Could you please tell us where the code in Appendix C of the paper is for using a quasi-random sampling sequence to reduce the variance of the rendering equation?

[yougushidedashan]

To further reduce the variance of the estimated value ofthe rendering equation (see Equation 15), we use quasirandom sampling sequences. Specifically, we use a Sobolsequence [36] with Owens scrambling [30], which gives theprocedural sequence necessary for assigning an arbitrary number of secondary ray samples to each primary ray sample. We then apply Cranley-Patterson rotation [12] to avoidneeding to redraw samples.

This part shows up where in your code I didn't find it

[half-potato]

https://github.com/half-potato/nmf/blob/7c79de00d111b5fdc0a6d0af2a9c00ee63bacb0f/brdf_samplers/base.py#L1