The modes we are studying have a small overall percent of the population who use them. This creates challenges when running a scenario with less than 100% of the population. We discussed a few ways forward on today's call:
- Throw a massive computer at the problem. @zneedell suggested if we had a TB of memory on a node, it would probably work. BYU has a computer with a 2TB node, and Tech has one with 800 GB.
- Come up with a sampling strategy that would model die-hard car users as background traffic, and only simulate the behavior of the population that might use ridehail.
The modes we are studying have a small overall percent of the population who use them. This creates challenges when running a scenario with less than 100% of the population. We discussed a few ways forward on today's call: