SIFI-CM 'simple'¹ Geant4 Simulation

• Prerequisites
• Installation
• Usage
  • Simulation
  • H-matrix calculation
• Geometry
• References

• Required ROOT version: 6.24/00
  For the installation guide look at root_official²
• Compiler supporting c++17
• CmdLineArgs >= 2.2.0
• mpich

• Clone the repository

git clone https://github.com/SiFi-CC/G4Simulation.git

cd G4Simulation

• Prepare submodules (spdlog) by running:

git submodule update --init --recursive

• Create build directory and compile the code

mkdir build && cd build
cmake ..
make -jN

• For using visual mode one will need to copy macro files to the build directory as well

cp ../*.mac .

• SD - distance from the source plane to the detector
• SM - distance from the source plane to the middle of the mask
• MD - distance from the middle of the mask to the detector (SD = SM + MD)
• M - size of the mask in X and Y directions
• M_z - mask thickness
• S - size of the source plane. This dimension is specified during the system matrix calculation. It determines a space of reconstructed images for a particular matrix.
• FCFOV (the fully coded field of view) - angular region which is defined as defined as "comprising all directions for which the detected flux is completely modulated by the mask"[Caroli et al].
• FOV (the field of view) - angular region where "only a fraction of the detected photons is coded by the aperture pattern"[Caroli et al].

The size of the detector plane is determined by the fibers size and amount. It is hardcoded that the length of a single fiber is 10 cm. Also it is not hardcoded but recommended to take a width of a fiber as 1.36 mm (despite the fact that default value is 1.3 mm which should be corrected).

So in 2D regime, the detector plane would be a square-shaped and one side of such a square would be equal to nFibres X fibre_width.

In 1D the size of the detector plane in Y direction is always the same as fiber length (10 cm), and in X direction it is equal to nFibres X fibre_width.

The thickness of the detector is always determined as nLayers X fibre_width

At the moment it is decided to test a nowallpet mask (or nowallpetcut) for the 2D mode (the practical implementation can be slightly different for the 1D). For that purpose the Tungsten rods had been ordered together with the first PET raster. It has been done for the 31 order mask, with the size M = 70 mm and M_z = 20 mm. So each rod is in size 70/31 ≈ 2.3 mm. The number of rods available is 480(or so?) This information should be taken into account when constructing another mask patterns - that is the single pixel size should kept the same (no matter it is 1D or 2D).

At the same time, it can be considered to use larger mask, but additional rods should be ordered than.

At this point experiments with geometry parameters led to finding optimal (or just better than other in some particular cases) values and set-ups.

64 fibers in total, tested with 4 layers by 16 (can be also arranged in 2 layers by 32 or 1 layer with 64 fibers)

• SM = 170 mm
• MD = 50 mm
• M = 70 mm
• M_z = 20 mm
• S = 70 mm
• Number of points in the source plane: 100x100=10000

7 layers with 55 fibers 2.01 mm fiberwidth

• SM = 170 mm
• MD = 50 mm
• M = 115.3 x 101.7 mm, 51 x 45 pixels, 467 order
• M_z = 20 mm
• FOV = 130 mm, 200 bins

The program operates in 2 modes. The first of them is used to simulate an experiment of γ-particles being shot towards the detector. And the second - for the calculation of the system matrix (needed for the reconstruction).

The general procedure is very similar for both modes. The basic difference is that the second performs a set of single simulations and combines them together in the form of system matrix.

./cmd/custom_simulation.cp output.root [optional arguments]

output.root is a name of the output root-file.

./cmd/custom_simulation simtest2.root -det 220:55:2.01 -mask 467:170:115.3:101.7:20 -n  1000 -masktype nowallpetcut -cutx 51 -cuty 45 -sourceBins 130:200 -nlay 7 -source -20:0 -1d -fullscat

(detailed description below the table)

Parameter -det

List of 3 values which define define geometry parameters, connected to the detector. First one is detector-source, that is a distance between the source and the detector. This parameter is followed by the number of fibers in the single layer and width of a single fiber.

Parameter -nlay

This parameter allows user to set a number of the layers in the detector.

Parameter -mask

List of 4 values which define define geometry parameters, connected to the MURA mask . First one defines an order of the mask (should be a prime number). Second value is mask-source, that is a distance between the source and the mask. The last 2 parameters are the width(and the length - as the mask is a square) and thickness.

