Help with the Cellular Automaton

  • Main tabs

  • Fixed windows

    • Variant window
    • Map window
      • Basic tab
        • Standard button
        • Alive at any point button
        • DE button
        • DE-4bin button
        • DE-5bin button
      • Original tab
        • Original seed button
        • DE-NN button
        • DE-NN-2bin button
        • DE-NN-4bin button
        • DE-NN-5bin button
        • DE-WB button
        • DE-WB-4bin button
        • DE-WB-5bin button
        • DE-Multi (shaded) button
        • DE-Multi2 (not shaded) button
        • DE-Multi3 button
      • Age tab
        • Fade newborn button
        • At least 10% button
        • All (100%) button
        • Cumulative age button
      • Variants tab
        • Multiple variants button
        • Variant change button
        • Change count button
        • Variant strength button
        • Variant top rank button
        • Variant dominance button
      • Memory tab
        • Status button
        • Capacity button
        • Capacity (including duplicates) button
        • Enforcement count button
      • Clusters tab
        • Connectivity button
        • Clusters button
        • Cluster split button
        • Regions button
      • Social tab
        • Similarity button
        • U/R button
        • Education button
    • Control window

      • Controls tab

        • Automaton info section
          • Play buttons
          • Reset button
          • Run until option
          • Draw interval option
          • Rnd.03 button
          • Calculate similarity button
          • Sketch outline button
        • Information section
        • Automaton settings (More controls)
          • Core processing selector
          • Neighborhood definition
          • Seed checkbox
          • Seed augmentation checkbox
          • Seed padding option
          • Social attribute lottery checkbox
          • Neighborhood size option
          • Global influence option
          • Random element option
          • Response data Check count button
          • Boundary detection option
          • Detect button
          • Clear button
          • Size (w x h) option
          • Zoom (cell width) option
          • Random birth/survival (%) option
          • Passive memory selector
          • Active memory selector
          • Slots option
          • Allow blocking checkbox
          • First update option
          • Update cycle option
          • Refresh active memory button
          • kNN similarity calculation neighborhood size option
          • Calculate similarities button
          • Scale similarity checkbox
          • Social similarity handling slide bar
          • Processing N option
          • Debug checkbox
        • Neighborhood rules
          • Neighborhood order option
          • Slider type buttons
          • GoL button
          • Bailey button
          • RBailey button
          • R9010
          • R8020
          • R7525
          • R6040
          • Birth slider
          • Survival slider

      • Context tab

        • Multitrack display
        • Data section
        • Neighborhood info display

      • Overview tab

        • Lifetime tracking display
        • Tallies by Age and Memory section
        • Tallies by Social on the live grid section
        • Tallies by Social in Active Memory
        • Seed statistics

 
  GISCA3 is a two dimensional cellular automaton which is used as a simulation of language diffusion. It has a square grid of locations, each of which is updated according to user-specified rules for each iteration (also called a generation) of the program. The map shows a picture of the Middle and South Atlantic States of the USA, from New York south to Florida including the eastern coastal states plus Pennsylvania and West Virginia. Only the eastern half of Georgia and the northern part of Florida are included for those states. The square grid has 22,801 (151 x 151) cells overall, of which only 8610 (shown in white rather than the grey of the others) are active in the simulation. Of these, actual data exists for 1162 locations, the number of speakers interviewed for the Linguistic Atlas of the Middle and South Atlantic States (LAMSAS). The map window plots information depending on the choices of the user, often making cells black or white, sometimes shading cells in shades of grey, and sometimes in shades of color (with the DE-Multi tabs under the Original tab, or the U/R and Education tabs under the Social tab).

 
 
 

Main Tabs

At the top of the display there are five main tabs: Data, Save, Measurement, View, and Boundaries. These tabs allow the user to select different options.

Data tab

The Data tab has a number of windows in which users can select characteristics for running the simulation.

Seed selector

The seed selector allows the user to choose from several different basemaps for the map window. The choices labeled lamsas-a through lamsas-e and lamsas-x each contain a basemap in which the 1162 cells with actual data (including all the social characteristics of the speakers) are used to provide social information for all of the other cells. The separate labels all have somewhat different assignments, so results of the simulation using the different selections can be expected to differ. The x-lags selection is a basemap for a different project, the Linguistic Atlas of the Gulf States, for which no data is currently available in the simulation.

