10 PRINT CHR$(205.5+RND(1)); : GOTO 10

I felt inspired to do 10 PRINT CHR$(205.5+RND(1)); : GOTO 10 in R.

knitr::opts_chunk$set(
  collapse = TRUE, comment = "#>", fig.width = 5, fig.height = 5)

library(ggplot2)
library(dplyr)
#> 
#> Attaching package: 'dplyr'
#> The following objects are masked from 'package:stats':
#> 
#>     filter, lag
#> The following objects are masked from 'package:base':
#> 
#>     intersect, setdiff, setequal, union

draw_print10 <- function(height = 100, width = 100, rows = 20, 
                         cols = 20, p = .5) {
  
  x_starts <- seq(0, height, length.out = rows + 1)
  x_ends <- seq(0, height, length.out = rows + 1)
  x_starts <- x_starts[-length(x_starts)]
  x_ends <- x_ends[-1]

  y_starts <- seq(0, width, length.out = cols + 1)
  y_ends <- seq(0, width, length.out = cols + 1)
  y_starts <- y_starts[-length(y_starts)]
  y_ends <- y_ends[-1]

  grid <- tidyr::crossing(
    tidyr::nesting(x_start = x_starts, x_end = x_ends),
    tidyr::nesting(y_start = y_starts, y_end = y_ends))

  grid <- grid %>%
    mutate(
      flip = runif(length(y_end)) <= p,
      y_temp1 = y_start,
      y_temp2 = y_end,
      y_start = ifelse(flip, y_temp2, y_temp1),
      y_end = ifelse(flip, y_temp1, y_temp2)) %>%
    select(x_start:y_end)

  ggplot(grid) +
    aes(x = x_start, y = y_start, xend = x_end, yend = y_end) +
    geom_segment(size = 1)
}

# Equal chance of flipping
draw_print10() +
  theme_void()

# Much lower chance
draw_print10(p = .10) +
  theme_void()

# Vectorized probabilities so that p(/) changes with x value
p <- seq(0, 1, length.out = 20 * 20)

draw_print10(p = p) +
  theme_void()

# Crank up the resolution
p <- seq(0, 1, length.out = 40 * 40)

draw_print10(rows = 40, cols = 40, p = p) +
  theme_void()

draw_print10(rows = 40, cols = 40, p = p) +
  theme_void() +
  aes(color = p) +
  guides(color = FALSE)

# Use Gaussian densities as probabilities
p <- dnorm(seq(-4, 4, length.out = 40 * 40))
draw_print10(rows = 40, cols = 40, p = p) +
  theme_void() +
  aes(color = p) +
  guides(color = FALSE)

# Use Cauchy densities as probabilities (fatter tail)
p <- dcauchy(seq(-4, 4, length.out = 40 * 40))
draw_print10(rows = 40, cols = 40, p = p) +
  theme_void() +
  aes(color = p) +
  guides(color = FALSE)

# Use flipped exponential densities as probabilities
p <- c(dexp(seq(0, 4, length.out = 40 * 20)),
       dexp(seq(4, 0, length.out = 40 * 20)))

draw_print10(rows = 40, cols = 40, p = p) +
  theme_void() +
  aes(color = p) +
  guides(color = FALSE)