are you sure palace is the right tool for this? it uses insane amount of resources (like most full wave solvers)
Gekko
fastercap, fasthenry and similar more focused/limited simulation tools are enough for a lot of things and are really cheap to run, also palace is really not suited for low frequency stuff so its not even universal
Gekko
imo, simulation probably needs to be like routing currently with a lot of solvers available and AI can choose where to apply what, otherwise its intractable computationally
0hmx
i like the idea of solver that the ai can use to call for doing simulations. i am just not sure if i can validate them all fast enough to ship; with palace things are already validated. and its just about having good intigration.
ig i am also not sure how good is ai at building workflows for itself for simulation. but i like the idea a lot.
fastercap and fasthenry i have not idea how they work i focsed on palace intigration for now.
thanks for the ideas, will try to share one i have something.
This is our planned API for running simulations, you add a component with props as the simulation config, is that what you're thinking?
Gekko
with fasthenry2 + fastercap you get R,L,C (resistance, inductance, capaticance) matrices extracted from trace geometry that you can drop into the already implemented spice netlist tscircuit can use
Seve
^ <@400684445054795796> we should look into this
Want to add to the conversation?
Reply in Discord so your notes stay connected to the source.