Celeritas is a new Monte Carlo transport code designed to accelerate scientific discovery in high energy physics by improving detector simulation throughput and energy efficiency using GPUs.
The BIH cost function fails in the presence of semi-infinite objects, which can happen given certain shapes. In theory, though, if the global boundary is finite, every bounding box should be finite. In fact we should be able to use this to determine strict bounding boxes on general combinatorial quadric surfaces using Millman 2014[^1].
With this change:
Global boundaries are validated to be finite (e.g. you can't use a negated sphere or a shape that doesn't yet have a strict bbox implementation)
Daughter universe bounding boxes start off as null but are expanded by their placement in parent volumes (using a parent-to-daughter transformation of the bounding box)
Objects being constructed take an "initial" bounding box (the unit's boundary) so that they're no larger than that box
[^1]: Millman, David L., David P. Griesheimer, Brian R. Nease, and Jack Snoeyink. “Computing Numerically-Optimal Bounding Boxes for Constructive Solid Geometry (CSG) Components in Monte Carlo Particle Transport Calculations.” In SNA + MC 2013 - Joint International Conference on Supercomputing in Nuclear Applications + Monte Carlo, edited by D. Caruge, C. Calvin, C.M. Diop, F. Malvagi, and J.-C. Trama, 02506. Paris, France: EDP Sciences, 2014. https://doi.org/10.1051/snamc/201402506.
The BIH cost function fails in the presence of semi-infinite objects, which can happen given certain shapes. In theory, though, if the global boundary is finite, every bounding box should be finite. In fact we should be able to use this to determine strict bounding boxes on general combinatorial quadric surfaces using Millman 2014[^1].
With this change:
[^1]: Millman, David L., David P. Griesheimer, Brian R. Nease, and Jack Snoeyink. “Computing Numerically-Optimal Bounding Boxes for Constructive Solid Geometry (CSG) Components in Monte Carlo Particle Transport Calculations.” In SNA + MC 2013 - Joint International Conference on Supercomputing in Nuclear Applications + Monte Carlo, edited by D. Caruge, C. Calvin, C.M. Diop, F. Malvagi, and J.-C. Trama, 02506. Paris, France: EDP Sciences, 2014. https://doi.org/10.1051/snamc/201402506.