Issue on running MCX

[Q W]

Dear Professor Fang,

I hope you are doing well.

 I am currently working on diffuse correlation spectroscopy （DCS）and running simulations using MMC. I have adjusted the MATLAB script several times, but it still fails. For example, 

%% **************************** Quan Adding**************

% MMC executable full path

mmc_exe = fullfile(mmc_path, 'bin', 'mmc.exe');

% JSON file full path (the one just written)

json_file = [scatterBrains_path filesep 'Subject' subject_num filesep 'example.json'];

% Check that mmc.exe exists

if ~exist(mmc_exe, 'file')

error('MMC executable not found: %s', mmc_exe);

end

% Check that example.json exists

if ~exist(json_file, 'file')

error('JSON file not found: %s', json_file);

end

% Build command (use double quotes for paths with spaces)

cmd = sprintf('"%s" -f "%s" -n 1e8 -D P -b 1 -d 1 --saveref 1 --momentum 1', mmc_exe, json_file);

fprintf('Running MMC:\n%s\n', cmd);

% Execute MMC

[status, result] = system(cmd);

if status ~= 0

error('MMC failed with status %d:\n%s', status, result);

else

fprintf('MMC completed successfully.\n');

disp(result);

End

It always gives me a Error:

Error using example

MMC failed with status 1:

Mesh error: can not open node file in unit mmc_mesh.c line#447

Do you have any suggestions for this?

Looking forward to your reply.

Best,

Quan

[Qianqian Fang]

hi Quan, the error message is quite clear - it could not find the mesh files. you must create those mesh files and pass on the filename stub (the ??? part of node_???.dat and elem_???.dat) to the input json file.

if you go to mmc/examples, all provided examples include externally stored node/elem data files. you should follow those examples if you have to use mmc.

however, since you use matlab, I strongly encourage you using mmclab instead of mmc - mmclab directly passes on matlab arrays/structs to mmc and saves all the overheads of reading/writing data files - file IO is the slowest part of modern computers.

--
You received this message because you are subscribed to the Google Groups "mcx-users" group.
To unsubscribe from this group and stop receiving emails from it, send an email to mcx-users+...@googlegroups.com.
To view this discussion visit https://groups.google.com/d/msgid/mcx-users/CAOvA73QZ%2BVxcAtzF%2BAS2Zz87XYYCciK0CZXHOQpQa_5HJQh2mA%40mail.gmail.com.

[Q W]

Thanks, Professor Fang,

I found an example that loads Colin27. In my application, I will be loading my own target head model. So, do I just need to load the Subject03_volume.mat file then ignore the XXmesh.mat and XXXloc.mat? 

Another question: in the following example, I am curious about how you can define the srcpos so precisely, e.g., 67.38, 167.5?

Also, for the detpos, what does the '3' mean?

In my application, for example, I only know that the source-detector separations are 10, 25, and 30 mm. I cannot just do simple math calculation for the detector position like in the layered head model.

%%

cfg.tstart = 0;

cfg.tend = 5e-09;

cfg.tstep = 2e-10;

cfg.srcpos = [75 67.38 167.5];

cfg.srcdir = [0.1636 0.4569 -0.8743];

cfg.srcdir = cfg.srcdir / norm(cfg.srcdir);

cfg.detpos = [75.0000 77.1900 170.3000 3.0000

75.0000 89.0000 171.6000 3.0000

75.0000 97.6700 172.4000 3.0000

75.0000 102.4000 172.0000 3.0000];

Best,

Looking forward to your reply

Quan

Qianqian Fang <q.f...@neu.edu> 于2026年4月15日周三 02:49写道：

[Q W]

Thanks, Professor Fang, 

Just to follow up on the previous question.

I found an example that loads Colin27. In my application, I will be loading my own target head model. So, do I just need to load the Subject03_volume.mat file then ignore the XXmesh.mat and XXXloc.mat? 

Another question: in the following example, I am curious about how you can define the srcpos so precisely, e.g., 67.38, 167.5 ( Definitely, we should consider the curvature of the human head model.)?

Also, for the detpos, what does the '3' mean?

In my application, for example, I only know that the source-detector separations are 10, 25, and 30 mm. I cannot just do simple math calculation for the detector position like in the layered head model.

%%

cfg.tstart = 0;

cfg.tend = 5e-09;

cfg.tstep = 2e-10;

cfg.srcpos = [75 67.38 167.5];

cfg.srcdir = [0.1636 0.4569 -0.8743];

cfg.srcdir = cfg.srcdir / norm(cfg.srcdir);

cfg.detpos = [75.0000 77.1900 170.3000 3.0000

75.0000 89.0000 171.6000 3.0000

75.0000 97.6700 172.4000 3.0000

75.0000 102.4000 172.0000 3.0000];

Best,

Looking forward to your reply

Quan

Qianqian Fang <q.f...@neu.edu> 于2026年4月15日周三 02:49写道：

hi Quan, the error message is quite clear - it could not find the mesh files. you must create those mesh files and pass on the filename stub (the ??? part of node_???.dat and elem_???.dat) to the input json file.

[Qianqian Fang]

hi Quan,

your question needs better clarity.

those *.mat files were not part of mmc - you must have created those yourself. I don't know what's in those files, so I can't answer your question how to handle those files.

mmc does not know how to read .mat file. it only cares the {node,elem,prop,facenb}_XX.dat files that were generated by savemmcmesh.m. whatever data you have, you need to pass those to savemmcmesh and generate those files that mmc can use.

Another question: in the following example, I am curious about how you can define the srcpos so precisely, e.g., 67.38, 167.5 ( Definitely, we should consider the curvature of the human head model.)?

why not? please read how mmc/mcx's detector works in this FAQ

https://mcx.space/wiki/index.cgi?Doc/FAQ#How_does_MCX_capture_detected_photons

Also, for the detpos, what does the '3' mean?

detector radius. 

https://github.com/fangq/mmc/blob/v2025.10/mmclab/mmclab.m#L89

and also the above FAQ

Qianqian

To view this discussion visit https://groups.google.com/d/msgid/mcx-users/CAOvA73S0d6hzeXjNW39_H6hxzFE9Xd4zBF0MZ0%3DUqfC-xQdOig%40mail.gmail.com.