pulsim 1.3.0

High-performance circuit simulator for power electronics

Pulsim

Power-electronics circuit simulator — C++23 kernel with a Python-first API.

Pulsim is a header-only C++ simulation engine (pulsim) wrapped by a flat Python module (import pulsim). It is built around a PWL state-space cache (for fast switched-converter dynamics), a Newton refinement on top of the cached linear factor (for non-linear devices), and a built-in event detector for diode/MOSFET commutations.

If you have legacy code from the pre-1.0 cycle (the old Circuit / Simulator / YamlParser surface), see docs/migration-guide.md for the device / analysis mapping table.

Why Pulsim

• PLECS-style PWL cache — switched-converter steady-state in milliseconds instead of minutes.
• Header-only C++23 kernel — drop pulsim/ into your own CMake target via pulsim::core; no static-library link step.
• Python-first ergonomics — CircuitBuilder API takes string node names and SI-unit parameters, returns the same SimulationResult whether you run a transient, an AC sweep, or a parameter sweep.
• Mixed-domain composable control — MixedDomainBlockChain runs PI/PID, comparators, rate limiters, op-amps, FOC blocks and thermal networks at kernel speed (no Python interpreter cost per step).
• Frequency-domain analysis included — small-signal MNA Bode + swept-sine FRA + closed-loop GM/PM measurement, all in the same surface.

Quick start

Build prerequisites

Pulsim has just two native dependencies — Eigen 3.4+ and a C++23 compiler. Everything else (sparse LU, YAML parser, tests) is header-only or vendored at configure time:

Dependency	Why
Eigen 3.4+	Header-only sparse linear algebra (matrices, vectors). Pulsim ships its own sparse LU solver on top — no third-party LU library is linked.
C++23 compiler	AppleClang 15+ / Clang 17+ / GCC 13+

Install on the supported platforms:

# macOS (Homebrew)
brew install cmake ninja eigen

# Debian / Ubuntu
sudo apt-get install -y cmake ninja-build libeigen3-dev

# Fedora
sudo dnf install cmake ninja-build eigen3-devel

Build + run

git clone https://github.com/lgili/Pulsim.git
cd Pulsim

# Build kernel + Python extension
cmake -S . -B build -G Ninja \
  -DCMAKE_BUILD_TYPE=Release \
  -DPULSIM_BUILD_PYTHON=ON
cmake --build build -j

# Use Pulsim from the source tree (no `pip install` required)
export PYTHONPATH="$(pwd)/build/python:$PYTHONPATH"

# First run — open-loop buck
python3 examples/scripts/run_buck.py

# Print available components + helpers
python3 -c "import pulsim as p; p.catalog()"

First simulation — 8 lines

import pulsim as p

b = p.CircuitBuilder()
b.add_voltage_source("Vin", "n0", "gnd", 5.0)
b.add_resistor      ("R1", "n0", "vc",   1000.0)
b.add_capacitor     ("C1", "vc", "gnd",  1e-6)

res = p.simulate(b, t_end=5e-3, dt=1e-5)
p.plot.scope(b, res, signals=["vc"])      # one-liner plot

What ships

• Builder: CircuitBuilder with 20+ helpers covering passives, sources, MOSFETs (SH1), IGBTs (Level 1), saturable inductors, transformers, op-amps, three-phase sources, …
• YAML loader: same surface, 13 ready-made example circuits in examples/.
• Solver: PWL state-space cache + Newton refresh + event detection + optional state-aware step_observer(t, x) callback for closed loops.
• Control library: PIController, PIDController, Comparator, RateLimiter, FirstOrderLowPass, …
• AC analysis: swept-sine Bode + auto-tuning (tune_pi_from_bode) with phase-margin / gain-margin extraction.
• Frequency-response analyser (FRA): closed-loop / nonlinear-Bode via the time-domain swept-sine path.
• Plot helpers: p.scope(), p.plot_bode() — one-line waveform + Bode plots with sensible defaults.

Where to learn more

• 6 narrative tutorials: docs/tutorials/
• Mental model: docs/mental-model.md
• API reference: docs/api-reference.md
• Gotchas: docs/gotchas.md
• 20 runnable scripts: examples/scripts/

Validation

# Python runtime tests
PYTHONPATH=build/python pytest python/tests -v

# C++ kernel tests (layer-by-layer Catch2 binaries)
ctest --test-dir build --output-on-failure

Reference CSV traces for the converter showcases live in benchmarks/baselines/; a 1.0-native regression runner that consumes them is not yet wired up.

Documentation

• Documentation site: https://lgili.github.io/Pulsim/
• Migration guide (from the pre-1.0 surface): docs/migration-guide.md

Docs deployment (GitHub Pages)

Docs are published by .github/workflows/docs.yml using MkDocs Material + mike:

• PR: strict docs build
• main: deploy dev docs channel
• vX.Y.Z tag: deploy release docs and update latest

In repository settings, set Pages Source to GitHub Actions.