-f filename.json
or
-f filename.inp
filename may contain absolute or relative path. If the filename has a '.json' extension, this input file is in the JSON format. If the input file name ends with '.inp', it is in the legacy format. If the file name contains space, one must use quotation marks to quote the full file name. In Windows, the quotation must be double quote (""). In Linux or Mac, either single or double quotation mark works.
If this flag is followed by an inline JSON string, such as -f '{...}', the parameter is parsed as an "inline JSON input file".
One can execute one of the built-in benchmarks by following the benchmark name, such as
--bench skinvessel
The default settings in each built-in benchmark can be overwritten by the additional command line flags, such as --bench cube60 -n 1e8 runs 1e8 photons instead of the default 1e6. Use --bench name --dumpjson to export the benchmark as a standalone JSON input file.
-n N
where N can be either an integer, such as 100000 or a floating point number, such as 1e5. If N is a floating point number, it will be rounded to the largest integer below the current value.
-r N
where N is a positive integer. When use this option, mcx runs 'nphoton*N' photons with nphoton per batch.
If N is a negative integer, this option splits the total simulation into abs(N) number of smaller batches; in each batch, only nphoton/abs(N) number of photons are simulated.
This option is useful when one does not have access to a dedicated GPU, i.e. the GPU is used for both display and computation. In such case, the mcx computation time is limited to only 5 to 10 seconds due to the graphics driver. One can use the -r option to make the execution time per batch under the time limit set by the driver.
'-b' or '-b 1' to enable (default) or '-b 0' to disable
When enabled, mcx will either reflect or transmit the photon at the boundaries based on the Fresnel's equation.
By default, mcx considers refractive index mismatch at the boundaries.
--bc option describes per-face boundary condition (BC). The first 6 letters
defines the boundary condition for bounding box faces at -x,-y,-z,+x,+y,+z axes; overwrite -b if given.
The first 6 letters can be one of the following:
if input contains additional 6 letters, the 7th-12th letters can be:
-d 1)
For example: --bc ______010 saves photons exiting at plane y=0
-u S
where S is a floating point number, denoting the edge length, in mm, of a voxel in the volume. For example, if the volume used in the simulation is 0.1x0.1x0.1 mm^3, then, one should use -u 0.1 in the command line.
Only isotropic voxels are currently supported by mcx.
'-U' or '-U 1' to enable (default) or '-U 0' to disable
For fluence, the normalization aims to generate a 'Green's function'. For other qualities, the normalization produces a stable solution by removing the dependency due to total photon numbers, voxel sizes, and time-gate settings. The detailed equations are explained in the MCX paper.
If one sets -U 2, the normalization is only applied to the fluence in the interior voxels, and does not apply to the diffuse reflectance (as negative numbers in the raw output volume) along the air voxels right outside of the non-zero voxels.
-E -1 // let MCX to automatically seed the CPU-RNG using system clock
-E n // n is a large positive integer, set the CPU-RNG's seed to n
-E filename.mch // replay detected photons using the seeds saved in the mch file
Setting a fixed RNG seed is expected to create reproducible results on NVIDIA cards if the thread/block size are kept the same.
-z 0 (default)// assumes the lower-bottom corner of the first voxel as [1 1 1]
-z or -z 1 // assumes the lower-bottom corner of the first voxel as [0 0 0]
All source and detector positions are referenced from the origin, determined by this flag.
-R -2 // this enables full atomic operations in the entire volume (default)
-R n // when n is a positive integer, mcx uses atomic operations in the
// shared memory for a n x n x n sub-cubic domain centered at the source.
-R 0 // disable all atomic operations, data racing may exist, particularly
// near the source.
-R -1 //use crop0/crop1 to determine atomic zone
Using full atomic operations was very slow in very early CUDA devices, but for most later NVIDIA GPUs, the use of atomic operations is as efficient as the non-atomic version.
-k 1 (default) // the time-of-flight of the photon starts at the launch time
-k 0 // the time-of-flight starts when a photon enters the first non-zero voxel.
If one sets "-V 0", all launched photon energy preserves after enters the domain.
