In article <83mci7F4suU3@mid.individual.net>, Dirk Bruere at NeoPax <firstname.lastname@example.org> wrote:
>Maybe you could do a bit on memristors?
Here's what I said about them in "week294":
Here's the last 1-port I want to mention:
5. The "memristor". This is a 1-port where the momentum p is a function of the displacement q:
p = f(q)
The function f is usually called the "memristance". It was invented and given this name by Leon Chua in 1971. The idea was that it completes a collection of four closely related 1-ports. In "week290" I listed the other three, namely the resistor:
p' = f(q')
q = f(p')
and the inductor:
p = f(q')
The memristor came later because it's inherently nonlinear. Why? A *linear* memristor is just a linear resistor, since we can differentiate the linear relationship p = Mq and get p' = Mq'. But if p = f(q) for a nonlinear function f we get something new:
p' = f'(q) q'
So, we see that in general, a memristor acts like a resistor whose resistance is some function of q. But q is the time integral of the current q'. So a nonlinear memristor is like a resistor whose resistance depends on the time integral of the current that has flowed through it! Its resistance depends on its history. So, it has a "memory" - hence the name "memristance".
Memristors have recently been built in a number of ways, which are nicely listed here:
Electrical engineering journals are notoriously resistant to the of open access, and I don't think there's a successful equivalent of the "arXiv" in this field. Shame on them! So, you have to nose around to find a freely accessible copy of Chua's original paper on memristors:
Memristors supposedly have a bunch of interesting applications, but I don't understand them yet. I also don't understand "memcapacitors" and "meminductors". The above PDF file also contains a New Scientist article on the wonders of these.