circuit simulation in 2026: there's got to be a better way!
I'm working on a project now where I want to programmatically generate a complicated electrical circuit (its actually serving as a model of a physical process), and I'm finding modern workflows for this surprisingly difficult. In python, there seems to be 3 options:
pyspice, pyltspice, and scikit-rf.
pyltspice is nothing more than a thin wrapper for ltspice. I still have to resort to ltspice's janky scripting language and netlist nomenclature which largely undermines its value as a python tool. Additionally, it hasn't been maintained in years.
I've played around with pyspice and it's a bit better - it does employ a somewhat more modern object-first programming approach. However it exposes only very limited analysis and plotting options meaning I have to code all that up myself. In particular, getting device currents out has proven remarkably arcane. I have to insert all these dummy voltage sources everywhere (which by the way end up getting assigned weird and inconsistent net names, so yay me if I change the topology) that make for ugly AF netlists and schematics. Oh, and its no longer maintained either.
Ok, scikit-rf. I've actually used this quite a bit for RF projects and for that its great. But as it uses an n-port architecture, working with discrete circuits is a clunky mess. Any component you might want to probe has be treated as a 2-port and doing any kind of ac/dc/transient analysis involves a bunch of annoying extra transformation steps (simulation of s parameters, vector fitting, etc). Ugghgh. Its a good package but its just not designed for this.
tl;dr: I don't want to use spice and netlists...why are there no python-first circuit simulation tools available in 2026? Would anybody be interested in a library offering a modern, python-first modeling workflow?