# Re: Constructing a random permutation on the fly

Ilmari Karonen <usenet2@xxxxxxxxxxxxxx> writes:
As for the number of rounds, there's a well-known result by Luby and
Rackoff saying that four rounds are enough (even for crypto purposes,
which your use case isn't) if your round function is random enough.

No, 4 rounds isn't enough for short word sizes like this, where the
probability of collision is non-negligible. I remember some paper
saying 7 rounds was enough but I don't remember the reason. I usually
use 10 or so rounds when I do stuff like this, but it's also been on
slightly larger inputs.
.

## Relevant Pages

• Re: Constructing a random permutation on the fly
... look up "Hasty Pudding cipher". ... Use a bunch of rounds to ... the desired range and iterate the encryption function until you get a ... which your use case isn't) if your round function is random enough. ...
(sci.crypt)
• Re: Triple AES (3AES)
... However it has been proven that the AES round function ... I can imagine that a weak cipher can ... generate the symmetric group. ... Enough rounds, with independent keys, can generate any ...
(sci.crypt)
• Re: Rounding
... When I use the ROUND function in VBA for ACCESS or EXCEL that does not happen. ... Am I doing something wrong or does VBA use 6 as the cutoff for rounding to ... rounds down; more than 5 rounds up; and a number ending in exactly 5 rounds to ...
(microsoft.public.access.modulesdaovba)
• Re: Is there a PROPER round function to use in ACCESS?
... Savvoulidis Iordanis wrote: ... > I can't believe that ACCESS doesn't have a round function that ... Assuming the error caused by translating between decimal and binary is ... rounds one way if odd and the other if even. ...
(microsoft.public.access.formscoding)
• Re: Rounding Problem
... > This is a typical problem where I have a column of numbers that when ... > formula that uses the ROUND function that rounds each cell prior to ...
(microsoft.public.excel)