Views

Accessing data and insights from simulation runs

HASH lets users access their simulation outputs in a variety of ways. These include custom viewers for the 3D rendering, geospatial display, and charting/plotting of outputted simulation state.

Using the playbar at the bottom of the right-hand view pane you can scrub forwards and backwards in time through any existing simulation state.

3D

Support for a full range of 3D models in Core's 3D viewer is coming soon. In the meantime, the following primitives can be used to give form to agents: box, cone, cylinder, dodecahedron , icosahedron, octahedron, plane, sphere, tetrahedron, torus, torusknot

Geospatial

Agents can be rendered on a map-view of the world by providing them with latlong co-ordinates. Various examples in the HASH Index showcase this (e.g. Local Competition)

Plots

Charts and plots are in many cases the best way to view simulation outputs in HASH. When you're not interested in the behavior of an individual agent, but aggregate trends over time or outcomes observed across the entire system, graphing variables of interest can enable you to quickly identify and analyze outcomes.

Raw Output

The underlying JSON of any individual step is accessible from the 'Raw Output' tab. This can be useful for debugging models during development, and allows for the raw state of simulations to be exported for further analysis.

Step Explorer

Dive into the statistics and distribution of any single step in the simulation. Select, filter, and visualize all the data of all the agents in your simulation with just a few clicks.

The Step Explorer has easy to use auto-generated charts to probe and understand what's happening with your agents on a given step. Unit visualizations are used which apply a one-to-one mapping between agent properties and the data points in the charts.

You can think of the Step Explorer as a counterpart to Plots, which tells you what's happening in aggregate across all of the time steps in a simulation run, while the Step Explorer lets you dive into a single time step and see what's happening within a population of agents in detail.

Step Explorer is useful for understanding the shape of the data within your simulation, and it's also helpful for debugging and noticing agent outliers, without having to write code or JSON.