search

CityScope / CS_Proxymix

ABM framework to study spatial configuration and human behaviour at the building scale
10 stars 7 forks source link

Problem with distance calculation #179

Closed agrignard closed 3 years ago

agrignard commented 3 years ago

Right now it seems that the at_distance operator behave weirdly.

If we use ask (ViralPeople at_distance largeDropletRange) we don't have results with this

ask ViralPeople where ((each.location.x-self.location.x)^2+(each.location.y-self.location.y)^2<largeDropletRange^2

Is it because the ViralPeople has a shape<-circle(1).

ptaillandier commented 3 years ago

You are right, It is because at_distance take into account the geometries of the agent and not only their location. In addition, with ask ViralPeople where ((each.location.x-self.location.x)^2+(each.location.y-self.location.y)^2<largeDropletRange^2, you consider the current agent as well (at_distance never returns the agent using this operator).

agrignard commented 3 years ago

What I don't get is that if I comment geometry shape<-circle(1); in ViralPeople then the distance should be between the centroid no?

We don't have an operator in GAMA that is computing the distance between location?

What's wrong with this formula?

ask ViralPeople where ((each.location.x-self.location.x)^2+(each.location.y-self.location.y)^2<largeDropletRange^2
ptaillandier commented 3 years ago

What I don't get is that if I comment geometry shape<-circle(1); in ViralPeople then the distance should be between the centroid no? yes.

We don't have an operator in GAMA that is computing the distance between location? You can always compute the distance between the centroid of the agent: agent1.location distance_to agent2.location. the at_distance operator always use the geometries of the agents and it is not possible to directly use the centroid.

What's wrong with this formula? Nothing. You can just write : ask ViralPeople where ((each.location distance_to location)<largeDropletRange) and you will get the same result. However, you should remove from the list the ViralPeople agent applying this statement to avoid autocontamination :p .

agrignard commented 3 years ago

When I comment geometry shape<-circle(1); the distance_to is still giving a wrong distance (0.82 where as it should give 2m)

agrignard commented 3 years ago

won't fix for now