More precisely, we show that any randomized one-sided-error bounded Monte Carlo algorithm for these problems must perform W(log N ) random accesses to external memory devices, provided that the internal memory size is at most O( 4√N /log N ), where N denotes the size of the input data.
Randomized Computations on Large Data Sets: Tight Lower Bounds
He needs to ensure that Eve cannot pre-calculate classical data that she can supply to her devices in order to pass his tests—otherwise she can predict all the output data that will be generated for any given random input, and so he cannot generate any new private randomness.
Private Randomness Expansion With Untrusted Devices
FNU-CIS would be deployed as thin client web-based application. Thin client applications display data collects input from the user and post it back to the server for processing. Thin clients don’t require installation on user devices.
Thin Client Web-Based Campus Information Systems for Fiji National University
***