Default Parameter List

This parameter list contains the default mean parameters under slot params_archetypes, their standard deviations under slot params_sigma, and their bounds under params_bounds. These parameters automatically come with the package.

Usage

params_from_csv

Format

An object of class list of length 3.

Details

# Content of the slots

Values under slots params_archetypes and params_sigma contain some variation as imposed through what we call the "archetypes", which for params_archetypes can be found in the column name and for params_sigma in the names of this list. These archetypes represent parameter sets that have been created to display a given type of behavior, such as rushing to get to the goals ("Rushed") or making very random moment-to-moment decisions ("DrunkAussie"). These archetypes thus represent a part of the individual variability that predped allows.

Another aspect of this variability is controlled by the values under params_sigma. For each of the archetypes a covariance matrix is defined in the list params_sigma that allows for variation around the values found in params_archetypes. However, do not mistake the matrices in params_sigma to be covariance matrices: Instead, these matrices have standard deviations on the diagonal and correlations between the parameters on the off-diagonal, allowing for users to more intuitively set up these matrices themselves. Under the hood, the covariance matrix COV is computed through the provided matrix X by defining:

\(SD = diag(X) ,\)

and by creating the matrix COR which consists of X with its diagonal turned to 1. We can then compute the covariance matrix by multplying both matrices:

\(COV = COR * SD * SD^T .\)

Importantly, the standard deviation and correlations should be defined between each of the parameters. We furthermore note that the standard deviations can also be equal to 0, allowing no variation in the selected parameters.

# Parameters

Each of the parameters in params_archetypes controls an aspect of the decisions pedestrians make when walking around in an environment, namely:

radius:

the radius of the agent

slowing_time:

the number of seconds the agent needs to slow down when approaching a goal

preferred_speed:

the speed at which the agent is comfortable walking

randomness:

the temperature parameter that controls the overall unpredictability of the nex decision an agent will make. Larger values make movement more deterministic (i.e., strongly determined by the utility functions).

stop_utility:

utility value of stopping instead of moving

reroute:

number of pedestrians that should be in the way for an agent to consider rerouting 50% of the time

b_turning:

slope that scales the effect of turning on the speed an agent can maintain, where 1 - b_turning denotes the maximal decrease in velocity in percent

a_turning:

exponent that determines the shape of the effect of turning on the speed an agent can maintain

b_current_direction:

slope that scales the utility of continuing waking in the current direction

a_current_direction:

exponent that determines the power to which the difference of not walking in the current direction is taken

blr_current_direction:

scales the preference for walking to the left or right when heading in a given direction. Done in such a way that b_current_direction defines the slope for the left side and is divided by blr_current_direction for the right side, meaning that the slope for the right side increases when blr_current_direction < 1 and decreases when blr_current_direction > 1

b_goal_direction:

slope that scales the utility of heading in the direction of your current goal

a_goal_direction:

exponent that determines the power to which the utility of not heading towards the current goal is taken

b_blocked:

slope that scales the extent to which agents will avoid directions that in the long run will lead to blockage (e.g., because of other agents)

a_blocked:

exponent that determines the power of the function that determines the utility for avoiding directions that will lead to blockage

b_interpersonal:

slope that scales the steepness of the utility for keeping an interpersonal distance

a_interpersonal:

exponent that determines the power to which the interpersonal distance is taken. Note that – in contrast to the other exponents – the exponent here concerns the exponent of a hyperbolic function, rather than a power function.

d_interpersonal:

increment added to b_interpersonal when pedestrians close to the agent are of a different social group, effectively increasing the interpersonal distance

b_preferred_speed:

slope that scales the effect of trying to walk at your preferred speed

a_preferred_speed:

exponent that determines the power to which the difference of not walking at your preferred speed is taken

b_leader:

slope that scales the effect of selecting and following in a leader's footsteps

a_leader:

exponent that determines the power of the function for the follow-the-leader effect

d_leader:

increment added to b_interpersonal when the leader is of the same social group as the agent, effectively increasing the tendency for the agent to follow this leader

b_buddy:

slope that scales the effect of selecting and walking besides a buddy

b_group_centroid:

slope that scales the effect of trying to maintain a small distance between an agent and their group members

a_group_centroid:

exponent that determines the power with which the distance of the agent to their group members is taken

b_visual_field:

slope that scales the effect of maintaining group members within the visual field

See also

predped-class, generate_parameters, load_parameters, utility-agent utility-data.frame