Survey selector

The survey selector currently offers only one choice, LAMSAS.

Question selector

The question selector allows the user to choose which question from the LAMSAS worksheets to display data. Choices that begin with a g identify grammatical questions, chiefly verb forms. Choices that begin with an l identify lexical questions, items in the world that are designated by different words. Choices that begin with a p identify pronunciation questions, so responses will be represented in fine IPA phonetic transcription (if you see unfamiliar symbols, you do not have the right font loaded). You can find out more about the pages (the first number of items in the selector) and lines (the second number) by looking at the Item Index and Worksheet spreadsheets athttp://www.lap.uga.edu/Projects/LAMSAS/Information/.

Save tab

This tab saves a screenshot to your computer. The screenshot is in PNG format, so it can be opened only with a program that can read that format, such as Paint (in Windows) or a browser. The name of the file is generated automatically. You can rename the file and add a .png file extension in your file handling program (typically File Explorer for Windows) or open the file in Paint (or a similar program) and use the Save As command to save it with the name you choose and a file extension permitted by the program you choose.

Measurement tab

This tab allows the user to control various functions related to clustering.

SAC buttons (Standard, Fade Newborn, Age)

SAC stands for spatial autocorrelation, which says whether the data distribution in the map window is clustered (the usual condition for linguistic variants), random, or regularly distributed (like the squares of a chessboard). These buttons all calculate the statistic Moran’s I for the chosen display. If you click any of the buttons (say, the Standard button), after a short delay a display will open under the buttons with the result. Here is the display if you click the Standard button with the thunderstorm variant selected in the variant window (the default when you start the simulation):

SAC-analysis: 0.047 POSITIVE CORRELATION (Nearby areas tend to be similar), Statistically: SIGNIFICANT (99% possible) (expected value: -0.00011615750958299454, z: 6.1), (“thunderstorm”-data, 735 active locations, of which 735 are age-appropriate)

The result provides the relevant numbers for Moran’s I, and highlights whether there is a positive correlation (a clustered distribution) and whether the result is statistically significant (if the Z score is higher than 1.96). You will see different results for each button, Standard/Fade Newborn/Age, because Moran’s I will be calculated in each case on a different set of live cells. Standard is all the live (black cells); Fade Newborn is only the live (black) cells that have been alive for more than one iteration (generation); Age is only the live (black) cells that have been alive for at least 10% of the simulation iterations (generations), say for 10 iterations out of 100.

Clusters by connectivity buttons (Standard, Fade Newborn, Age)

This is an experimental section that counts the number of clusters found in the data display given the button pushed and the choices in two selectors:

Con. Threshold selector

This selector sets how many connections there have to be in a cluster, at minimum, for a cluster to be counted. The default value is 1; users can enter other numbers in the selector, say “4,” which would only count the number of clusters that had at least four connections.

Con. Neighborhood selector

This selector allows the user to choose what counts as a connection: Rook’s Case or Queen’s Case. Rook’s Case is only orthogonal connections (up to four), those just above or below, or just to the right or the left, of the target cell. Queen’s Case also includes diagonal connections (up to eight), those diagonally above right and above left, or diagonally below right and below left, of the target cell, in addition to the orthogonal connections.

If you click any of the buttons (say, the Standard button), after a short delay a display will open under the buttons with the result. Here is the display if you click the Standard button with the thunderstorm variant selected in the variant window (the default when you start the simulation):

Cluster detection result: 94 clusters found. Average size: 3.0, min size: 2, max size: 14. In all, 38.6% (284 cells) of alive cells were approved, or 3.3% of all 8610 non-boundary cells). Note: Cluster detection was performed for value: “thunderstorm”, using connectivity threshold at 1 (neighborhood: rook) and minimum cell age at 0.

The result provides the relevant numbers given the button clicked, including the number of clusters and their min/max/average size. You will see different results for each button, Standard/Fade Newborn/Age, because the number of clusters will be calculated in each case on a different set of live cells. Standard is all the live (black) cells; Fade Newborn is only the live (black) cells that have been alive for more than one iteration (generation); Age is only the live (black) cells that have been alive for at least 10% of the simulation iterations (generations), say for 10 iterations out of 100.

Clusters by kernels (Phase 1, Phase 2)

This is an experimental section that identifies kernels, situations where a target cell has a given number of neighbors, and then “grows” them. Two selectors are available:

Kernel size selector

This selector allows the user to set the minimum number of live neighbors for a cluster to be identified. The default is 4; users may enter other numbers between 1 and 8.

Minimum age selector

This selector sets how long a cell mush have been alive in a simulation run before it can be included in a cluster.

Phase 1 button

If you click the Phase 1: Find kernels button, after a short delay a display will open under the buttons with the result. Here is the display if you click the Phase 1: Find kernels button with the thunderstorm variant selected in the variant window (the default when you start the simulation):

Cluster detection result: 12 clusters found. Average size: 3.7, min size: 1, max size: 19. In all, 6.0% (44 cells) of alive cells were approved, or 0.5% of all 8610 non-boundary cells). Note: Cluster detection was performed for value: “thunderstorm”, using connectivity threshold at 4 (neighborhood: unavailable) and minimum cell age at 0.

The result provides the relevant numbers given the button clicked, including the number of kernels and their min/max/average size.

Phase 2 button

If you then click the Phase 2: Grow kernels button, after a short delay the display will change:

Cluster detection result: 12 clusters found. Average size: 7.4, min size: 3, max size: 24. In all, 12.1% (89 cells) of alive cells were approved, or 1.0% of all 8610 non-boundary cells).

Note: Cluster detection was performed for value: “thunderstorm”, using connectivity threshold at 1 (neighborhood: rook) and minimum cell age at 0.

The new result still shows the relevant numbers, but the numbers have changed to reflect the fact that the kernels have “grown” (ie, what the whole cluster is, given that a kernel was identified). There may be fewer kernels after Phase 2 because some kernels may have merged.

View tab

The view tab allows the user to select the ages of the live cells shown. There are two selectors. The default for both of these is -1, which means that the function is turned off and all live cells are shown.

Minimum age

Users may enter 0 or any positive integer to set the lowest age of live cells to be displayed.

Maximum age

Users may enter 0 or any positive integer to set the highest age of live cells to be displayed.

Boundaries tab

The Boundaries tab controls how the boundary of the cells in the map window is set. The default condition of the map in the map window provides a buffer of cells around the cells that would otherwise be on the map, so that proximity rules work correctly even on cells at the very edge of the map (ie, every active cell on the map has a full set of eight neighbors in the cells next to it). You might want to trim the boundaries if you wanted to save an image of a map, to make it look better.

Detect boundaries

Clicking the Detect Boundaries button restores the full boundary buffer if either one of the other buttons has been clicked.

Time boundaries (extreme)

Clicking the Trim boundaries (extreme) button takes away all of the extra cells in the buffer.

Trim boundaries (soft)

Clicking the Trim boundaries (soft) button takes away some of the extra cells in the buffer.

 
 
 

Fixed windows

There are three fixed windows: the variant window, the map window, and the control window.

Variant window

The variant window shows the responses gathered in the LAMSAS survey for the question selected. They are presented in order by frequency. The NR response type means “no response” and the NA response type means “not asked.” It is normal for large surveys to have missing data like this. Clicking on any of the responses will show in the map windows where the responses of that kind were collected. Holding down the CTRL key while clicking allows the user to select multiple responses from the list. The simulation runs faster the fewer response types are selected. Clicking the Play button in Controls without removing any of the variant reponses makes the simulation run the slowest—for the default “thunderstorm” question (l6#2) the simulation processes 109 responses at once!

Keep button

Clicking the Keep button will keep the selected responses and get rid of the others, so that the simulation will only run with the user’s selected responses. The simulation runs the fastest when one response is selected and the Keep button is clicked.

Merge button

Clicking the Merge button will put the data from the selected responses together, so that running the simulation will process the merged responses as a single response type. Users might want to do this to combine responses that are only different for a superficial reason, like “thunderstorm” and “thunderstorms.”

Clean button

Clicking the Clean button will remove the NR, NA, and empty responses from the data before running the simulation.

T50, T20, T10 buttons

Clicking any of these buttons will remove any response that did not occur at least the number of times specified (eg, clicking T50 removes responses that did not occur 50 times).

Map window

The map window shows general descriptive statistics at the bottom, the map in the middle, and a number of tabs at the top. The tabs allow the user to change the display in the middle map.

Basic tab

This tab determines how live cells are displayed on the map.

Standard

This button shows the cells on the map that are currently alive in black, as opposed to cells that are not alive in white.

Alive at any point

This button shows the cells on the map in black that have been alive during the current run of the simulation, even if they are not alive in the current generation.

DE (current)

This button displays cells in shades of grey according to the probability that the selected variant might be found there, as of the current generation of the simulation run. The darker the grey, the higher the probability. Density estimation (DE) is a statistic that, for every cell, considers the status of its nearest neighbors to estimate how likely it is that a selected variant might be found there. Displays of actual survey data, such as those produced in the simulation, are sparse (ie, under twenty percent of cells can have data in them), and DE produces estimates that cover all of the cells in the grid. The display for this button shows a continuously variable shade of grey for the estimates. These DE plots use the Kernel method, which uses a smoothing function to make the shading look more continuous.

DE-4bin (current)

This button displays cells in four shades (black, dark grey, light grey, white) to show the probability that the selected variant might be found in them, as of the current generation of the simulation run. Each shade corresponds to a quartile: the highest probability quartile is black, the next dark grey, the next light grey, and the lowest probability quartile is white.

DE-5bin (current)

This button displays cells in five shades (black, very dark grey, dark grey, light grey, white) to show the probability that the selected variant might be found in them, as of the current generation of the simulation run. Each shade corresponds to a quintile: the highest probability quintile is black, the next very dark grey, the next dark grey, the next light grey, and the lowest probability quintile is white.

Original tab

All of the buttons under this tab produce visualization of the original data distribution for the selected variant or for all of the variants for the selected question.

Original seed button

This button display the data locations where the selected variant was originally found in the survey data.

DE-NN

This button displays cells in shades of grey according to the probability that the selected variant might be found there in the original survey data. The darker the grey, the higher the probability. Density estimation (DE) is a statistic that, for every cell, considers the status of its nearest neighbors to estimate how likely it is that a selected variant might be found there. Displays of actual survey data, such as those produced in the simulation, are sparse (ie, under twenty percent of cells can have data in them), and DE produces estimates that cover all of the cells in the grid. The display for this button shows a continuously variable shade of grey for the estimates. These DE plots use the Nearest Neighbors method, which does not use the smoothing function in the DE plots under the Basic tab. Shading will be somewhat choppier, but perhaps more aligned with the estimate for every cell.

DE-NN-2bin

This button displays cells in two shades (black, white) to show the probability that the selected variant might be found in them in the original survey data. Each shade corresponds to a whether the probability of finding the selected variant location is 50% or greater (black) or less than 50% (white).

DE-NN-4bin

This button displays cells in four shades (black, dark grey, light grey, white) to show the probability that the selected variant might be found in them in the original survey data. Each shade corresponds to a quartile: the highest probability quartile is black, the next dark grey, the next light grey, and the lowest probability quartile is white.

DE-NN-5bin

This button displays cells in five shades (black, very dark grey, dark grey, light grey, white) to show the probability that the selected variant might be found in them in the original survey data. Each shade corresponds to a quintile: the highest probability quintile is black, the next very dark grey, the next dark grey, the next light grey, and the lowest probability quintile is white.

DE-WB

This button displays cells in shades of grey according to the probability that the selected variant might be found there in the original survey data. The darker the grey, the higher the probability. These DE plots use the Kernel method, which uses the smoothing function in the DE plots under the Basic tab in order to make the shading more continuous.

DE-WB-4bin

This button displays cells in four shades (black, dark grey, light grey, white) to show the probability that the selected variant might be found in them in the original survey data. Each shade corresponds to a quartile: the highest probability quartile is black, the next dark grey, the next light grey, and the lowest probability quartile is white.

DE-WB-5bin

This button displays cells in five shades (black, very dark grey, dark grey, light grey, white) to show the probability that the selected variant might be found in them in the original survey data. Each shade corresponds to a quintile: the highest probability quintile is black, the next very dark grey, the next dark grey, the next light grey, and the lowest probability quintile is white.

DE-Multi (shaded)

This display displays cells in different colors corresponding to different variants for the selected question. DE-Multi is a new proportional density estimate plot that observes the multiple variant densities at each location and produces a color-coded plot of the strongest variants at each location. Thus, the more intense the color, the higher probability for the variant assigned that color. This DE plot reflects the competition between different variants, as it is revealed by different probabilities for finding multiple variants at any location. In order to look for the variant associated with each color, click on the Context button in the Control window and mouse over different colors.

DE-Multi2 (not shaded)

This display displays cells in different colors corresponding to different variants for the selected question. DE-Multi is a new proportional density estimate plot that observes the multiple variant densities at each location and produces a color-coded plot of the strongest variants at each location. This plot does not use different intensities of color as shading, so that a color is displayed for the highest probability variant at each location. In order to look for the variant associated with each color, click on the Context button in the Control window and mouse over different colors.

DE-Multi3

This display displays cells in different colors corresponding to different variants for the selected question. DE-Multi3 is a new proportional density estimate plot that observes the multiple variant densities at each location and produces a color-coded plot of the strongest variants at each location. Thus, the more intense the color, the higher probability for the variant assigned that color. DE-Multi3 differs from DE-Multi in that it uses a measure of distance to the next most popular variant to create the shading. In order to look for the variant associated with each color, click on the Context button in the Control window and mouse over different colors.

Age tab

Buttons for this tab change the display of live cells according to how old they are in the current run of the simulation. In order to change back to the default view of all the cells currently alive in this run of the simulation, click the Basic tab and click the Standard button.

Fade Newborn button

Clicking this button changes the display of cells that just became alive in the current generation from black into a light grey color. Since many cells in the simulation change frequently between not alive and alive, this button makes it easier to see cells whose alive status is persistent.

At least 10% button

Clicking this button changes the display of live cells from black into a grey color depending on their relative age in the current run of the simulation. Since many cells in the simulation change frequently between not alive and alive, this button makes it easier to see cells whose alive status is persistent. Black cells have been alive for at least 10% of the simulation (eg, for at least 10 generations if the simulation has run for 100 generations). As the run of the simulation continues, there will eventually be multiple shades to designate different ages of live cells (black for oldest, dark grey, light grey, lighter grey, white for the newest cells).

All (100%) button

Clicking this button changes the display of live cells from black into a grey color depending on their relative age in the current run of the simulation. Since many cells in the simulation change frequently between not alive and alive, this button makes it easier to see cells whose alive status is persistent. As the run of the simulation continues, there will eventually be multiple shades to designate different ages of live cells: black for oldest cells, dark grey, light grey, lighter grey, white for the newest cells. This button differs from the At least 10% button in that black cells are those in the oldest quintile, so that as the simulation continues there will be fewer black cells than there are in the At least 10% display.

Cumulative age

Clicking this button changes the display of live cells from black into a grey color depending on their cumulative age in the current run of the simulation. Since many cells in the simulation change frequently between not alive and alive, this button makes it easier to see cells whose alive status is persistent. Darker color cells have been alive for for the most generations of the simulation. As the run of the simulation continues, there will eventually be multiple shades to designate different ages of live cells (black for oldest, dark grey, light grey), all of them at least a little grey since all cells will have been alive at least some of the time.

Variants tab

Visualizations for this tab consider relative status among many variants. If the current run of the simulation has only one variant selected, these visualizations will not be useful. You can see the variants alive at particular cells if you click the Context button in the Controls window and mouse over cells.

Multiple variants button

This button shows the number of variants that are alive at every location, the darker the cell the more variants present. Cells with five or more variants are black. As the run covers more generations, the map will appear more and more black.

Variant change

This display uses three colors to show whether the variant originally present at a location is still there (black), has been lost (red), or has been added (green). Since over 80% of cells did not have data at the beginning of a simulation run, green will be the most common color. The display will be different if you select different variants in the Variant window.

Change count

This display shows the number of changes between not alive and alive for each cell. The more changes, the darker the color. As the simulation run continues, the display will get darker and darker overall. The display will be different if you select different variants in the Variant window.

Variant strength

This button shows the strength of a variant at the locations on the map. Strength corresponds to age, how long the variant has been alive at a location. The stronger a variant is at a location (ie, the longer it has been alive there), the darker the color of the cell. The display will be different if you select different variants in the Variant window.

Variant top rank

This button shows whether the selected variant is top ranked (ie, the oldest living cell) at each location on the grid. Dark green means that the variant is top ranked by itself; light green means that the variant is tied with other variants for the top rank. The display will be different if you select different variants in the Variant window.

Variant dominance

This button shows whether the selected variant is top ranked (ie, the oldest living cell) at each location on the grid. Ties are not allowed. Dark green means that the variant is top ranked. Light green means that the variant exists with other variants at the location. Yellow means that the variant has existed at the location but is not currently alive there. The display will be different if you select different variants in the Variant window.

Memory tab

Options under this tab all display different aspects of the memory function of the simulation. During a run of the simulation with multiple variants (not with a single variant), the simulation keeps track for every location of the variants that have been alive there. This models the passive memory of speakers. In addition, whenever a variant has been alive for many generations (by default, 10% of the run), it is added to a different list that models speakers’ active memory, those variants that speakers would not just know but use. Active memory is subject to different controls (see the appropriate Help section) that set the number of variants that can be in the list (by default, 5) and whether or not they can be forgotten.

Status button

This display shows which locations have the selected variant in active memory, as of the current generation in the simulation run. The display will be different if you select different variants in the Variant window.

Capacity button

Capacity (including duplicates)

This display shows the number of active memory slots used, out of the number available (by default, 5). The color of a cell depends on what proportion of slots are used, where darker colors are more and lighter colors are fewer. Red cells are at capacity. This button does consider duplicate entries in the active list.

Enforcement count

The simulation checks to see whether entries in the active memory list still have been alive long enough to be eligible for listing. If so, an entry scores one point in its enforcement score; if not, one point is subtracted from its enforcement score. This button colors cells according to the enforcement score of the selected variant, darker colors as the score gets higher.

Clusters tab

This tab is used in conjunction with the Measurement tab. Before trying to use the Clusters tab, you should detect clusters under the Measurement tab by using one of the buttons (Standard, Fade Newborn, Age) of the Clusters by connectivity option.

Connectivity button

This button makes the result of the Clusters by connectivity result more grey.

Clusters button

This button returns the map to the back/white colors of the Clusters by connectivity result.

Cluster split button

This button assigns a unique color to the clusters detected in the Clusters by connectivity result.

Regions button

This button colors many locations according to clusters found there. It is only useful early in a simulation run (the colored locations cease to appear later in the run).

Social tab

Options here display information about the social status of locations.

Similarity button

Not currently functional.

U/R button

Clicking this button displays urban locations in red and rural locations in green. Which locations are red and green are different between the different seeds. All seeds preserve the proportion of urban to rural locations in the original survey data. An urban location is one that has a population of at least 2000 in the 1940 US Census.

Education button

Clicking this button displays the level of education at each location, from yellow for little education through orange to red for higher education. The color of each location differs between the different seeds. All seeds preserve the proportion of educational levels at locations in the original survey data.

Control windows

 
The control window allows the user to set many parameters for a simulation run. It also allows the user to display statistics in lists, graphs, and charts for the current simulation run.

Controls tab

This tab contains the basic settings for a simulation run, as well as the means to start and stop the simulation run.

Update statistics

This button brings the statistics shown up to date for the current simulation run.

Automaton info section

This section shows the number of generations of the current simulation run, and offers different controls for running the simulation and showing the results.

Play buttons

The Play button starts the simulation run, and also stops it if you click on it another time. The simulation run can also be started and stopped by pressing the space bar on your keyboard. Each of the buttons with a + and a number will run the simulation for a certain number of generations. The simulation run occurs for all of the variants in the variants window, so the simulation runs much slower if many variants are present there and much faster if the Keep button or one of the T50, T20, T10 buttons has been used to limit the number of variants for processing.

Reset button

Clicking the reset button reloads the simulation to its initial default state. You can also use CTRL+r on your keyboard to reset the simulation.

Run until option

The default setting for the Run until option is 0, which lets the simulation continue to run as long as the user does not press the spacebar or click the Play button another time. The user can enter any positive integer in the Run until box to set an exact number of generations for the simulation to run.

Draw interval option

The default draw interval is 1, which means that the simulation will update the map window after every generation. Since redrawing takes time and computer resources, you may want to update the map window less often, say every 10 generations or every 100 generations. The user can enter the number of generations between map updates in the box. Entering -1 in the box causes the simulation to update the map window only at the end of the simulation run.

Rnd.03 button

Clicking this button cause the simulation randomly to overturn the decision for a cell, say from become alive or stay alive to turn off, in 3 cases out of every 10000. This level of randomness slows down the formation of clusters and allows low frequency variants to remain alive over the course of a long simulation run.

Calculate similarity button

This button calculates social similarity scores according to the parameters entered under the More controls button lower down in the Controls tab.

Sketch outline button

This button puts a black outline around colored clusters, if these are present in the map window.