-Y n // n is a positive integer, denoting the index of the detectors to be replayed
-Y 0 // all detected photons will be replayed regardless of detector
If -Y is not specified, MCX replays all detected photons; otherwise, MCX only replays the detected photons from the specified detector. See -E for more details.
-P flag is followed by a JSON-formatted string. For example
-P '{"Shapes":[{"ZLayers":[[1,10,1],[11,30,2],[31,60,3]]}]}'
This defines a 3-layer medium: z slices 1-10 is filled with tissue label 1, z slices 11-30 is filled with label 2, and 31-60 is filled with label 3.
The shape definition always starts an array object named 'Shapes'. Each element in the 'Shapes' object defines a primitive object. The supported primitives include Name/Origin/Grid/Sphere/Box/Subgrid/{XYZ}Layers/{XYZ}Slabs/Cylinder/UpperSpace. Most objects have a sub-field 'Tag', specifying the tissue label (index to the property list). The details of the Shapes objects can be found in this link
-j option to dynamically define simulation parameters to overwrite/modify the default settings as specified in the .inp/.json file provided after the -f flag. The -j flag is followed by a JSON-formatted string. The format of the JSON construct is the same as in an MCX JSON input file. For example
-j '{"Optode":{"Source":{"Type":"fourier","Param1":[40,0,0,2]}}}'
This changes the source type, whatever it was defined in the input file, to "fourier" source, and sets the source parameter 1 to [40,0,0,2] while keeping everything else unchanged.
if -f, -j, -P and other parameters (such as -n, -s ...) all present in the command line, the priorities are
all other command line flags > -j > -P > -f
where the settings in -n/-s/-d/... overwrite the settings in -j, which also overwrites the input file in -f
byte: 0-128 tissue labels
short: 0-65535 (max to 4000) tissue labels
integer: integer tissue labels
muamus_float: 2x 32bit floats for mua/mus
mua_float: 1 float per voxel for mua
muamus_half: 2x 16bit float for mua/mus
asgn_byte: 4x byte gray-levels for mua/s/g/n
muamus_short: 2x short gray-levels for mua/s
-e f // where f is a floating point number between 0 and 1.
Use -g to split a simulation containing many time gates into smaller but multiple sequential simulations. Format:
-g n // n is a positive integer
Only use this option when the GPU global memory can not hold the data for all required time gates, which only happens when one trys to simulate a very large domain with very dense time gates (very rare). If one's GPU can only hold the data for n time gates, while one has to simulate a total of N > n time gates, use -g n to split the total simulations into multiple runs: in the first run, MCX will record the results for 1~n time gates, in the second run, MCX will launch a new kernel to simulate photons for 1~2*n time gates, but only records the photon fluence for n+1 ~ 2*n time gates and so on. As one can see, this method introduces overhead in the later simulations. Therefore, it is recommended to avoid.
-a 0 (default) // the input volume is a MATLAB-array
-a 1 // the input volume is a C-array
If an MATLAB array is used, the fastest loop index is the left-most index of an array; while for a C array, it is the right-most index.
$ ../../bin/mcx -L ============================= GPU Infomation ================================ Device 1 of 2: GeForce RTX 2080 Compute Capability: 7.5 Global Memory: 8370061312 B Constant Memory: 65536 B Shared Memory: 49152 B Registers: 65536 Clock Speed: 1.81 GHz Number of SMs: 46 Number of Cores: 2944 Auto-thread: 94208 Auto-block: 64 ============================= GPU Infomation ================================ Device 2 of 2: GeForce GTX 1050 Compute Capability: 6.1 Global Memory: 3165323264 B Constant Memory: 65536 B Shared Memory: 49152 B Registers: 65536 Clock Speed: 1.52 GHz Number of SMs: 6 Number of Cores: 768 Auto-thread: 12288 Auto-block: 64
-t N
where N is a positive integer. The thread number N must be a multiple of 32 - the size of a warp - on all CUDA devices. If N is not a multiple of 32, mcx will round it to the nearest multiple less than N.
