TestFlatTop.cpp File Reference

#include <gtest/gtest.h>
#include <mpi.h>
#include <algorithm>
#include <cmath>
#include <memory>
#include "Distribution/FlatTop.h"
#include "Ippl.h"
#include "PartBunch/BunchStateHandler.h"
#include "Utility/IpplTimings.h"

Include dependency graph for TestFlatTop.cpp:

Go to the source code of this file.

Classes
class	FlatTopTest

Functions
	TEST_F (FlatTopTest, UniformDiskStatisticsAndBounds)
	Testing generatiioin of uniform sampples on an elliptical disk.
	TEST_F (FlatTopTest, CountEnteringParticles_NoDomainDecomp)
	This test counts the number of particles entering the domain per rank for a FlatTop pulse.

Function Documentation

TEST_F() [1/2]

TEST_F	(	FlatTopTest	,
		CountEnteringParticles_NoDomainDecomp
	)

This test counts the number of particles entering the domain per rank for a FlatTop pulse.

This test verifies that FlatTop::countEnteringParticlesPerRank correctly counts the number of particles entering the domain within a given time window when domain decomposition is disabled.

A total of \(N_{\mathrm{tot}}\) particles are emitted over the full pulse duration according to a normalized FlatTop temporal profile \(f(t)\). The expected number of particles entering during a time window \([t_0, t_f]\) is computed analytically.

The number of particles emitted in the interval \([t_0, t_f]\) is

\[ N_{\mathrm{new}} = \left\lfloor N_{\mathrm{tot}} \frac{ \int_{t_0}^{t_f} f(t)\,dt }{ \int f(t)\,dt } \right\rfloor. \]

The integral over the time window is approximated using a trapezoidal rule:

\[ \int_{t_0}^{t_f} f(t)\,dt \approx \frac{1}{2} \left( f(t_0) + f(t_f) \right) (t_f - t_0). \]

With domain decomposition disabled, particles are distributed evenly across all MPI ranks. The expected number of particles per rank is therefore

\[ N_{\mathrm{new,local}}} = \left\lfloor \frac{N_{\mathrm{new}}}{{\mathrm{\#ranks}}} \right\rfloor, \]

with the remainder

\[ remainder = N_{\mathrm{new}} - \#{\mathrm{rank}} N_{\mathrm{new,local}} \]

assigned to rank 0, leading to

\[ N_{\mathrm{new,local}}^{(0)} = N_{\mathrm{new,local}} + remainder. \]

and \(N_{\mathrm{new,local}}^{(r)} = N_{\mathrm{new,local}}\) for \(r > 0\).

The total number of particles entering the domain is therefore

\[ N_{\mathrm{new,global}} = \sum_{r=0}^{N_{\mathrm{ranks}}-1} N_{\mathrm{new,local}}^{(r)}. \]

The test verifies that

\[ N_{\mathrm{new,global}} = N_{\mathrm{expected}}, \]

The test verifies that the number of particles counted on each rank matches this analytic expectation exactly.

Definition at line 252 of file TestFlatTop.cpp.

References FlatTop::allocateParticles(), FlatTop::countEnteringParticlesPerRank(), FlatTop::FlatTopProfile(), FlatTop::getDistArea(), and FlatTop::setWithDomainDecomp().

Here is the call graph for this function:

TEST_F() [2/2]

TEST_F	(	FlatTopTest	,
		UniformDiskStatisticsAndBounds
	)

Testing generatiioin of uniform sampples on an elliptical disk.

This test verifies that FlatTop::generateUniformDisk generates particle positions uniformly over the elliptical region

\[ \mathcal{D} = \left\{ (x,y) \in \mathbb{R}^2 : \frac{R_x^2}{\sigma_x^2} + \frac{R_y^2}{\sigma_y^2} \le 1 \right\}. \]

The test checks the following properties:

• Support: All generated particles satisfy

  \[ \frac{R_x^2}{\sigma_x^2} + \frac{R_y^2}{\sigma_y^2} \le 1. \]

• Zero momentum:

  \[ \mathbf{P}_i = \mathbf{0}. \]

• First moments: By symmetry of the uniform disk,

  \[ \mathbb{E}[R_x] = \mathbb{E}[R_y] = 0. \]

• Second moments: For a uniform distribution over an elliptical disk,

  \[ \mathbb{E}[R_x^2] = \frac{\sigma_x^2}{4}, \qquad \mathbb{E}[R_y^2] = \frac{\sigma_y^2}{4}. \]

The empirical sample moments are compared against these analytic values using a finite-sample tolerance.

Definition at line 104 of file TestFlatTop.cpp.

References FlatTop::generateUniformDisk().

Here is the call graph for this function: