rank distribution plots

[ajrominger]

two things:

• let's not have a color gradient, instead let's just animate the current timestep with a single color and not show all the timesteps at once
• let's have the axes be somewhat dynamic so they respond to the changing numbers of species (x axis) and changes to max abundance (y axis)

[ajrominger]

re: dynamic axes, here's what seems to be a python and plolty solution: https://stackoverflow.com/questions/69584171/is-there-a-way-to-dynamically-change-a-plotly-animation-axis-scale-per-frame

[ajrominger]

here might be a solution based on this post

# make some pretend data ----

# `foo` is a data.frame with 
#     `gen` = generations
#     `x` = data for x-axis (i.e species rank)
#     `y` = data for y-axis (i.e. species abund)

set.seed(1)
foo <- lapply(1:10, function(i) {
    x <- 1:(rpois(1, 10) + 1)
    y <- rev(x)
    gen <- i
    
    data.frame(gen = gen, x = x, y = y)
})

foo <- do.call(rbind, foo)

# make the plot ----
# i don't know how to do that so i'm bs'ing
graph_animated <- ggplotly # stuff

# build
plt <- plotly_build(graph_animated)


# modify the layout for each frame ----

# get moving averages of x and y limits
for(i in 1:max(foo$gen)) {
    ii <- i + (-2:2) # get a range of indeces centered at the current gen
    ii <- ii[ii %in% foo$gen] # limit to only valid indeces
    
    # set x and y lims for frame `i`
    plt$x$frames[[i]]$layout <- list(xaxis = list(range = range(foo$x[ii])), 
                                     yaxis = list(range = range(foo$y[ii])))
})

[connor-french]

Quick update and notes:
I got it at least running! When I translated it to my code, I introduced a couple indexing bugs that I finally worked out...

Issue: up until generation 4, the axes reset every time step (0.2 generations) and it's a headache to watch. After generation 4, the axes stop resetting.
Ideas: I think I can play around with the indices to get this working. Maybe the range where we get the min/max limits isn't appropriate.

The latest commit has my current implementation (f06cc2a)

[ajrominger]

i was playing with this this morning with Renata and we decided best to leave the axes static for now--in no small part because she needs to record a lecture with it for class! so that's what i did with this commit: 6b718f8

[connor-french]

Damn, after trying with the dynamic axes for a while, I found out it's impossible to do in plotly right now. We keep getting wonky axis-resetting because dynamically updating the frame dimensions and data dimensions is currently not supported by the source plotly.js. The solution referred to in the post you linked @ajrominger only works for a single axis change. There are other animation possibilities that I can look into, but there is the ability of users to zoom in with the existing plotly framework that I can show you in our next meeting