# Re: can we find one-way trapdoor funcation family from the theory of calculus

From: Andrew Swallow (am.swallow_at_btopenworld.com)
Date: 11/21/05

• Next message: David Wagner: "Re: Provable security of independent encryptions"
```Date: Mon, 21 Nov 2005 20:19:58 +0000 (UTC)

```

Unruh wrote:

> Andrew Swallow <am.swallow@btopenworld.com> writes:
>
>
>>Unruh wrote:
>
>
>>>Andrew Swallow <am.swallow@btopenworld.com> writes:
>>>
>>>
>>>
>>>>Unruh wrote:
>>>>
>>>>
>>>>>"Pubkeybreaker" <Robert_silverman@raytheon.com> writes:
>>>>
>>>>[snip]
>>>
>>>
>>>>>>Trying to determine, therefore, if it can somehow be made into a
>>>>>>one-way function
>>>>>>is just total nonsense.
>>>>>
>>>>>
>>>>>No, but then that was not what they were saying. An integral is a
>>>>>transformation on a functions. Can such transformations be used to make a
>>>>>one way function?
>>>
>>>
>>>>You could use something like the sin function. If your number is bigger
>>>>than 360 then sin(x) cannot be inverted because it is a m:1 function.
>>>
>>>
>>>Unfortunately a trapdoor function is one that can be inverted, just not
>>>very easily.
>>>
>>
>>Try sin(x + 13N) where N is large
>>To get from arcsine to x you need to know N.
>
>
> If you know the value for one x (well, actually two x), you know the value for all x. This is a
> pretty useless trapdoor function. Ie, a single known plaintext/encrypted
> pair breaks the scheme.
> Not only do you need unpredictability, you need resistance to known pairs.
>

>>In a strong system N would be a function rather than a constant.
>>Replacement of sin by a binary or integer function would also help.
>
>
>>Andrew Swallow

That is why N has to be a function.

Andrew Swallow

• Next message: David Wagner: "Re: Provable security of independent encryptions"