Often times on my way home, I find myself walking North on Stone Way, attempting to catch a Westbound bus #44 on 45th St. at Stone Wy. I can see the buses go past from afar on Stone Way, and when I see a bus go past, I wonder what the implications are for my wait time at the stop. When I'm far enough away, I want a bus to go past already, so that the next bus might show up just as I get to the stop. But as I get nearer the stop, I don't want a bus to go past, as that would mean a longer wait.

The worst scenario is a bus whizzing past just before I get to the stop, while the best is a bus showing up just as I do. So there's a discontinuity there, and that made me want to model the problem, and help figure out how I should adjust my walking speed to synchronise my arrival and a bus being there.

Let

So it takes me `l/v` time to get to the stop, and `l/v mod (1/nu)` time since the last bus passed by. Thus, it's `(-l/v) mod (1/nu)` time for the next bus. I thus want to minimise (the non-negative quantity) `(-l/v) mod (1/nu)`.

Let: `-l/v mod (1/nu) = epsilon_0`, for some `epsilon_0 >= 0`

`=> -l/v = -k/nu + epsilon_0`, for some `k in ZZ_(>=0)`

`=> v = -l/(-k/nu + epsilon_0) = l/(k/nu - epsilon_0) = (l nu)/k + epsilon_1`, for some `epsilon_1 >= 0`

Thus, I should pick speeds just greater than `(l nu)/k`, for some (positive) integral `k`. Let's say I don't want to go any faster than `v_0`, but I don't want to get home soon. I'd then pick `k` such that `(l nu)/k` is just less than `v_0`, i.e. minimising the positive quantity `v_0 - (l nu)/k`.

Let's let: `v_0 - (l nu)/k = v_delta`, with `v_delta > 0`

`=> k = (l nu)/(v_0 - v_delta) = ceil((l nu)/v_0)`.

Therefore, the speed I should target, that would

The worst scenario is a bus whizzing past just before I get to the stop, while the best is a bus showing up just as I do. So there's a discontinuity there, and that made me want to model the problem, and help figure out how I should adjust my walking speed to synchronise my arrival and a bus being there.

Let

- `l` be the distance from the stop when I sight a bus go past,
- `nu` the frequency of Westbound buses along 45th St.,
- `v_0` my current speed, and
- `v` my new (for simplicity) constant speed once I see the bus.

So it takes me `l/v` time to get to the stop, and `l/v mod (1/nu)` time since the last bus passed by. Thus, it's `(-l/v) mod (1/nu)` time for the next bus. I thus want to minimise (the non-negative quantity) `(-l/v) mod (1/nu)`.

Let: `-l/v mod (1/nu) = epsilon_0`, for some `epsilon_0 >= 0`

`=> -l/v = -k/nu + epsilon_0`, for some `k in ZZ_(>=0)`

`=> v = -l/(-k/nu + epsilon_0) = l/(k/nu - epsilon_0) = (l nu)/k + epsilon_1`, for some `epsilon_1 >= 0`

Thus, I should pick speeds just greater than `(l nu)/k`, for some (positive) integral `k`. Let's say I don't want to go any faster than `v_0`, but I don't want to get home soon. I'd then pick `k` such that `(l nu)/k` is just less than `v_0`, i.e. minimising the positive quantity `v_0 - (l nu)/k`.

Let's let: `v_0 - (l nu)/k = v_delta`, with `v_delta > 0`

`=> k = (l nu)/(v_0 - v_delta) = ceil((l nu)/v_0)`.

Therefore, the speed I should target, that would

- minimise my wait time at the stop
- minimise the overall time
- not exceed my current walking speed