You might have seen John Scholes’ quietly famous implementation (in Dyalog APL) of Conway’s Game of Life. We’re going to explore just one of the array-language features in it, indexing one array with another.

You can experiment with the expressions at To animate them you need a Dyalog interpreter, with its editor and delay function.

The principle seems simple enough.

      'abcdef'[1 3 4 6]

The hidden surprise is that the result of such an expression has the same shape as the array in the index brackets.

      2 2⍴1 3 4 6
1 3
4 6
      'abcdef'[2 2⍴1 3 4 6]


Snow Crash, the title of Neal Stephenson’s acclaimed debut novel, is a reference to the random visual ‘white noise’ on the screen of a completely crashed computer. Let’s make some snow.

      '⎕⌹'[?2 2⍴2]
      '⎕⌹'[?2 2⍴2]

Now let’s follow John’s example and animate it.

      ⎕ed 'pic'
      {} {pic∘←'⎕⌹'[?30 100⍴2]⊣⎕DL÷⍵ ⋄ ⍵} ⍣40 ⊢8


Five seconds of ‘snow crash’. Perhaps it would look more snowy with snowflakes – for which we’ll use the five-pointed APL star.

      {} {pic∘←' *'[?30 100⍴2]⊣⎕DL÷⍵ ⋄ ⍵} ⍣40 ⊢8


Nice. Except real snow doesn’t flicker like that. It falls. It drifts.

Park that thought. Raise the temperature.

Oh, westron wind, when wilt thou blow
That the small rain down may rain?
Christ, that my love were in my arms
And I in my bed again!

wrote Henry VIII. Here we go.

      {} {pic∘←' |'[?30 100⍴2]⊣⎕DL÷⍵ ⋄ ⍵} ⍣40 ⊢8


Uh. Wasn’t that wind supposed to be blowing from the west?

{} {pic∘←' \'[?30 100⍴2]⊣⎕DL÷⍵ ⋄ ⍵} ⍣40 ⊢8


Enough rain. Back to snowflakes. And fewer of them.

      pic←' *'[1+1=?30 100⍴20]


Flakes must fall.We need a skyfall algorithm. Drop the bottom row, prefix new flakes at the top:

      pic←{' *'[1+1=?100/20]⍪¯1 0↓⍵} pic


And some animation.

      next←{' *'[1+1=?100/20]⍪¯1 0↓⍵} 
      {} {pic∘←next ⍵⊣⎕DL÷8} ⍣40 ⊢pic

This snow falls straight down, which isn’t quite right. We expect a little wind. From the west, perhaps.

   (L T)←{' *'[1+⍵]}¨1=?((⍴⍵)-0 1)⍴¨20
   L,T⍪¯1 ¯1↓⍵
{} {pic∘←next ⍵⊣⎕DL÷8} ⍣200 ⊢pic

The snowfield is 3-D. Nearer flakes should be larger – and fewer – than distant ones.

   (L T)←{'⍟**∘∘∘...... '[⍵]}¨13⌊?(0 ¯1+⍴⍵)⍴¨100
   L,T⍪¯1 ¯1↓⍵
{} {pic∘←next ⍵⊣⎕DL÷8} ⍣200 ⊢pic


There we go. 25 seconds of simulated snowfall.

We’re missing some features. Turbulence produces a certain amount of random jiggling for the flakes. Distant flakes should fall more slowly than near flakes. We’ve reached the limit of our simple representation of a snowfield as 2-D character array. We might (in a later post) represent the flakes differently, and project their representations onto a visual plane.

Watch the whole thing on YouTube!

For now, a Merry Christmas and – let it snow!

7 thoughts on “Let it snow

      1. Yes,
        at 800 meters MSL we had +8 °C today, 😦
        we are in desperate need of all the snow we can get

        Greetings from Austria.


Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s