add ability to not distribute lumped elements

[dmarek-flex]

At high frequencies and mostly in the case of high impedances, distributing the lumped element over multiple grid cells breaks the assumption that the electric field is uniform within the lumped element. We now offer the ability to restrict the network portion of the lumped element to one grid cell and use automatically created PEC connections to connect the network portion of the lumped element to the bounds of the lumped element. We also have the ability restrict the network portion along the voltage axis but distribute the element laterally.

These PEC connections introduce parasitic inductances and capacitances that depend on the discretization used and which effectively modify the simulated network. In some cases, it will be possible to account for these, so we offer estimates for the values of these inductances and capacitances.

Example usage has been added to the associated notebook tutorial.

[dmarek-flex]

Here are some pictures of the different strategies we have for distributing lumped elements at the moment:

DistributeSpec.Off and lateral size 0 gives wire type connections (high parasitic inductance)

DistributeSpec.Off and lateral size !=0 gives sheet type connections (lower inductance, higher capacitance)

DistributeSpec.LaterallyOnly and lateral size !=0 gives sheet type connections and distributes network portion (slightly less parasitic inductance)

DistributeSpec.On this is the default approach and has the lowest parasitic inductance and capacitance

Overall DistributeSpec.On seems to be the best approach in most cases. The exception is when you would like to model a network with a high inductance at high frequencies, then the parasitic inductance can be accounted for in the network model and its effect minimized.

[dmarek-flex]

Let's start with an RF team round of reviews.

[dmarek-flex]

And here is a circuit model for the parasitic effects that the connections add:

$Y(s)$ is the network that you would like to model in the lumped element.

[weiliangjin2021]

First pass through lumped_element. All look quite clear! Will take another look after break

[QimingFlex]

Looks great, @dmarek-flex! I have a few questions, which I’ll address as I finish reviewing the code.

[dmarek-flex]

Got a basic structure question for @daquinteroflex @tylerflex @yaugenst-flex . You will notice in this PR I added a tools/microwave directory where I plan on adding small standalone type calculations for microwave specific applications.

For example, I plan on moving things like the path integrals from the microwave plugin to this folder in the future. The reason being is that I find that I will need to use these calculations when adding basic components like Voltage or Current monitors, so making those tools a part of a plugin doesn't really make sense to me.

[daquinteroflex]

So I can't actually respond to this publicly so I've tagged the private issue to discuss this

One approach is simply stick it in the plugins (for now) till we begin this migration effort. This is my preference actually @dmarek-flex if that sounds ok. We can do the migration effort together in another PR when the new structure is more defined if that sounds ok? What's your timeline and we can plan to do it for this since Charge is wrapping up?