Why does Landauer define erasure of information as resetting

[voila]

Hi. I hope this is the right forum for the question.

Landauer's principle says that the erasure of one bit of information (1 or 0) requires k_BTln2 of heat production. When he talks about erasure, it means "resetting back to 0 (or 1)" after measuring. He says something along "this is needed as to not to bias the device". I guess this means "for the device to perform a measure, it needs to previously be in a given state". So that if you need to measure 0's and 1's, it intrinsically needs to be restored to 0 before each new measure. Why?

 

[rcgldr]

Wiki article seems to have a different interpretation, and note that there are challenges, which my guess depends on the type of device used to store information.

http://en.wikipedia.org/wiki/Landauer's_principle

A web search finds other articles, such as this one.

http://www3.imperial.ac.uk/pls/portallive/docs/1/55905.PDF

 

[voila]

I have an answer to the question, but still there's something quite open to me. Erasure means resetting so that it doesn't interfere the device operation. For example, if a device with a one bit memory controls a door and is instructed to "OPEN DOOR whenever a measure gives 1", it wouldn't work if it wasn't resetted to 0.

However, focus now on the picture I'm uploading:

The right side of the red line is the resetting procedure, which resets a bit to 0 no matter what its state is. Calculations have shown this procedure implies, on average, a work of W=kTln2 on the particle (particle is brownian, and so disippates it into heat).

The left side is what Bennett proposed for a measure of a bit of information, saying measurement requires no work done. Worry not about the procedure he thought of, the important part is that it can be represented as a double well potential being modulated like is indicated next to the figure.

...Notice something similar? I find both modulations of the potential to be quite the same, yet the left one is an example of no expenditure of energy and the right one is just the opposite. What am I missing?

Edit: "Medida" means measurement.

 

[Dominik Tugend]

I am a noob, but I have a feeling that you might need to define what measurement means for you:

https://en.wikipedia.org/wiki/Uncertainty_principle

It must be emphasized that measurement does not mean only a process in which a physicist-observer takes part, but rather any interaction between classical and quantum objects regardless of any observer

https://en.wikipedia.org/wiki/Observer_effect_(physics)

 

[voila]

Yes, I know. A process where the memory of the device takes a state (1 or 0 if its a one bit memory) which corresponds to a given physical variable of the system being measured.