Program design: beehive simulation

Learn this with friends

ComputingIntro to computer science - Python5:545 check-ins

About this video

The video designs a simulation of how much honey a beehive makes in a week. Bees are modelled as a state machine moving between needs_nectar and has_nectar, with random numbers of pickups and drop-offs each hour inside a for loop, and a subtle bug is fixed by updating the states only after both counts are calculated. Total nectar deposits divided by 90 gives the grams of honey.

You'll be able to

  • explain how a simulation simplifies a real-world process using states and randomness.
  • decide which variables to initialise before a loop and which updates belong inside it.
  • spot a bug caused by updating variables too early inside a loop iteration.

Key words

simulation
A program that models a real-world process to answer a question about it.
state machine
A model where things move between a fixed set of states, like a bee needing nectar or having nectar.
state transition
A move from one state to another, such as a bee picking up nectar.
simplification
An assumption that leaves out some real-world detail to make a model easier to build.
Transcript

- [Instructor] Let's design a simulation with for loops. We wanna answer the question, how much honey does a beehive produce over a certain period of time? Now there are a lot of variables that might impact honey production like the geography, the weather, what flowers are nearby.

These are all important factors to consider in designing our simulation. But we need to think about what simplifications we can reasonably make. Based on observation, we know that a typical forager bee will make several trips to collect nectar over the course of a day.

I'm thinking the easiest way to model this might be to use a state machine, where a bee moves between different states. There's a state where they've left the hive and are collecting nectar, and there's a state where they have nectar and they're returning to the hive to drop it off.

So say maybe every hour a bee can optionally move from the needs nectar state to the has nectar state if it's picked up from enough flowers, or it can move from the has nectar state to the needs nectar state if it's dropped off at the hive. Our simulation then can just repeat these state transitions over some number of hours to calculate the total amount of nectar deposited.

We can then use that nectar amount to calculate how much honey that typically produces. To account for some environmental factors, we can use randomness to decide how many bees move from one state to the other each hour. To start, let's decide what variables we'll need.

We'll wanna keep track of the total number of bees in each state. So we'll want a needs_nectar and a has_nectar. And then we also wanna count the total number of nectar deposits that we've made, so we can calculate the amount of honey at the end. All of our bees are gonna start out in the needs nectar state.

They're waiting in the hive ready to go. None of them have nectar yet. We'll initialize that to 20,000 'cause that's approximately how many forger bees we have. Before we add the loop, let's just try and get one hour of collection working. Each hour, a random number of bees in the needs nectar state are going to finish collecting nectar.

Sounds like we're gonna need the random module. So I'm gonna add that import statement at the top. The number of bees that pick up each hour is influenced by a lot of factors, but we're just gonna simplify that down and say, it's somewhere between one quarter and three quarters of the bees in the needs_nectar state.

Alright, now we need to handle the state transitions. Any these that pick up move from the needs nectar state to the has nectar state. So we need to update both of those variables. Has nectar increases by the number of pickups and needs nectar decreases by the number of pickups.

We're not adding or removing bees here, so the total across both states should still equal 20,000. Let's add some print statements to check that and run it a few times. Okay, let's add our loop now to repeat this hour of activity. We wanna calculate how much honey we can produce in a week so it makes sense to use a for loop here since we already know the number of repetitions.

We'll name our loop variable hour since each iteration represents an hour and we wanna repeat for seven times 24 hours. That's not quite accurate 'cause bees do sleep. They're not gonna be collecting for all 24 hours. So let's change that to seven times 18. We wanna initialize all our variables before the loop, so we'll leave them up here, but we do want to repeat the state transition on every loop iteration.

So we're gonna end in all of these lines inside the loop. Now when we run our simulation, we see that all of our bees move into the has nectar state. They all eventually complete a pickup. And now we just need to model the drop off so the bees can move back from the has nectar state to the needs nectar state.

Each hour a random number of bees who have nectar drop off that nectar at the hive. We're gonna say it's at the same rate so between a quarter and three quarters of the bees in the has nectar state. Every drop off we make is a permanent nectar deposit that other bees in the hive can then use to produce honey.

So every hour we wanna increment nectar deposits by the number of drop-offs. Okay? But when I run it, all of my bees are still stuck in the has nectar state. We still need to transition the bees from the has nectar state to the needs nectar state when they drop off.

So has nectar should decrease by the number of drop-offs and needs nectar should increase by the number of drop-offs. But I think there's kind of a subtle bug here. What do you think? At the top of the hour, we let some bees drop off, which means they move back to the needs nectar state, and then at the bottom of the hour, we let some bees in the needs nectar state pick up.

That means the same bee could both drop off and pick up within the same hour, which doesn't seem realistic. To fix this, we need to make sure we're not updating the number of bees in each state until after we've calculated the number of Dropbox and the number of pickups because those should be calculated based on the number of bees in each state at the start of the hour.

So I'm just gonna move all of the variable updates to the bottom of the loop. Awesome. Our state transitions are looking good, so let's check in on our nectar deposits. It looks like they're steadily increasing by a random amount on each iteration, which is what we'd expect.

It should be increasing by the number of drop-offs. So if that total's looking good, it's time to calculate how much honey we produced. We only wanna calculate this once at the end of the week, so we don't wanna put this inside of the loop body. We want it to go after the loop indented outside of it.

From our research, it seems like it takes a single bee about 90 trips to make a single gram of honey. So we're gonna divide our total number of nectar deposits by 90, which would give us approximately the number of grams of honey. And I'm gonna round that so we don't get a weird decimal.

And if we run our simulation now a few times, we have our answer. Our hive can produce around 7,000 grams of honey a week. This simulation, of course, made a lot of assumptions. There's a lot more things we can model here with more time in research. So as with most simulations, we should think critically about the results that we get.

Video © Khan Academy, used under CC BY-NC-SA. Watch on Khan Academy. Puzzicle is not affiliated with Khan Academy.