To achieve the best efficiency, it is recommended to launch a large number of threads to maximize the utility of the GPU resources; a thread number larger than 10000 is generally sufficient.
You should not manually specify the thread number using this option when you use the autopilot mode (-A).
-T N
where N is a positive integer. The block size N must be a multiple of 32 - the size of a warp - on all CUDA devices. If N is not a multiple of 32, mcx will round it to the nearest multiple less than N.
Because mcx does not need inter-thread communication, a small block size, such as 32 or 64, is generally recommended.
You should not manually specify the block size using this option when you use the autopilot mode (-A), which by default is on.
-A 1
When the autopilot mode is enabled, mcx will compute the 'optimal' thread number and block size using a heuristic algorithm.
-G 1 // use only the first GPU device (device orders based on 'mcx -L' output)
-G n // n is a positive integer, use only the n-th GPU device
-G 1110 // when -G is followed by a string made of only 0s and 1s, it specifies
// a mask for active GPUs, for example, 1110 means to use GPU 1-3
// together while the 4th GPU is not used.
When multiple GPU devices are specified, one need to use the -W/--workload flag to optimally partition the total photons to be simulated. By default, photons will be evenly distributed among multiple GPU devices.
-G)
-W w1,w2,w3,... // w_i is a numerical value, corresponding to the relative portions
// of the workload of the i-th GPU device. The total load
// is proportional to the sum of w_i.
For example, -W 10,20,20 indicates a 1:2:2 workload split between 3 active GPUs.If one needs to simulate 1e6 photons, the 1st GPU will run 2e5 photons, while the 2nd and the 3rd GPUs will run 4e5 photons each.
The -I option lists all available GPU, same as -L, but it also run the actual simulation.
-s session_name // session_name is a string, it can not contain <>:"/|?*
If -s is set, the output fluence file (.mc2), detected photon file (.mch) and the log file (.log) will be named as session_name.{mc2,mch,log}. Otherwise, the input file name following the -f option will be used in the place of session_name.
'-d' or '-d 1' to enable (default) or
'-d 0' to disable
When this option is enabled, a binary file, with a suffix of .mch, will be produced in additional to the '.mc2' output. The .mch file contains the partial path data for all photons enters the apertures of the detectors.
For example, -w dspmxvw asks mcx to save all supported field. If a domain
contains 2 tissue types (#media=2), this results in a 2D floating point array
made of 14 columns and #detected photon rows.
-x to 1 also implies setting '-d' to 1
This is the same as appending letters 'xv' with -w/--savedetflag.
-X.
-M to dump the modified volume data for debugging purposes. The dumped volume is saved in a binary file with Nx x Ny x Nz bytes, each byte contains the tissue label as the input volume, and the highest bit of each byte denotes whether the voxel is next to a detector. If -F jnii or '-F nii' is used, the output file will be a JSON/JData or NIfTI file, respectively.
-m to save the momentum transfer for all detected photons. This is the same as appending letter 'm' with -w/--savedetflag. One can use this output for diffuse correlation spectroscopy (DCS) simulations. This information is stored in the .mch file. One can load the data using loadmch.m and process the saved data using the mcxdcsg1.m script, both matlab scripts can be found under mcx/utils.
-H n // n is a positive integer, signifying mcx to allocate a buffer to hold n detected photons
By default, mcx can save up to 1e6 detected photons. If the detected photons exceed this limit, mcx will show a warning. Users may use the -H option to rerun the simulation and use the number in the warning to reallocate the buffer.
-S to enable or disable saving volumetric fluence distributions (or the requested output specified by the -X flag).
