How laurentnevou/q_kp_multiband_zb works
Project Type and Purpose
• Type: A scientific computing / simulation tool implemented in MATLAB.
• Purpose: Compute electronic band structures of bulk zinc-blende (ZB) semiconductors using various k·p models (3-, 4-, 6-, 8-, 14-, 16-band).
High-Level Architectural Breakdown
A. Main Control Layer
– kp_multiband_ZB_Main.m: the entry point script that orchestrates the workflow.
B. Parameter Extraction Layer
– ExtractParameters.m: reads material parameters from CSV databases.
C. Database Layer
– materialDB.csv, materialDB_kp16bands.csv: contain preset parameters for III-V and II-VI semiconductors.
D. Model Implementation Layer
– Library.m: core helper functions (e.g., numerical utilities, matrix builders).
– kZB_f.m: generic zinc-blende Hamiltonian builder.
– kp_*_f.m (e.g., kp_3bands_DKK_f.m, kp_8bands_Kane_Fishman_f.m, kp_16bands_Luttinger_Fishman_f.m): individual k·p model functions that assemble and diagonalize the Hamiltonian for each band model.
E. Visualization / Output Layer
– MATLAB built-in plotting routines invoked by the main script to display band-structure graphs.
Relationships and Data Flows
1. User invokes kp_multiband_ZB_Main.m (Main Control).
2. Main Control calls ExtractParameters to read one of the CSVs → returns physical constants & k·p parameters.
3. Main Control selects the appropriate kp_*_f model function based on user input (number of bands & model variant) and passes parameters to it.
4. The model function may in turn call Library.m utilities (e.g., matrix operations).
5. kp_*_f returns eigenvalues (band energies) vs. k-points to Main Control.
6. Main Control feeds results into MATLAB’s plotting engine and displays or exports the figures.
Architectural Patterns and Design Principles
• Layered separation:
– Data layer (CSV database)
– Logic layer (parameter extraction + model functions + utility library)
– Presentation layer (main orchestration + plotting)
• Extensibility points: new kp_*.m modules can be dropped in to add more band models.
• Single-process, monolithic MATLAB application—no microservices, but well-modularized by function.