There's some function f() which has some "error", so "error" grows when decryption is failed every time and then finally it won't be able to decrypt. But, the "error" will be initialized when its decryption is succeed.

Is there any kind of crypto functions like this?

To prevent getting access grant in the password-based scheme from adversary, one of considerable option is using limit login attempts. This works only if the adversary can access via providing way such as APIs. When the attacker takes raw data from Database, it's useless. If there's a function like I mentioned above, it could be helpful to keep users' password.

  • $\begingroup$ No, but I could think of a scheme using a (trusted) third party easily. You could simply keep a counter or something similar $\endgroup$ – Maarten Bodewes Mar 25 '17 at 23:02
  • $\begingroup$ Right, what is being asked for isn't really a mathematical concept at all. They want a function, decryption, to mutate the input ciphertext (and any copies, I assume). This basically requires the same capabilities as DRM. $\endgroup$ – Thomas M. DuBuisson Mar 27 '17 at 1:55
  • $\begingroup$ I guess with some constraints on the input functions we can build a Functional encryption scheme for this property. $\endgroup$ – user38956 Mar 28 '17 at 7:35
  • $\begingroup$ Still thinking will answer in some time $\endgroup$ – user38956 Mar 28 '17 at 7:35

I do not belive such a thing exists if the raw data is taken because there would be no way of the cipher text 'knowing' if it was trying to be decryted add adding noise.


OK this may sound naive. But isn't noise intrinsic to any system until one can isolate it completely from the environment.. Flee thy into solitude? error functions in your case related to erf and erfc functions? Are you talking in terms of the Exploration-exploitation tradeoff? Adversarial problems with partial feedback? Intermittency in search strategies with heavy tailed distributions? A system which gives a noisy signal will have a broad spectrum.. And if it's getting buffeted as in simple Brownian motion it will forget its history being dictated by fast relaxation time scale with delta correlations. So here it is effectively an overdamped dynamics with white noise. But a system driven by a colored noise or fractional gaussian noise have inbuilt memory in correlations. Then one have longer persistence lengths. So it ultimately boils down to modelling and the exact problem at hand. Unfortunately I don't understand crypt-decrypt terminology. But you do get a lot of chips. Hope that helps.


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.