Parameter -masktype

Specify the type of implementation of the MURA mask. It should be one of the following options:

• standard - ideal (theoretical) case when the mask is truly binary and has no walls between the pixels

• round - the mask is constructed as a Tungsten monoblock, where in places of empty pixels are round-shape holes

• pet - the mask has a PET raster where it is possible to put Tungsten rods and prepare any pattern. The size of rods is hardcoded as a 90% of the full pixel size.

• nowallpet - similar to pet but the raster is prepared specifically to the particular pattern. It means that empty pixels are filled with PET and it allows to avoid having walls.

• nowallpetcut - the same as nowallpet but allows option -cut to be activated

Parameter -cut

If -masktype=nowallpetcut, the cut parameter specifies a size of mask's central area to be used. For example if mask order is 467 and -cut=31, then only 31X31 central pixels of the 467's order MURA mask will be used.

Parameter -er

Specifies the error in the size of Tungsten rods. If used, than from the each dimension of each rod is subtracted a random number from the distribution: er/2 - N(er/2, er/6)

Parameter -source

The list of 2 numbers, which specifies a coordinates of the source (or the center of the source if not point-like).

Parameter -sourceBins

The list of 2 values (float and int) to specify a source histogram range and number of bins. The first of them is a size of the plane in [mm] and the second is a number of source positions in each direction (positions are uniformly distributed among the source plane).

Parameter -n

Number of generated events.

Parameter -e

Energy of generated γ-particles [kEv].

Parameter -theta

This parameter allows to set the θmin angle [deg]. θmax is always 180 [deg] (parallel to the source-detector axis). It means that particles will be shot with angles [θmin, 180].

Parameter -sMac

Path to the mac-script with additional specifications of the source. It allows to change type and shape of the source distribution.

Parameter -sFile

Path to tsv PhaseSpace file with eache row containing values: ID:energy:x:y:z:p_x:p_y:pz parameters of simulated gamma. In such case parameter -n will specify how many times each gamma will be simulated.

Parameter -vis

If this flag is used, the programs runs in visual mode³. That is Geant4 graphical interface is opened with specified geometry setup.

IMPORTANT you need to copy all the *.mac files to the build catalog.

Parameter -1d

If this flag is used, the detector and mask are constructed in a singe-dimensional mode along x-axis

Parameter -detshift

Two values (X and Y direction) which determine the shift of the detector with respect to the center of mask. If one value is inserted - the shift will be applied only in X direction.

Example for full-scale prototype

parallel -j ${njobs} --lb --progress "./cmd/custom_simulation ./sim_PMMA90p7MeV_energy_smear/sim220_170-det55_2p01_7lay-nowallpetcut-mask467_cut51_45-70mm_1d_PMMA90p7MeV{/.}_0_0.root -det 220:55:2.01 -mask 467:170:115.3:101.7:20 -n  1 -masktype nowallpetcut -cutx 51 -cuty 45 -sourceBins 130:200 -1d -nlay 7 -sFile {} -source 0:0 >>| logfile.log || echo {} error; echo {} >>| error_file.log" ::: ${phasespace_files}.tsv

mpirun -np ${n_cores} ./cmd/mpi_gen_sim_matrix.cp output.root [optional arguments]

mpirun -np 6 ./cmd/mpi_gen_sim_matrix matr220_170-fullscat-nowallpetcut-mask467_cut51_45-70mm_1d_layerwise.root -det 220:55:2.01 -mask 467:170:115.3:101.7:20 -n  1000 -masktype nowallpetcut -cutx 51 -cuty 45 -source 130:100 -1d -nlay 7 -fullscat

n_cores - is a number of parallel processes (number of cores used).

output.root is a name of the output root-file.

The most of the optional arguments are the same as for the simulation, but some of them are not used or have different meaning.

The list of 2 values (float and int) to specify a source plane parameters. The first of them is a size of the plane in [mm] and the second is a number of source positions in each direction (positions are uniformly distributed among the source plane).⁴

All the rest parameters have the same meaning and usage as for Simulation

[Caroli et al] : Caroli, E., Stephen, J.B., Di Cocco, G. et al. Coded aperture imaging in X- and gamma-ray astronomy. Space Sci Rev 45, 349–403 (1987). https://doi.org/10.1007/BF00171998