Since Davis is intended to be used with many simultaneous windows, several optimizations are applied to the views. For example, this view keeps track of the previous color of each element and does not redraw a pixel unless it is in the foreground and the color changes. The visualization also uses a settable quantization to reduce the number of possible colors and therefore also the number of pixels that are redrawn on each frame. These simple optimizations reduce the computational load by a factor of 10 or more for some of the neural data tested with Davis. The display uses tool tips to identify individual elements and their current values as the user moves the cursor over the figure.

The three KL phase portraits shown in this graphic give low-dimensional caricatures that reveal the structure and timing of the response for different types of neurons. The red dot in each KL window specifies the position of the current frame in the caricature. In each picture, the origin corresponds to t = 0, since the first frame was subtracted from all of the other frames before doing the KL decomposition. The current frame may be interpolated if the timer is not running at a rate commensurate with the sampling frequency of the measurement group. The upper right KL window depicts the soma voltage for the lateral neurons, the lower left KL window depicts the soma voltage for medial neurons, and the lower right KL window depicts the soma voltage for the horizontal cells.

Notice that at 97 ms, the response for the strengthened AMPA is much more developed than that for a model with the normal AMPA conductances. The KL decomposition does not show the linear growth followed by a sharp change in direction. Also, the response shows no sign of damping, but rather, it oscillates for the full 1.5 seconds encompassed by the run. The response for a system with AMPA coupling that is twice as strong is qualitatively similar to that shown for multiplication by 4. When AMPA is halved, the response is much less vigorous, but the wave does reach the edge of the brain before dying away.

Environments are available through the pull down menu at the top of the selection window. The viewer has selected the NGU model with sorted coordinates (Nenadic-Ghosh-Ulinski Model Sort) as the environment. The environment name is used as a key into a configuration file that holds the file names of the XML files representing the environments. The viewer uses these XML files to build the menu and to input the data. The visualization is based entirely on this XML file rather than on hard-coded information.

The coordinate group forms an organizational unit of elements that are to be treated in the same way by visualization. Examples of coordinate groups include neurons of a particular type, calcium channels, grid points for interpolated data, photodiodes, or even virtual points that have been placed to obtain a particular visual effect. Each element in a coordinate group has a spatial position and is associated with one or more time series.

A single data set can be partitioned into coordinate groups in many different ways. For example, stellate neurons may be placed in a single coordinate group, specifying that each stellate neuron is to be treated as a unit. Alternatively, data from individual compartments of the stellate neurons can be dumped separately. In this case, each type of compartment in the stellate neurons would form a distinct coordinate group with its own set of positions.