Supplementary Code for: Modeling the Binding-Site Barrier in Tumor Spheroids: Arithmetically Stable Admittance Mapping
Description
This repository contains the complete Python codebase required to reproduce the numerical benchmarks and Bayesian inference pipelines presented in the associated manuscript. It includes: The O(N) steady-state admittance propagator. The hybridized Picard-Admittance solver for non-linear Michaelis-Menten kinetics. The Laplace-domain Gaver-Stehfest transient solver. Continuous FEM collocation algorithms used for ground-truth generation. Scripts generating the arithmetic viability maps and MCMC posterior predictive checks.
Files
Steps to reproduce
Steps to reproduce: 1. Environment Setup Ensure a standard Python 3.8+ distribution (such as Anaconda) is installed. The codebase relies exclusively on standard, highly stable open-source libraries. Required packages: numpy, scipy, matplotlib, and emcee (for the MCMC sampler). 2. Execution Protocol Download and extract the provided ZIP archive. Maintain all scripts in the same root directory. The scripts can be executed independently via any standard IDE (e.g., Spyder, VSCode) or command-line interface. 3. Figure Generation and Expected CPU Times The scripts are explicitly designed to reproduce the manuscript's benchmarks. Runtimes are estimated for a standard desktop CPU: To reproduce the deterministic admittance inversion tutorial: Execute toolbox_admittance_propagator.py. Expected runtime: < 1 minute. To reproduce the algorithmic stability benchmark (Table 1): Execute benchmark_fem_convergence_richardson.py. Expected runtime: < 1 minute. To reproduce the non-linear MCMC inference and parameter unobservability: Execute solver_nonlinear_picard_mcmc.py. Expected runtime: 3-5 minutes. To reproduce the transient Gaver-Stehfest penetration dynamics: Execute solver_transient_laplace_stehfest.py. Expected runtime: < 1 minute. To reproduce the temporal algorithmic viability maps: Execute benchmark_stability_viability_map.py. Expected runtime: < 2 minutes. 4. Outputs Upon successful execution, all scripts are hard-coded to automatically generate and export the corresponding publication-ready figures as 1:1 scale .pdf files directly into the working directory. No further post-processing is required.