Skip to content

debug_guide.md

Latest commit

 

History

History
125 lines (95 loc) · 3.8 KB

debug_guide.md

File metadata and controls

125 lines (95 loc) · 3.8 KB

User Guide for the Notebook src/debug.ipynb

GeoclassificationMPS Project
Branch: debug

📝 Summary

This notebook is designed to test and debug the geostatistical simulation pipeline within the GeoclassificationMPS project. It enables configuration of simulation parameters, generation of Training Images (TI) and Conditioning Data (CD), execution of the simulation via DeeSse, and subsequent analysis and visualization of results.

🗺️ General Structure

1. Initialization and Imports

from launcher import *
from interface import *
from geone.imgplot import drawImage2D
import os
os.chdir(r"C:\users\amen0052\documents\personal\geoclassificationmps")
print("Current directory:", os.getcwd())

2. Simulation Parameters

Key parameters to adjust as needed:

  • arg_seed: Random seed.
  • arg_n_ti: Number of Training Images.
  • arg_ti_pct_area: Percentage of area used for TI.
  • arg_num_shape: Number of shapes (for TI generation).
  • arg_aux_vars: Auxiliary variables.
  • arg_output_dir: Output directory.

3. Advanced Configuration

The central function get_simulation_info_custom() gathers all necessary parameters:

  • Data CSV path
  • TI generation methods available:
    "DependentCircles", "DependentSquares", "IndependentSquares", "Customised", "ReducedTiSg"
  • Management of auxiliary, conditional, and simulation variables
  • DeeSse settings (number of realizations, threads, etc.)
  • Options for saving and visualizing outputs

🛠️ Practical Usage Guide

⚙️ 1. Prepare Your Environment

  • Ensure all required files (.csv, .npy, custom masks) are present in the expected folder (./data/).
  • Adjust file paths as needed in the import and configuration cells.

📋 2. Define Your Parameters

In the relevant cell:

verbose = True
arg_seed = 123
arg_n_ti = 1
arg_ti_pct_area = 90
arg_num_shape = 15
arg_aux_vars = ["grid_lmp", "grid_grv"]
arg_output_dir = "./output/tmp"

Modify these values according to your study case.

🏗️ 3. Configure the Simulation

Call the function:

params, shorten, nvar, sim_var, auxTI_var, auxSG_var, condIm_var, names_var, types_var, outputVarFlag, nr, nc = get_simulation_info_custom(
    arg_seed, arg_n_ti, arg_ti_pct_area, arg_num_shape, arg_aux_vars, arg_output_dir)

🧩 4. Generate Training Images and Conditioning Data

Select the TI generation method, for example:

ti_methods = params['ti_methods']
if "DependentSquares" in ti_methods:
    # Generation and visualization...

Other methods are available based on your requirements.

🖼️ 5. Visualize Masks and Training Images

Use plotting functions to display intermediate results:

plot_mask(ti_frame_DS[0], masking_strategy="Dependent Squares")
drawImage2D(ti_list[0], iv=0)
save_plot("aux_var")

▶️ 6. Run the DeeSse Simulation

Prepare the simulation object:

deesse_input = gn.deesseinterface.DeesseInput(...)
deesse_output = gn.deesseinterface.deesseRun(deesse_input, nthreads=nthreads, verbose=2)

💾 7. Save the Results

if saveOutput:
    save_simulation(deesse_output, params, output_directory=deesse_output_folder_complete)

📊 8. Analyze Indicators

Compute and visualize entropy, divergence, etc.:

ent, dist_hist, dist_topo_hamming = calculate_indicators(...)
std_ent, realizations_range1 = calculate_std_deviation(ent, 1, numberofmpsrealizations)
plot_realization(deesse_output=deesse_output)

🧠 Recommendations and Best Practices

  • File Paths: Enter your own project path in the first cell.
  • Documentation: Follow the notebook’s in-line examples and formats for preparing your data files (.csv, .npy).
  • Modularity: Experiment with different TI generation methods to compare their impact on results.