Domain Overview

Domains are plugins that encapsulate the evaluation logic, physics models, and default configurations for a specific class of materials. MatCraft ships with 16 built-in domains spanning energy, environmental, structural, and electronic materials.

What a Domain Provides

Each domain plugin implements:

Evaluation function: Maps parameter vectors to objective values using physics models, empirical correlations, or pre-trained ML models.
Default parameters: Sensible parameter definitions and ranges for the material class.
Default objectives: Common optimization targets for the domain.
Validation rules: Domain-specific checks (e.g., composition constraints for alloys).
Templates: Pre-built MDL files for common optimization scenarios.

Built-In Domains

Using a Domain

Specify the domain in your MDL file:

yaml

domain: water

Or when initializing a project:

bash

materia init my-project --domain battery

Listing Available Domains

bash

materia init --list-domains

Output:

Available domains (16):
  water          Water membrane optimization
  battery        Lithium-ion battery design
  solar          Solar cell efficiency
  catalyst       Heterogeneous catalysis
  hydrogen       Hydrogen storage materials
  thermoelectric Thermoelectric generators
  polymer        Polymer blends and composites
  ceramic        Structural and functional ceramics
  superconductor High-temperature superconductors
  steel          Steel alloy composition
  aluminum       Aluminum alloy optimization
  glass          Glass composition design
  concrete       Low-carbon concrete
  adhesive       Structural adhesives
  biomaterial    Biocompatible materials
  coating        Protective coatings

Domain Evaluation Models

Each domain uses one or more evaluation approaches:

Analytical models: Closed-form equations derived from physics (e.g., Hagen-Poiseuille for membrane permeability).
Empirical correlations: Data-driven regression models fitted to published experimental data.
Pre-trained ML models: Neural networks trained on DFT databases or large experimental datasets.
Hybrid models: Combinations of physics-based and data-driven components.

The evaluation function is transparent — you can inspect the source code of any built-in domain in the materia.plugins package.

Custom Domains

If none of the built-in domains matches your problem, you can create a custom domain plugin. See Custom Plugin Development for a step-by-step guide.

Domain-Specific Documentation

Detailed documentation for each domain is available: