How to Read This Site
Purpose
Seção intitulada “Purpose”This chapter is a navigation guide. It describes the site’s sidebar sections, explains what each section covers, and suggests reading paths for different readers. A reader deciding whether this site matches their needs should read this chapter first.
1. Site Structure
Seção intitulada “1. Site Structure”The site is organised into the sections shown in the sidebar, in this order: Get Started, Model entities, Methodology reference, Running studies, Execution reference and Worked Examples. Each reference section follows the section it complements: Methodology reference follows Model entities, and Execution reference follows Running studies. Both reference sections start collapsed.
Get Started answers what Novomodelo computes (What Novomodelo Solves) and holds How the model decides, the model’s decision loop in one page, and this navigation guide. Readers deciding whether Novomodelo fits their problem should start here.
Model entities explains the model one entity at a time, with one page for each of the 16 entities: what each entity is, what the model decides about it and where its formulas and files are. Four subsections follow the columns of the figure on the home page: Inflows (Inflows and the PAR(p) model, Scenarios and openings), Generation (Hydro plants, Thermal plants, Non-controllable sources, Pumping stations), Network (Buses and subsystems, Lines and interchanges, Energy contracts) and Consumption (Load, Deficit). The time structure, the water value and the rules of the game (generic constraints, penalties and risk aversion) close the section.
Methodology reference holds the five methodology blocks described below, followed by the Glossary and the Bibliography. Readers of the entity pages turn to it for the complete formulation behind each entity.
Introduction (the first block of Methodology reference) establishes the conceptual and notational foundation for the rest of the site: a one-page framing of the SDDP algorithm for readers new to stochastic dynamic programming (The SDDP Framework in One Page), the complete notation used throughout (Notation Conventions), and, as a companion, this navigation guide (kept in Get Started). Its chapters can be read in order or consulted individually.
System Modelling defines the power system entities and the stage LP that Novomodelo solves at each iteration. This block is the mathematical foundation for everything that follows: hydro plants, thermal units, transmission lines, penalty structures, the stage LP and the state it carries between stages. Readers who want to understand how the physical system is represented in the optimisation should begin here.
Stochastic Modelling covers the inflow uncertainty model. Novomodelo uses periodic autoregressive models (PAR) to generate scenario trees; this block explains the policy graph that links the stages, the PAR model structure and its fitting procedure, the treatment of negative inflow realizations, the scenario generation pipeline, and studies that mix stage resolutions. Readers interested in how uncertainty is handled, without being concerned with the optimisation algorithm, can read this block independently of The SDDP Algorithm.
The SDDP Algorithm is the core of the methodology. It covers the forward and backward passes, cut generation and management, the LP layout and its scaling, warm-start strategies, risk measures, stopping rules, upper-bound evaluation, and determinism guarantees. Readers coming from the SDDP literature will find the connections to standard treatments here; readers new to SDDP should read Introduction and System Modelling first.
Coupling & Boundary Conditions covers how a study’s horizon is closed: the horizon mode, the post-study boundary that values the state left at the end of the horizon (chained studies included), and how discounting interacts with the cut approximation. See Horizon Modes for the entry point.
Running studies follows the steps of a study. Installation opens the section. Execution modes covers the three ways to run Novomodelo: from the CLI (Quickstart), from Python (Python Quickstart) or through an AI agent (AI Agents). Preparing the case covers the input files (Case Format), the run settings (Configuration) and converting an existing case (Case Conversion). Running the study covers training and simulation (Running Studies) and reusing a trained policy (Policy Management). Performance and scale covers Performance Accelerators and HPC & Cluster Deployment. Analysing results covers the output files (Output Format) and reading convergence (Convergence & Diagnostics). Troubleshooting closes the section and covers failures at every step. Readers who run studies return to this section after their first study.
Execution reference holds the lookup pages for running studies: the Generic Constraints language, the JSON Schemas, the Error Codes, the FlatBuffers Policy Schema, the CLI Reference and the Python API. Use this section as a lookup resource, not as a starting point.
Worked Examples provides two pedagogical walkthroughs of the SDDP loop: a single-reservoir toy case and a four-reservoir toy case, both small enough to verify by hand. These chapters trace the forward pass, the backward pass, cut construction and the lower-bound update with concrete numbers. The novomodelo repository ships the runnable cases examples/1dtoy and examples/4ree; each walkthrough states how its numbers relate to its case. Readers learning SDDP by doing should start with the worked examples after reading Introduction.
2. Tools and Interfaces
Seção intitulada “2. Tools and Interfaces”Novomodelo runs from the novomodelo command-line program and from the novomodelo-python package. What Novomodelo Solves §4 describes both interfaces and the files a run writes. The methodology chapters use “case directory” and “configuration” as concrete terms for the artefacts a user manages; these map directly to both interfaces.
Readers interested in the design commitments that underpin these interfaces — reproducibility, determinism, declaration order invariance and agent-readability — should read What Novomodelo Solves §5.
3. Reading Paths
Seção intitulada “3. Reading Paths”Different readers enter this site from different directions. Find the row that matches you:
| Reader | Reading path |
|---|---|
| Run studies | Installation, then the Quickstart, then the Running studies section, starting with Running Studies and Convergence & Diagnostics. |
| Coming from other software? | What Novomodelo Solves §6, then Converting an existing case; for equivalent terms in other planning tools, see the Glossary. |
| Work in Python | The Python Quickstart, then the Python API reference. |
| New to SDDP | Read What Novomodelo Solves and all of Introduction, in Methodology reference, then read System Modelling to understand the LP, then read The SDDP Algorithm for the algorithm itself. The worked examples in Worked Examples reinforce the concepts. |
| Familiar with SDDP, new to Novomodelo | Skim What Novomodelo Solves and Introduction, read Notation Conventions carefully, then read The SDDP Algorithm. Cross-reference System Modelling for the stage LP and its state when The SDDP Algorithm refers to stage variables. |
| Looking for a specific topic | Use the sidebar to navigate to the relevant section. The SDDP Algorithm covers the algorithm; System Modelling covers the LP and system model; Stochastic Modelling covers the inflow uncertainty model; Coupling & Boundary Conditions covers boundary conditions. Notation Conventions resolves notation questions, and Running studies covers the software in use. Execution reference covers the CLI, the Python API, the schemas and the error codes. |
Cross-References
Seção intitulada “Cross-References”- What Novomodelo Solves — the problem statement, algorithm name, methodology guarantees, user-facing capability summary, and an orientation for readers coming from other software
- The SDDP Framework in One Page — one-page algorithmic framing for readers new to stochastic dynamic programming
- Notation Conventions — complete symbol table for index sets, parameters, decision variables, and dual variables
- System Element Modeling Overview — the elements of the system and their variables: System Modelling entry point
- PAR(p) Inflow Model — periodic autoregressive inflow model: Stochastic Modelling entry point
- SDDP Algorithm — forward pass, backward pass, cut generation, convergence: The SDDP Algorithm entry point
- Horizon Modes — boundary conditions and horizon-mode design: Coupling & Boundary Conditions entry point