Skip to content

SimWeave

A hybrid discrete/continuous simulation engine for Python. SimWeave gives you a single environment in which differential equations, queueing systems, A* agents over graphs, multi-SKU warehouses, currency-aware money flows and Monte Carlo replication all share the same clock.

Why pick SimWeave over a single-paradigm library

Most Python sim libraries do one thing well: SimPy is a process-based discrete-event scheduler, SciPy gives you ODE integration, NetworkX gives you graphs, and so on. SimWeave's pitch is the seam between them:

  • Hybrid by default. A continuous mass-spring system, a queueing network, and an A* agent can all share the same SimEnvironment and the same dt. You don't have to bridge two clocks or two event loops.
  • Fixed-shape outputs that aggregate trivially under Monte Carlo. Every simulate() call returns a SimulationResult with predictable array shapes, so 1 000 replicates collapse into a single (n_runs, n_time) ndarray that the fan-chart helper consumes directly. No pandas wrangling, no per-run schema drift.
  • Cheap core, opt-in extras. The base install is just NumPy. Plotly, SciPy, NetworkX, Numba, Babel and OSMnx all sit behind named extras — the package stays small even when the surface area is broad.
  • Plotly-first visualisation that round-trips through JSON. Every helper returns a plotly.graph_objects.Figure, which means a downstream web frontend can render the same chart natively without re-implementing it.
  • Composable primitives. Queues, services, generators, agents, warehouses and recorders all subclass one Entity interface. New domain objects plug in without touching the scheduler.

Getting started in 60 seconds

pip install simweave[viz]

import numpy as np
import simweave as sw

# Continuous: a damped mass-spring system
msd = sw.MassSpringDamper(mass=1.0, damping=0.4, stiffness=4.0)
res = sw.simulate(msd, t_span=(0.0, 12.0), dt=0.01, x0=np.array([1.0, 0.0]))

# Plot it (requires the [viz] extra)
fig = sw.plot_state_trajectories(res, title="Damped MSD")
fig.write_html("msd.html", include_plotlyjs="cdn")

Fleet operational availability

SimWeave's reliability module lets you model fleets of assets — taxis, military vehicles, industrial machines — where each vehicle has subsystems that can fail, consume parts from a warehouse, and pass through a repair centre. The stacked area chart below shows a simulated year for an 8-vehicle taxi fleet with real calendar dates on the x-axis, produced by a single line:

fig = sw.plot_fleet_availability(recorder, time_axis=sw.SimTimeAxis("2027-01-01", "days"))

Where to start

Worked demos

The demos/ folder in the repository is the fastest way to see SimWeave end-to-end. Every script is self-contained — pick one and run it directly, e.g.

pip install simweave[dev]
python demos/01_simple_queue.py

The two-digit prefix is just a stable ordering hint; the scripts have no dependency on each other.

Demo Topic
01_simple_queue.py M/M/1 queue with arrivals + a sink
02_chained_services.py Two services in series
03_resource_pool.py Multi-channel service capacity
04_supply_chain_basic.py Warehouse + reorder logic
05_supply_chain_optimise.py Poisson + DE cost optimisation
06_agent_astar.py A* agent on a grid
07_monte_carlo.py Replication and MCResult
08_hybrid_continuous_discrete.py Continuous physics + discrete events
09_mass_spring_damper.py Canonical MSD
10_quarter_car.py Two-mass road model
11_series_rlc.py LCR electrical circuit
12_thermal_system.py Thermal RC network
13_money_cashflow.py Money + FX over time
14_viz_tour.py Every plot helper, end-to-end
15_units_dimensional.py SI-units algebra (e.g. m/s = velocity)
16_inventory_optimisation.py Optimisation from estimated demand
17_half_car_pitch.py Extending suspension to front and rear
18_half_car_roll.py Extending suspension to left and right
19_full_car_dynamics.py Extending suspension to 7 DOF
20_PID_thermal_system.py PID control in thermal RC example
21_reliable_fleet.py Taxi fleet MC + stacked availability
22_sensitivity_analysis.py 2-D parameter sweep + 3-D surface
23_time_axis_calendar.py Calendar dates on plot axes
24_signalised_intersection.py 4-way crossroads with traffic signal
25_roundabout.py UK roundabout vs signalised comparison

Status

SimWeave is currently <v1.0.0 The public surface re-exported from simweave is considered stable; submodules may add features but are unlikely to break existing imports. See the design notes for the rationale behind the major subsystems.