AgentSheets 4.0 Release Notes


Transition to Java Programming - Gradually...

How do you transition from visual programming to more traditional forms of programming such as Java in a way that’s relevant to high school or undergraduate non-major courses? AgentSheets 4 shows you the Java source code for your project. See how your agents turn into Java classes and learn how your rules are transformed into Java IF THEN statements.


Compute Three Dimensionally

AgentSheets always allowed you to pile agents up in stacks, but it was difficult to compute across these stacks. We’ve extended our Visual AgenTalk programming language to support 3D access of agent attributes across agents on stacks. Access values above, below, at the top, and at the bottom of stacks. Use this feature to build exciting three- dimensional simulations.





Track your Agents for Data Analysis

If you build simulations like ecosystems where you need to keep track of your agents, there’s a good chance that you need to count agents. New extensions to our Visual AgenTalk formula language let you easily and efficiently determine the number of agents of a certain type or with a certain depiction.


Smarten up your Artificial Intelligence

With the new HillClimb action you not only build Artificial Intelligence such as Collaborative Diffusion much more quickly, you’re also able to build smarter AI. Create games such as Pacman where agents are first attracted to each other like ghosts to the Pacman, and then are repelled (when the Pacman eats the powerpill). For pathfinding, select from working with Von Neuman (4 adjacent neighbors) and Moore (8 neighbors) neighborhoods.



Interact with your Data in real Time

Instantly turn any simulation property into a slide-controllable real time interface to your simulation. Set the bounds and enable the slider by ticking a checkbox. Done.

General Improvements

We have fixed reported bugs, and made AgentSheets 4 faster and easier to use. We also made AgentSheets simpler to install on the Mac by using new installers and code signing.


This material is based in part upon work supported by the National Science Foundation under grant numbers 0833612 and DMI-0712571. Any opinions, findings, and conclusions or recommendations expressed in this material are those of the authors and do not necessarily reflect the views of the National Science Foundation.