Overview¶

What pydae-core does¶

pydae-core takes a symbolic description of a DAE system and turns it into a fast, standalone simulation object. Internally the pipeline is:

        flowchart LR
    A[sys_dict<br/>SymPy expressions] --> B[Builder]
    B --> C[Generated C code]
    C --> D[Compiled .so / .dll<br/>ctypes or CFFI]
    D --> E[Model<br/>runtime API]
    E --> F[ini / run / post<br/>analysis]

The key idea is that building is symbolic and slow, simulation is numerical and fast. You pay the compilation cost once and then run arbitrarily many scenarios through the compiled binary.

Main objects¶

`Builder`¶

Consumes a sys_dict of SymPy expressions and produces compiled C code plus a Python wrapper.

from pydae.core import Builder

bld = Builder(sys_dict, target="ctypes")
bld.build()

`Model`¶

The runtime interface to a built system. Supports steady-state initialisation (ini), time-domain simulation (run), and post-processing (post).

from pydae.core import Model

model = Model("my_system")
model.ini({...}, xy_0={...})
model.run(10.0, {"u_1": 1.0})
model.post()

Diagnostics¶

Jacobian-health utilities (rank, conditioning, sparsity) live in pydae.core.diagnostics and are useful for debugging ill-posed DAEs before trying to run them.

Typical sys_dict shape¶

sys_dict = {
    "name": "my_system",
    "params_dict": {"L": 5.21, "G": 9.81, "M": 10.0},
    "f_list": [...],          # differential equations dx/dt = f
    "g_list": [...],          # algebraic equations 0 = g
    "x_list": [...],          # differential state variables
    "y_ini_list": [...],      # algebraic unknowns at initialisation
    "y_run_list": [...],      # algebraic unknowns during simulation
    "u_ini_dict": {...},      # inputs fixed at ini
    "u_run_dict": {...},      # inputs that may vary during run
    "h_dict": {...},          # output expressions to record
}

See Getting started for a complete worked example.

Where to go next¶

Hands-on walkthrough: Getting started.
API reference: API.
Building a balanced power system instead of hand-writing equations: pydae-bps.
Building an unbalanced distribution system: pydae-uds.