Learning goals¶
Confirm Python/NumPy fluency for array workflows
Integrate a simple ODE and plot trajectories
Fit a small neural regression model and compute gradients
Choose your bridge¶
Preparation is optional when you already know the missing side. Pick the primer that matches you, then use the notebooks below as drills.
| If this is you | Read |
|---|---|
| New to programming | Programming from zero, then Your first lab |
| Machine learning, new to power systems | Power systems for ML people |
| Power systems, new to machine learning | Machine learning for power-systems people |
| Choosing a longer route | Start here and Learning paths |
Prep package (four notebooks)¶
The four notebooks are also in the left sidebar under Preparation. Today they are stubs (scaffold cells only); expand them before relying on this section for a full course offering.
| Notebook | Focus | Status |
|---|---|---|
| Prep 01 — NumPy basics | Arrays, broadcasting, plotting | Stub |
| Prep 02 — ODE integration | solve_ivp on a damped oscillator | Stub |
| Prep 03 — Tiny neural regression | Tiny MLP curve fit (NumPy or PyTorch) | Stub |
| Prep 04 — Gradients | Manual vs AD gradients | Stub |
Download a local copy: use the notebook page’s download control when the theme shows one, or open the files from the repository folder book/200_prep/. Lab launch paths (local / Binder / Colab) are documented under Running notebooks. Prep notebooks stay stubs for now — Binder/Colab badges for Prep arrive when those notebooks are filled.
Diagnostic¶
After the four notebooks, simulate a damped oscillator and fit a simple regression model to its samples. If that is painful, work through Programming from zero and Your first lab, then return to Lab 1.
Maths refresher (optional, external)¶
If linear algebra or multivariable calculus feels rusty, these external sections from the EECS 245 course notes (Suraj Rampure, University of Michigan; CC BY-NC-ND 4.0) are useful background. We link only — we do not copy or adapt their text (ND forbids distributing derivatives). Read them on their site; then return here for power-system examples.
| Topic | EECS 245 section | Used in SciML for PS |
|---|---|---|
| Partial derivatives | Partial derivatives | Automatic differentiation |
| Least squares | Least squares | Classical estimation, Lab 3 |
| Gradients and gradient descent | Gradient descent, convexity | Training and inverse fits |
| Eigenvalues | Eigenvalues and eigenvectors | Small-signal / modal ideas (later) |
| SVD / low-rank / PCA | Low-rank approximation, PCA | Identifiability, model reduction |
Source: Rampure (2026).
What is not in Preparation¶
ANDES (network power-system simulator) is not part of this section. Lab 1 uses a small SMIB helper without ANDES. You meet ANDES later if a stability / CCT lab or network chapter needs it (separate environment). Tool docs: Reference — Reading list.
Next¶
- Rampure, S. (n.d.). EECS 245 course notes. Retrieved October 14, 2026, from https://notes.eecs245.org/