As can be seen from visualtest_2d.py (as of b7729a1) , the histogram is prone to rather large errors depending on what side of the bin divider a group of agents falls. Specifically, if a grid of agents covers the domain, sometimes a histogram bin will have one more row than other histogram bins. Plotting a smoothed density in place of the histogram would be far more accurate and more useful in practice.
As can be seen from visualtest_2d.py (as of b7729a1) , the histogram is prone to rather large errors depending on what side of the bin divider a group of agents falls. Specifically, if a grid of agents covers the domain, sometimes a histogram bin will have one more row than other histogram bins. Plotting a smoothed density in place of the histogram would be far more accurate and more useful in practice.