Timeline for Diffusion of arithmetic and bit operations
Current License: CC BY-SA 4.0
8 events
| when toggle format | what | by | license | comment | |
|---|---|---|---|---|---|
| Oct 18, 2022 at 5:35 | comment | added | kelalaka | Ok, the figures are confusing. X-or is fine, however, modular or not, addition can effect more than one bit and that is really depend on the size of modulus $n$ where it seems $2ˆ{64}$ | |
| Oct 18, 2022 at 5:24 | comment | added | Tom | @kelalaka diagrams are wrong? | |
| Oct 18, 2022 at 5:18 | vote | accept | Tom | ||
| Oct 17, 2022 at 6:16 | comment | added | Tom | 4. It looks like in last diagram all pixels are green, but some are lighter. How to interpret that? Does it shows some bits are mixed less often? | |
| Oct 17, 2022 at 5:29 | comment | added | Tom | 3. It seems like this diffusions don't add to each other. I wanted to do thought experiment and calculate diffusion of n multiplications (e.g. by adding toegether n times diffusions of single multiplication), but of course some bits may remain unchanged, especially if low input bits would be all zeros. So to increase diffusion we have "move" green bits into black bits positions - somehow, by bit shifts or arithmetic operations. So it is hard to caclulate analytically final result. Can I use this method for scientific purposes, to estimate diffusion of some algorithm (is this accepted method)? | |
| Oct 17, 2022 at 5:29 | comment | added | Tom | Ok, I have many questions. 1. Did you make this diagrams in some program or generate it in c++ with some graphic library? What is the name of it? 2. Authors of Skein wrote about full diffusion, is it possible to get 100% diffusion? I thought about (a >> 1) * b. It could achieve full diffusion after 64 iterations, istn't it? | |
| Oct 17, 2022 at 5:04 | vote | accept | Tom | ||
| Oct 18, 2022 at 5:18 | |||||
| Oct 17, 2022 at 0:43 | history | answered | Leo | CC BY-SA 4.0 |