'X' - output time-resolved fluence rate (1/mm^2), i.e. TPSF 'F' - output time-resolved fluence rate integrated in each time-gate, 'E' - energy deposit at each voxel (normalized or unnormalized, depends on -n) 'J' - Jacobian (replay mode), 'P' - scattering event counts at each voxel (replay mode only) 'M' - partial momentum transfer
mc2 - MCX mc2 format (binary 32bit float) (default)
nii - Nifti format (fluence after taking log10())
jnii - JNIfTI format (http://openjdata.org)
bnii - Binary JNIfTI (http://openjdata.org)
hdr - Analyze 7.5 hdr/img format
tx3 - GL texture data for rendering (GL_RGBA32F)
the bnii/jnii formats support compression (-Z) and generate small files load jnii (JSON) and bnii (UBJSON) files using below lightweight libraries in MATLAB or Python:
-F jnii or --dumpjson is used (when saving data to JSON/JNIfTI format)
-f
--dumpjson 2 prints configurations before validation
--dumpjson 3 prints configurations after validation
Use the -h flag to print the built-in help of all supported command line flags
Use the -v flag to print the version of MCX
'-l' or '-l 1' to enable (default) or
'-l 0' to disable
The output file name is 'session_name.log' where 'session_name' is the string specified by the -s flag. If -l is used, no message will be printed in the command line.
mcx -i < input.inp
mcx -f input.inp
can produce the same answer.
1 R debug RNG 2 M store photon trajectory info (saved in a.mctfile, can be loaded withloadmch.m) 4 P print progress bar
combine multiple items by using a string, or add selected numbers together
Starting from version 0.7.9, MCX accepts a JSON-formatted input file in addition to the conventional tMCimg-like input format. JSON (JavaScript Object Notation) is a portable, human-readable and "fat-free" text format to represent complex and hierarchical data. Using the JSON format makes a input file self-explanatory, extensible and easy-to-interface with other applications (like MATLAB).
A sample JSON input file can be found under the examples/quicktest folder. The same file, qtest.json, is also shown below:
{
"Help": {
"[en]": {
"Domain::VolumeFile": "file full path to the volume description file, can be a binary or JSON file",
"Domain::Dim": "dimension of the data array stored in the volume file",
"Domain::OriginType": "similar to --srcfrom0, 1 if the origin is [0 0 0], 0 if it is [1.0,1.0,1.0]",
"Domain::LengthUnit": "define the voxel length in mm, similar to --unitinmm",
"Domain::Media": "the first medium is always assigned to voxels with a value of 0 or outside of
the volume, the second row is for medium type 1, and so on. mua and mus must
be in 1/mm unit",
"Session::Photons": "if -n is not specified in the command line, this defines the total photon number",
"Session::ID": "if -s is not specified in the command line, this defines the output file name stub",
"Forward::T0": "the start time of the simulation, in seconds",
"Forward::T1": "the end time of the simulation, in seconds",
"Forward::Dt": "the width of each time window, in seconds",
"Optode::Source::Pos": "the grid position of the source, can be non-integers, in grid unit",
"Optode::Detector::Pos": "the grid position of a detector, can be non-integers, in grid unit",
"Optode::Source::Dir": "the unitary directional vector of the photon at launch",
"Optode::Source::Type": "source types, must be one of the following:
pencil,isotropic,cone,gaussian,planar,pattern,fourier,arcsine,disk,fourierx,fourierx2d",
"Optode::Source::Param1": "source parameters, 4 floating-point numbers",
"Optode::Source::Param2": "additional source parameters, 4 floating-point numbers"
}
},
"Domain": {
"VolumeFile": "semi60x60x60.bin",
"Dim": [60,60,60],
"OriginType": 1,
"LengthUnit": 1,
"Media": [
{"mua": 0.00, "mus": 0.0, "g": 1.00, "n": 1.0},
{"mua": 0.005,"mus": 1.0, "g": 0.01, "n": 1.0}
]
},
"Session": {
"Photons": 1000000,
"RNGSeed": 29012392,
"ID": "qtest"
},
"Forward": {
"T0": 0.0e+00,
"T1": 5.0e-09,
"Dt": 5.0e-09
},
"Optode": {
"Source": {
"Pos": [29.0, 29.0, 0.0],
"Dir": [0.0, 0.0, 1.0],
"Type": "pencil",
"Param1": [0.0, 0.0, 0.0, 0.0],
"Param2": [0.0, 0.0, 0.0, 0.0]
},
"Detector": [
{
"Pos": [29.0, 19.0, 0.0],
"R": 1.0
},
{
"Pos": [29.0, 39.0, 0.0],
"R": 1.0
},
{
"Pos": [19.0, 29.0, 0.0],
"R": 1.0
},
{
"Pos": [39.0, 29.0, 0.0],
"R": 1.0
}
]
}
}
A JSON input file requiers several root objects, namely "Domain", "Session", "Forward" and "Optode". Other root sections, like "Help", will be ignored. Each object is a data structure providing information indicated by its name. Each object can contain various sub-fields. The orders of the fields in the same level are flexible. For each field, you can always find the equivalent fields in the *.inp input files. For example, The "VolumeFile" field under the "Domain" object is the same as Line#6 in qtest.inp; the "RNGSeed" under "Session" is the same as Line#2; the "Optode.Source.Pos" is the same as the triplet in Line#3; the "Forward.T0" is the same as the first number in Line#5, etc.
