There are a couple issues that would be nice to address:
time complexity of the algorithms #26
scalability #139
I'm thinking that the package needs a rework. I lent very heavily into making it a feasible teaching tool and ended up with some very difficult innards...
Is there a way to restate the problems (and algorithms) as matrices? My instinct is yes.
If there is, we can use vectorised operations to speed things up (everything becomes index comparisons, I think). We could also expand these matrix-based solutions to use something like dask for enormous, out-of-memory instances...
There are a couple issues that would be nice to address:
I'm thinking that the package needs a rework. I lent very heavily into making it a feasible teaching tool and ended up with some very difficult innards...
Is there a way to restate the problems (and algorithms) as matrices? My instinct is yes.
If there is, we can use vectorised operations to speed things up (everything becomes index comparisons, I think). We could also expand these matrix-based solutions to use something like
dask
for enormous, out-of-memory instances...