Knight tour using geom_path, unexpected plot for matrix with a 5 digit number in it

[rgosik]

Hi, I use the following code to create kinght tour visualizations:

df <- read.csv("D:\\knightTours\\RothN.csv")
Mat = as.matrix(df)
var.size = n
coords <- tibble(col = rep(1:var.size, var.size),
                 row = rep(1:var.size, each = var.size))

coords %>% 
  mutate(order = Mat[var.size * col + row]) %>%  
  arrange(order) %>% 
  ggplot(aes(x = col, y = row )) + 
  geom_path() + 
  coord_equal()

Each csv file contains n * n amount of numbers: 0, 1, 2, 3, ... n * n

However when n > 99, the plot is not looking as expected, more precisly, I found out that the chaos occurs when there is any 5 digit number in the .csv file.
here is the plot for n = 99:

and here is the plot for n = 100:

in the same n = 100 plot, replacing 10000 digit with 0, while connecting 0 with 1, produces a plot correct with the numbering within the csv file, which cemented me in the belief of 5 digit number being a problem:

the issue grows only larger with bigger n, here is n = 125:

The plot looks as expected for all n <= 99, why the appearance of 5 digits number in the matrix destroys the whole plot ?

[AlexisW]

@rgosik

mutate(order = Mat[var.size * col + row]) %>%

Are you sure about that line? col and row start counting at 1, so the smallest coordinate you look up is n*1+1 = n+1 and you never read the Mat cell at [1,1]. I think you should use:

mutate(order = Mat[var.size * (col-1) + row]) %>%  

or equivalently:

mutate(order = map2_int(row, col, ~ Mat[.x, .y])) %>%

Not totally sure without data to test on. If that's not the solution, you should also double-check the type of the order column, make sure it doesn't get arranged by alphabetical order or something.