How changing a single byte/word in a substitution box affects the inverse result?

alpominth

4/12/24, 6:34 AM

Let's suppose I have a 8-bit substitution box composed of 256-bytes and invert it.

Does changing one byte in the substitution box will make the inverse result completely different than if I didn't change it?

Or will only the changed byte be different in the inverse S-box?

279

2 + 0

s-boxes

Score:3

Crypto

kodlu

4/12/24, 7:12 AM

If the forward Sbox is

$$[S(0),\ldots,S(255)]$$ it is normally a permutation. You cannot change only one byte and keep the permutation property. You need to change at least two bytes. THus the inverse will also be changed only in two bytes. To give an example if you had $$ [S(0),S(1),\ldots,S(255)]=[7,31,21,\ldots,72] $$ then you might do $$ [S(0),\ldots,S(255)]=[21,31,7,\ldots,72] $$ with only the first and third bytes changed. So in the inverse direction now you have $$ S^{-1}(21)=0,S^{-1}(7)=2, $$ while you originally had $$ S^{-1}(21)=2,S^{-1}(7)=0, $$ with everything else unchanged.

+ 0

Score:2

Crypto

fgrieu

4/12/24, 7:10 AM

I we can inverse an 8-bit substitution box, this implies it is a permutation (example: the AES S-box).

If we change one byte in that table, then it is no longer a permutation, and can no longer be fully inverted: for one element (the new value) there will be two antecedents/premimages, and for one (the old value) there will be none.

If we exchange two (different) bytes in that table, then it remains a permutation, and can be fully inverted. In the table for the inverse, two bytes are exchanged (those which indexes are the values of the two bytes inverted in the direct table).

In both cases, the other 254 elements in the inverse table are unaffected.

+ 0

Elon Musk

I sit in a Tesla and translated this thread with Ai:

EN: How changing a single byte/word in a substitution box affects the inverse result?

Post an answer

Most people don’t grasp that asking a lot of questions unlocks learning and improves interpersonal bonding. In Alison’s studies, for example, though people could accurately recall how many questions had been asked in their conversations, they didn’t intuit the link between questions and liking. Across four studies, in which participants were engaged in conversations themselves or read transcripts of others’ conversations, people tended not to realize that question asking would influence—or had influenced—the level of amity between the conversationalists.