An MCX JSON input file must be a valid JSON text file. You can validate your input file by running a JSON validator, for example http://jsonlint.com/ You should always use "" to quote a "name" and separate parallel items by ",".
MCX accepts an alternative form of JSON input, but using it is not recommended. In the alternative format, you can use
"rootobj_name.field_name": valueto represent any parameter directly in the root level. For example
{
"Domain.VolumeFile": "semi60x60x60.json",
"Session.Photons": 10000000,
...
}
You can even mix the alternative format with the standard format. If any input parameter has values in both formats in a single input file, the standard-formatted value has higher priority.
To invoke the JSON-formatted input file in your simulations, you can use the "-f" command line option with MCX, just like using an .inp file. For example:
mcx -A -n 20 -f onecube.json -s onecubejson
The input file must have a ".json" suffix in order for MCX to recognize. If the input information is set in both command line, and input file, the command line value has higher priority (this is the same for .inp input files). For example, when using "-n 20", the value set in "Session"/"Photons" is overwritten to 20; when using "-s onecubejson", the "Session"/"ID" value is modified. If your JSON input file is invalid, MCX will quit and point out where the format is incorrect.
Historically, MCX supports an extended version of the input file format used by tMCimg. The difference is that MCX allows comments in the input file. A typical MCX input file looks like this:
1000000 # total photon, use -n to overwrite in the command line
29012392 # RNG seed, negative to generate
30.0 30.0 0.0 1 # source position (in grid unit), the last num (optional) sets srcfrom0 (-z)
0 0 1 # initial directional vector
0.e+00 1.e-09 1.e-10 # time-gates(s): start, end, step
semi60x60x60.bin # volume ('unsigned char' binary format)
1 60 1 60 # x voxel size in mm (isotropic only), dim, start/end indices
1 60 1 60 # y voxel size, must be same as x, dim, start/end indices
1 60 1 60 # y voxel size, must be same as x, dim, start/end indices
1 # num of media
1.010101 0.01 0.005 1.37 # scat. mus (1/mm), g, mua (1/mm), n
4 1.0 # detector number and default radius (in grid unit)
30.0 20.0 0.0 2.0 # detector 1 position (real numbers in grid unit) and individual radius (optional)
30.0 40.0 0.0 # ..., if individual radius is ignored, MCX will use the default radius
20.0 30.0 0.0 #
40.0 30.0 0.0 #
pencil # source type (optional)
0 0 0 0 # parameters (4 floats) for the selected source
0 0 0 0 # additional source parameters
Note that the scattering coefficient mus=musp/(1-g).
Starting from v0.7.9, MCX can also use a shape description file in the place of the volume file. Using a shape-description file can save you from making a binary .bin volume. A shape file uses more descriptive syntax and can be easily understood and shared with others.
Samples on how to use the shape files are included under the example/shapetest folder.
