Is there such a thing as a circuit (or more generally, system of differential equations) where the output frequency is always orthogonal (anharmonic) to the input?
Example response would be, you have an arbitrary input frequency, and you want to pick an output frequency that *doesn't* line up with the input. If you use a fixed oscillator, it'll work for some inputs, but there will be special frequencies where they start to line up. Suppose that, as the input frequency varies, the output gets pushed away from input harmonics, so that they never line up. A phase-UNlocked loop, if you will!
The behavior in my mind is a slowly varying oscillator, so if you looked at the output on the oscilloscope, it would always be a reasonably clean wave (sine or square), but as you sweep the input, it drifts and hops in just such a way as to avoid the input. Though I think cycle-to-cycle consistency is not a requirement, in which case a possibly simpler solution could be random noise with the input filtered out of it.
An example use would be, suppose you have a two channel analog oscilloscope and you want to add horizontal cursors to it. Easy solution, just put a square wave on the other channel, and as long as it's not triggered, it'll blur into two horizontal lines most of the time. Vary the "high" and "low" voltages and there's your cursors. Trouble is, when the input is one of those lucky frequencies where the trigger lines up, the illusion breaks down. This also applies to CHOP mode in the scope itself. The actual limitation is evidently more sloppy: if they line up within persistence of vision (~20Hz BW), you get a flickering, rolling or locked display.
I have no application in mind, just a Sunday afternoon musing. It seems rather useless, but would probably find application in FDMA or something like that where orthogonality is handy. Actually, such a function probably has even deeper theoretical applications from communications to encryption (e.g., use a discretized version to produce a message digest which is not just a scrambling of the input, but deterministically orthogonal in all dimensions) and number theory. Which means it might fit into one of those "computable but infinite" or "unknowable" categories... hmm...
Tim