Changelog • multimod

multimod 0.4.0.9007 (development version, 2025-12-08)

Model workspace system (save_model(), load_model()):
- Apache Arrow IPC format (fast, type-safe, binary)
- Apache Parquet format (compressed, columnar)
- CSV format (human-readable, portable)
- Lazy loading support for large datasets
- See vignette("model_workspace") for details
Data import and linking:
- import_energyRt_data(): Import from energyRt package structures
- populate_sets_from_scenario(): Extract set members from parameters
- link_scenario_data(): Link to Arrow/Parquet files for lazy loading
- Support for OSeMOSYS example model data
Flexible data modes in generated code:
- Embedded: Data included directly in generated Python/Julia code (for small models)
- External: Code connects to Arrow repository (for large models)
Example datasets: Consolidated into unified example_models structure
- Access: data(example_models) provides both energyRt and OSeMOSYS examples
- Structure: example_models$energyRt$multimod, example_models$OSeMOSYS$gmpl, etc.
- Models used: OSeMOSYS-Utopia (standard test case), energyRt-DEMO (BASE_UTOPIA scenario)
- Old datasets removed; use example_models$energyRt$multimod instead of utopia_multimod

Pyomo support: Complete Python/Pyomo code generation
- Abstract model design with external data loading
- Multiple solver support (HiGHS, Gurobi, CPLEX, etc.)
- Both embedded and external data modes
- See vignette("pyomo") for workflow details
JuMP improvements (write_jump()):
- Modern Julia/JuMP syntax with named constraints
- Tuple indexing: eqName[(h,r,y) in mapping]
- Short index aliases: h (tech), r (region), c (comm), y (year), ts (slice)
- Clean get() pattern for parameter access with defaults
- Variable bounds from GAMS types: >= 0, <= 0, Bin, Int, free
- Comprehensive diagnostics with CSV logging
- Both embedded and external data modes
GMPL support (write_gmpl(), write_gmpl_data()):
- Full GMPL/MathProg syntax support
- Proper formatting and indentation
- Lazy loading from Arrow/Parquet datasets
- Memory-efficient data writing

Configuration system: Unified configuration for solver paths with 4-tier priority:
- R options > Environment variables > YAML config > Auto-detection
- New functions: get_multimod_python(), get_multimod_julia(), get_multimod_glpsol()
- YAML support: multimod_config_write(), multimod_config_read(), multimod_config_show()
- See vignette("configuration") for details
In-memory parsing: All reader functions (read_gams(), read_gmpl(), read_gmpl_data(), import_gmpl_data()) now accept character vectors in addition to file paths, enabling direct parsing from R objects without temporary files.

Parameter folding: Automated dimensionality reduction for uniform parameters
- fold_model(), create_fold_spec(), get_fold_summary()
- See vignette("folding") for details
Model trimming: Remove unused sets, parameters, variables, and equations
- trim_model(), get_trim_summary()

JuMP fix: 1-dimensional mappings now correctly loaded as Set{String} instead of Set{Tuple{String}}. This was causing membership checks like t in mTradeCapacityVariable to silently fail, resulting in missing constraint terms and incorrect model formulations.
- 1D mappings: Set{String} for direct membership checks
- Multi-D mappings: Set{Tuple{...}} for tuple membership
- Parameters follow same pattern for consistent dictionary key types
- Generated models now produce identical results to reference implementations
Export functions: Updated export_model_source() to use new example_models structure
- export_osemosys() and export_energyrt() now work with consolidated datasets

New vignette: vignette("configuration") - Configuration system and solver setup
New vignette: vignette("pyomo") - Complete Pyomo workflow
New vignette: vignette("model_workspace") - Model workspace and data management
Updated vignette: vignette("jump") - Complete JuMP workflow
Updated vignette: vignette("gmpl") - Complete GMPL workflow
Updated vignette: vignette("latex") - LaTeX generation with new dataset
Updated: Development roadmap with realistic implementation status