The sample shape file, shapes.json, is shown below:
{
"MCX_Shape_Command_Help":{
"Shapes::Common Rules": "Shapes is an array object. The Tag field sets the voxel value for each
region; if Tag is missing, use 0. Tag must be smaller than the maximum media number in the
input file.Most parameters are in floating-point (FP). If a parameter is a coordinate, it
assumes the origin is defined at the lowest corner of the first voxel, unless user overwrite
with an Origin object. The default origin of all shapes is initialized by user's --srcfrom0
setting: if srcfrom0=1, the lowest corner of the 1st voxel is [0,0,0]; otherwise, it is [1,1,1]",
"Shapes::Name": "Just for documentation purposes, not parsed in MCX",
"Shapes::Origin": "A floating-point (FP) triplet, set coordinate origin for the subsequent objects",
"Shapes::Grid": "Recreate the background grid with the given dimension (Size) and fill-value (Tag)",
"Shapes::Sphere": "A 3D sphere, centered at C0 with radius R, both have FP values",
"Shapes::Box": "A 3D box, with lower corner O and edge length Size, both have FP values",
"Shapes::SubGrid": "A sub-section of the grid, integer O- and Size-triplet, inclusive of both ends",
"Shapes::XLayers/YLayers/ZLayers": "Layered structures, defined by an array of integer triples:
[start,end,tag]. Ends are inclusive in MATLAB array indices. XLayers are perpendicular to x-axis, and so on",
"Shapes::XSlabs/YSlabs/ZSlabs": "Slab structures, consisted of a list of FP pairs [start,end]
both ends are inclusive in MATLAB array indices, all XSlabs are perpendicular to x-axis, and so on",
"Shapes::Cylinder": "A finite cylinder, defined by the two ends, C0 and C1, along the axis and a radius R",
"Shapes::UpperSpace": "A semi-space defined by inequality A*x+B*y+C*z>D, Coef is required, but not Equ"
},
"Shapes": [
{"Name": "Test"},
{"Origin": [0,0,0]},
{"Grid": {"Tag":1, "Size":[40,60,50]}},
{"Sphere": {"Tag":2, "O":[30,30,30],"R":20}},
{"Box": {"Tag":0, "O":[10,10,10],"Size":[10,10,10]}},
{"Subgrid": {"Tag":1, "O":[13,13,13],"Size":[5,5,5]}},
{"UpperSpace":{"Tag":3,"Coef":[1,-1,0,0],"Equ":"A*x+B*y+C*z>D"}},
{"XSlabs": {"Tag":4, "Bound":[[5,15],[35,40]]}},
{"Cylinder": {"Tag":2, "C0": [0.0,0.0,0.0], "C1": [15.0,8.0,10.0], "R": 4.0}},
{"ZLayers": [[1,10,1],[11,30,2],[31,50,3]]}
]
}
A shape file must contain a "Shapes" object in the root level. Other root-level fields are ignored. The "Shapes" object is a JSON array, with each element representing a 3D object or setting. The object-class commands include "Grid", "Sphere", "Box" etc. Each of these object include a number of sub-fields to specify the parameters of the object. For example, the "Sphere" object has 3 subfields, "O", "R" and "Tag". Field "O" has a value of 1x3 array, representing the center of the sphere; "R" is a scalar for the radius; "Tag" is the voxel values. The most useful command is "[XYZ]Layers". It contains a series of integer triplets, specifying the starting index, ending index and voxel value of a layered structure. If multiple objects are included, the subsequent objects always overwrite the overlapping regions covered by the previous objects.
There are a few ways for you to use shape description records in your MCX simulations. You can save it to a JSON shape file, and put the file name in Line#6 of yoru .inp file, or set as the value for Domain.VolumeFile field in a .json input file. In these cases, a shape file must have a suffix of .json.
You can also merge the Shapes section with a .json input file by simply appending the Shapes section to the root-level object. You can find an example, jsonshape_allinone.json, under examples/shapetest. In this case, you no longer need to define the "VolumeFile" field in the input.
Another way to use Shapes is to specify it using the -P (or --shapes) command line flag. For example:
mcx -f input.json -P '{"Shapes":[{"ZLayers":[[1,10,1],[11,30,2],[31,60,3]]}]}'
This will first initialize a volume based on the settings in the input .json file, and then rasterize new objects to the domain and overwrite regions that are overlapping.
For both JSON-formatted input and shape files, you can use the JSONlab toolbox [4] to load and process in MATLAB.