Lambertian/arbitrary radiation pattern source for MMC

[Tigrane-CM]

Hello,

I want to simulate light penetration in the brain using MMC, with off-the-shelf LED components.

I've been looking at how to define a Lambertian source in MMC. I would have expected a planar source with -infinity focus to handle it, as it seems to do in mcx (from line 1571 of mcx_core.cu) but it doesn't seem to be handled in MMC. is that correct ?

Is there an easy way to fix it, and/or could it be possible to define a radiation pattern function to better simulate LED emitters, that we could directly infer from the datasheets ?

Thanks for your help,

Tigrane

[fangq]

hi @Tigrane-CM, sorry, meant to do this sooner.

the focal length feature is now supported in mmc/mmclab. same as mcx, just set cfg.srcdir(4), -inf for Lambertian launch, nan for isotropic launch. only works for area/line sources.

[Tigrane-CM]

Hi Dr Fang,

thank you for your answer. I've updated my setup to pmmc v 0.3.6 and tried using this parameter, but I get errors:

• if I use srcdir=[0,0,1,-inf], python tells me that inf is not defined, of course.
• if I go through numpy with srcdir=[0,0,1,-np.inf], I get

RuntimeError: PMMC terminated due to an exception!Error from thread (0): Error: Failed to build program executable!

and in more detail,

:1377:19: error: use of undeclared identifier 'inff'
if (isnan(GPU_PARAM(gcfg, focus))) { // if beam focus is set, determine the incident angle
^
:368:29: note: expanded from macro 'GPU_PARAM'
#define GPU_PARAM(a,b) (a ## b)
^
:61:1: note: expanded from here
gcfgfocus
^
:14:20: note: expanded from here
#define gcfgfocus -inff

Could you please provide a little help with this ?
I also note that it's not always easy to understand the keywords and options that are available in the cfg dict or as arguuments to pmmc.run, as they don not always correspond to the ones used in the CLI. Is there anywhere a reference for these ?

Thanks again,
Tigrane

[fangq]

@Tigrane-CM, this is because of a new flag enabled in the v2025.10 release. to be able to use inf as the focal length, you need to set cfg['optlevel'] to 1. If set to 3 or above, you will see this error.

[fangq]

@Tigrane-CM, the above issue was now fixed in

e55aa08

pmmc is also updated to 0.3.7 to allow srcdir(4) (focal-length) to be nan or -inf.