Hacker News Reader

Comment by dexwiz | Hacker News Reader

This and all the examples linked elsewhere in the thread seem to be electrochemical batteries where the anode or cathode are used as a resistor in another circuit. A current is applied and a metal oxidizes. It’s pretty well known that as a metal oxidizes, it’s resistive properties will change.

As most people know batteries have a limited number of cycles. Every reaction cycle adds some entropy/side reactions, and eventually it will become irreversible. Magnetic storage reorients a crystal or metal, which is more repeatable than a chemical reaction. Our neurons have similar systems to “clean” themselves so they can reset.

I imagine finding a viable memresistor is more about its ability to cycle.

View on HN

(OP here) - the backplane is already a resistor, we're plating aluminum differentially along the resistive backplane to alter resistance depending on current flow. The resistor is X Ohms/cm.. when we plate out Aluminum near the negative terminal, that shorts a section of the resistor, lowering resistance there (making a "shorter" resistor). The aluminum is forced back and forth depending on current flow.

Isn't this just what dexwiz was saying, restated?

No.

Related to this, but why do people claim that the memristor is a "fundamental" circuit element? It seems to me that it's just a generalization of a resistor where the resistance depends on the integral of the current. There are already all sorts of circuit components that have the property where the resistance is a function of X, so why is the memristor particularly deserving of its own class?

Unless I'm missing something the flux-charge equation that "dPhi = M dq" seems to be the same as that of the resistor as well (since dPhi = vdt we have vdt = Mdq => v = M*i which is just ohm's law assuming a constant M).

https://en.wikipedia.org/wiki/Memristor#The_memristor_as_a_f...

> Chua in his 1971 paper identified a theoretical symmetry between the non-linear resistor (voltage vs. current), non-linear capacitor (voltage vs. charge), and non-linear inductor (magnetic flux linkage vs. current). From this symmetry he inferred the characteristics of a fourth fundamental non-linear circuit element, linking magnetic flux and charge, which he called the memristor. In contrast to a linear (or non-linear) resistor the memristor has a dynamic relationship between current and voltage including a memory of past voltages or currents. Other scientists had proposed dynamic memory resistors such as the memistor of Bernard Widrow, but Chua introduced a mathematical generality.

Wikipedia has a pretty good explanation to your question how the memristor completes the quartet of inductor, capacitor and resistor among the differentials I, V, Φ, and q.

The equation for the memristor seems to differ from that of the others though in that R, C, and L are constants whereas M is a function of charge. The explicit charge-dependence of M isn't noted in the diagram, but without it you would just get a resistor.

No, it's no different than the capacitor in that respect. Capacitance is dq/dV.

A (fixed) capacitor has a constant capacitance C just like a fixed memristor has a constant memristance M. Both M and C can be used in differential equations involving charge.

The basic idea is that you cannot "implement" (emulate) memristance using resistors, capacitors and inductors and that is what causes it to be fundamental property of the circuit element.

Since memristance is just the charge-dependent flux I'm not sure you couldn't get it by just throwing a resistor capacitor and inductor into a single circuit. In fact memristance is just the voltage divided by the current, which is impossible to be 0 for any circuit with some voltage. So by this definition almost any circuit has memristance.

Note that it's not unfair to consider any circuit with nonzero memristance as 'containing' a memristor, after all the whole point of fundamental circuit elements is that you can add them to model stuff like wire resistance and stray capacitance. However while you can't create resistance, capacitance and inductance by combining any of the others you can have memristance by combining all 3 (technically you can create any of the 4 by combining the 3 others, but 3 of them can be built easily so those make more sense as fundemental circuit elements).

So while a linear memristor would be rather unusual, I don't quite see what's so fundamental about it as a non-linear circuit element.

> In fact memristance is just the voltage divided by the current

In fact, voltage divided by current is resistance, BY DEFINITION.

Hey, you can also take a bunch of resistors, wire them in a loop and proclaim you have successfully emulated inductor with just resistors. That is what you get if you decide to set aside rules.

Important part of the description of what is memristor is its dependence on past charge flown through it, you can't decide to ignore parts of definition whenever it suits your argument.

>In fact, voltage divided by current is resistance, BY DEFINITION.

For a linear resistor sure, but otherwise resistance is better defined as the slope.

And well if you don't like that definition of memristance I suggest you update the corresponding wikipedia article.

> Hey, you can also take a bunch of resistors, wire them in a loop and proclaim you have successfully emulated inductor with just resistors. That is what you get if you decide to set aside rules.

A circle of resistors has 0 inductance by definition. In practice it's pretty tricky to build such a circuit using 0 inductance wire in an impermeable environment so it remains an approximation.

Key to a memristor is hysteresis -- the cycling of resistance, based on past current. This makes them fundamentally oscillatory.

What I wonder is: if a PID can be modelled with resistor, capacitor, and inductor, what’s giving us the memristor in practical control terms?

Hysteresis?

But the integrative part already takes care of the signal “history”, doesn’t it?

Reply on news.ycombinator.com