Information section

This section shows the data associated with different cells in the map as the mouse passes over them. Right-clicking a cell in the map window allows the user to set a particular cell for display, in order to perform further functions on it. Some cells have a survey ID number; most cells are “artificial,” meaning that they do not have values from the original survey associated with them. Other boxes show the x/y coordinates of the cell in the map window and data about the status of the cell in the current run of the simulation. The last six boxes, those with outlines, show the social information from the original survey if such data is associated with the cell, or show the social information assigned to an artificial cell in the selected seed. Clicking one of the last six cells shows the weight assigned to the social value and the ± range, as assigned under the More controls button lower down in the Controls tab.

Automaton settings (More controls)

This section allows the user to specify a large number of parameters for a simulation run. Clicking the More controls button reveals the list of parameters; clicking the Less controls button hides the list of parameters.

Core processing selector

There are three possibilities here: single track with multiple variants, multiple tracks and variants, and legacy mode.

Neighborhood definition

The default is proximity alone. Socially similar neighbors can be searched, by default within two cells (24 cells evaluated). Users can assign different weights to different social factors, and scores from the resulting formula are used to define similarity. Four options apply social options: socially similar neighbors, socially similar neighbors with rounding, socially similar neighbors by count, and socially similar neighbors for and against.

Seed checkbox

Checking the Relocate stacked cells box just means that original survey speakers whose locations put them in the same cell (thus stacked) will be moved to adjacent cells so that there is one speaker per cell.

Seed augmentation checkbox

??

Seed padding option

??

Social attribute lottery checkbox

These options are associated with the creation of new seeds, which is not possible in this online version of the simulation.

Neighborhood size option

Global influence option Random element option

Response data Check count button

This button counts how many cells the selected variant has alive on the map. A small window will open on the simulation to give the number from the count.

Boundary detection option

This option lets the user decide how many cells of padding are added to the boundary (default is 3). Not possible in this online version of the simulation.

Detect button.

Applies the number from the Boundary detection option to the map. Not possible in this online version of the simulation.

Clear button

Clears the padding from the boundary. Not possible in this online version of the simulation.

Size (w x h) option

Sets the size of the map. Not possible in this online version of the simulation.

Zoom (cell width) option

Set the size of a cell. Not possible in this online version of the simulation.

Random birth/survival (%) option

This option sets the amount of randomness to be used in evaluation of the rules, either for a cell to become live if it was not alive, or to stay alive if it was already alive. Percentages below .06% will maintain complex behavior (cluster formation) during a simulation run. Numbers higher than .06% will throw the simulation into a chaotic state (no cluster formation).

Passive memory selector

There are four choices: no memory, unbiased, eager to learn new, eager to remember. There are also options for Factor, minimum value, and maximum value. This selector is not typically used in the online simulation. Passive memory is chiefly used in the simulation to determine whether a variant has been persistent enough to be included in Active memory.

Active memory selector

There are four choices: no memory, nothing forgotten, continuous memory loss, and loss after capacity is reached. The default is nothing forgotten. The continuous loss setting means that variants will be lost from Active memory if their persistence score in passive memory drops to zero. The last option is the same, except that variants are not lost unless the maximum capacity is reached.

Slots option

This user selectable number represents how many variants can be stored in Active memory.

Allow blocking checkbox

If this box is checked, Active memory cannot have duplicate entries of the same variant.

First update option

This user selectable number sets the generation at which the first memory check is carried out.

Update cycle option

This user selectable number sets the number of generations between updates of memory.

Refresh active memory button

This button forces an update of memory.

kNN similarity calculation neighborhood size option

This option sets the number of rows from the target cell to search for social similarities. The default is 2, which searches a total of 24 cells (in two rows around the target cell).

Calculate similarities button

Not used in the online version of the simulation.

Scale similarity checkbox

This box adjusts the similarity score by the size of the neighborhood. Normallly not checked.

Social similarity handling slide bar

This slide bar sets the weight of social variables. Cells in the first row around a target cell have a weight of 1; the default social similarity weight is .25.

Processing N option

The default value is 2. Not normally changed in the online version of the simulation.

Debug checkbox

Not used in the online version of the simulation.

Neighborhood rules

This section allows the user to make the rules for whether a target cell will be alive or not in the next generation of a run of the simulation. A cell may be born (turned on if it was off), or a cell may survive (stay on if it was already on), and rules for these two conditions are set separately. There are buttons for commonly used sets of rules, and also the means to make any kind of rule using options and sliders.

Neighborhood order option

This option allows the user to set the number of rows around a target cell to consider for rules. The default is 1 (8 cells). If the user enters a different number, the simulation will show a separate slide for each possible row.

Slider type buttons

There are two choices here, use of a slider that shows a proportion of cells (in percent), and a slider that uses a count of cells. The default setting uses a count of cells.

GoL button

This button sets the rule to having 3 neighbor cells alive for the birth of the target cell, and having 2 or 3 neighbor cells alive for the survival of the target cell. GoL stands for “Game of Life,” an early application of a cellular automaton.

Bailey button

This button sets the rule to having 2, 3, or 4 neighbor cells alive for the birth of the target cell, and having 5, 6, 7, or 8 neighbor cells alive for the survival of the target cell. This is the only count rule known to produce complex behavior (cluster formation).

RBailey button

This button is the equivalent of the Bailey button, only using proportions rather than counts. Use the Cluster tab and Region button to see how it forms colored regions from the list of variants (black dots are locations where more than one variant survives).

R9010 button

This button uses proportions, 90% of neighbor cells alive for birth, 10% of neighbor cells alive for survival. Use the Cluster tab and Region button to see how it forms colored regions from the list of variants (black dots are locations where more than one variant survives).

R8020 button

This button uses proportions, 80% of neighbor cells alive for birth, 20% of neighbor cells alive for survival. Use the Cluster tab and Region button to see how it forms colored regions from the list of variants (black dots are locations where more than one variant survives).

R7525 button

This button uses proportions, 75% of neighbor cells alive for birth, 25% of neighbor cells alive for survival. Use the Cluster tab and Region button to see how it forms colored regions from the list of variants (black dots are locations where more than one variant survives).

R6040 button

This button uses proportions, 60% of neighbor cells alive for birth, 40% of neighbor cells alive for survival. Use the Cluster tab and Region button to see how it forms colored regions from the list of variants (black dots are locations where more than one variant survives). This button produces the regions with fewest black dots.

Birth slider

Users can move the sliders manually to set the rule for birth, either in proportions or in counts (the default).

Survival slider

Users can move the sliders manually to set the rule for survival, either in proportions or in counts (the default).

 
 
 

Context tab

This tab shows information about the neighborhood of each location in the map as the user mouses over the cell.

Multitrack display

This display shows where variants occur in the neighborhood of the selected cell. This includes the variant at the target cell (if there is one), plus any variants in the neighborhood (the default is first-order neighbors, so 8 additional cells). A bar chart shows the relative numbers for each variant.

Data section

This section gives the counts for different variants in the selected area, and offers a table that shows additional information about each variant such as age, number of changes at the location, and memory status.

Neighborhood info display

This section shows a map of the cells in the selected neighborhood, showing the selected cell as black and other dells with data as white outlined cells.

 
 
 

Overview tab

This tab shows charts and text data about the status of the simulation run. Clicking the triangle to the left of each heading either expands or compresses the section.

Lifetime tracking display

This section shows two displays. The first is a table for the number of active cells at the current stage of the run. Initially only one line is shown, for the top ranked variant. It shows the number of cells for the variant, in red or green or black), for each generation (iteration) of the run. Green means the number has increased, red decreased, and black stayed the same. If you click on the row, rows for all the live cells will be displayed. The second display shows a chart with one line per live variant (mousing over a line will show which variant it represents). These lines typically rise from the x axis until they reach about 4000 cells, at which point they level off. In order to preserve more variants, all the ones with low initial numbers of cells, it is necessary to include a “random” component in the run, as by clicking the RND.03 button

Tallies by Age and Memory section

This section first shows the a-curve for counts of the cumulative age of each variant, i.e. the number of times any cell has been alive for the variant over the course of the run. The chart is followed by a table that gives the numbers for cumulative ages for every variant, including those that were never alive. The next display shows the relative strength of live variants in both an a-curve chart and a table with numbers. Next, values for the active memory for each variant are shown both as an a-curve chart and a table. Finally, an a-curve chart and table are given for the enforced active memory tally.

Tallies by Social on the live grid section

This display is only active if social factors have been programmed into the run.

Tallies by Social in Active Memory

This display is only active if social factors have been programmed into the run.

Seed statistics

This displays the original counts for variants, and gives the number of responses per speaker.