Yes, from the description:

The return value should be a 2d array cropped around all of the living cells.

thanks for the hint

Hi, did you ever figure this out? Your input is clearly length four and should have two bytes of padding.

This is most likely because you are mutating the input array, i suggest you just create a copy of fhe array and you should pass this issue

Only cells with exactly 3 neighbors become alive. The second cell has three neighbors that are alive (diagonals are included). The third cell does indeed have two neighbors, but only two, so it maintains whatever stat it had.

Yup this is the case

I accidently changed that back with my fork. Sorry about that.

I am getting the same thing.
Is there a way to get past this?

Edit: Ok all you need is to round(num, 4) - in Python.

Not a mathematical set. I'll change it to say "group".

For encoding the autokey is basically the key concatenated the plain text (message below).

E.g. with a key of "password":

message: my secret code i want to secure
key:     pa ssword myse c retc od eiwant

In decoding, you have to infer it from what you've already decoded. I.e. when you decode with password you get the first few characters of the rest of the autokey and so on.

You have to use the plain text as part of the key when the plain text is longer than the key. Otherwise what you produce is the non-autokeying variety of the Vigenere Cipher.

Read the description again.

yet both 103 and 107 are primes and the gap is 6

the gap is 4. For [101, 107] the step is 6 but it is not a "gap" because there is 103 in between.
No problems with the tests. Did you make the distinction between "gap" in primes and "step" in primes? To see the difference look at: http://www.codewars.com/kata/5613d06cee1e7da6d5000055. Maybe you could begin with "Steps in Primes" kata. Hope that helps!