config in f…

12df5e0

…el/, fix default config path

FEL: add module README

1048d98

aaadelmann requested a review from s-mayani June 25, 2026 10:24

s-mayani reviewed Jun 25, 2026

View reviewed changes

Comment thread fel/MithraBunch.h Outdated

*/

Double e;

// Charge();

s-mayani Jun 25, 2026

Copy link
Copy Markdown

Collaborator

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

remove commented line

s-mayani reviewed Jun 25, 2026

View reviewed changes

Merge branch 'master' into FEL

61abec7

FEL: address review comments in MithraBunch.h

0cc796e

klappi-s mentioned this pull request Jul 8, 2026

Unified Demos Folder #563

Open

		// printmessage(std::string(__FILE__), __LINE__, std::string("The bunching factor can not be
		// larger than one or a negative value !!!") ); exit(1);

Uh oh!

Conversation

mbkkunze commented Jun 25, 2026

Summary

Build & run

Notes

Uh oh!

mbkkunze commented Jun 25, 2026

Uh oh!

s-mayani Jun 25, 2026

Choose a reason for hiding this comment

Uh oh!

s-mayani Jun 25, 2026

Choose a reason for hiding this comment

Uh oh!

s-mayani Jun 25, 2026

Choose a reason for hiding this comment

Uh oh!

s-mayani Jun 25, 2026

Choose a reason for hiding this comment

Uh oh!

s-mayani Jun 25, 2026

Choose a reason for hiding this comment

Uh oh!

s-mayani Jun 25, 2026

Choose a reason for hiding this comment

Uh oh!

s-mayani Jun 25, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

s-mayani Jun 25, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

s-mayani left a comment

Choose a reason for hiding this comment

Uh oh!

aaadelmann commented Jun 30, 2026

Uh oh!

aaadelmann commented Jun 30, 2026

Uh oh!

aaadelmann commented Jun 30, 2026

Uh oh!

aaadelmann commented Jun 30, 2026

Uh oh!

aaadelmann commented Jun 30, 2026

Uh oh!

s-mayani commented Jun 30, 2026

Uh oh!

aaadelmann commented Jun 30, 2026

Uh oh!

aaadelmann commented Jun 30, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

s-mayani commented Jun 30, 2026

Uh oh!

mbkkunze commented Jul 1, 2026

Uh oh!

aaadelmann commented Jul 2, 2026

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

3 participants

s-mayani Jun 25, 2026 •

edited

Loading

s-mayani Jun 25, 2026 •

edited

Loading

aaadelmann commented Jun 30, 2026 •

edited

Loading