Simulation data for inhibitory budget matching in competitive STDP networks

This dataset contains the simulation outputs required to reproduce the analyses and figure panels for the preprint “Inhibitory budget matching constrains homeostatic plasticity in competitive spiking networks” by Fabrizio Musacchio and Martin Fuhrmann (bioRxiv, forthcoming). The dataset accompanies the GitHub repository:https://github.com/FabrizioMusacchio/inhibitory_budget_stdp Included files: long_30k_sweeps_manifest_HPC.zipPrimary 30,000-training-example and 5,000-test-example HPC sweep outputs. These runs were used for the fixed-inhibition reference grid, matched and mismatched inhibitory-plasticity comparisons, and the Vogels-style inhibitory STDP parameter map. slow_budget_maps_30k_manifest_HPC.zipAdditional 30,000-training-example HPC parameter maps for slow homeostatic inhibition and budget-constrained slow homeostatic inhibition. long_visualization_60k_2ep.zipLonger fixed-inhibition diagnostic run used to evaluate whether extending training beyond the 30,000-example, one-epoch protocol materially changes the qualitative receptive-field and performance conclusions. To reproduce the preprint figure panels, download and extract these archives into the runs directory of the code repository so that the following folders exist: runs/long_30k_sweeps_manifest_HPC/runs/slow_budget_maps_30k_manifest_HPC/runs/long_visualization_60k_2ep/ Then run the figure-generation script from the repository root: MPLCONFIGDIR=/tmp/mpl conda run -n diehl_cook_euler python additional_scripts/preprint_generate_figures.py The simulation code, analysis scripts, HPC workflow, and setup instructions are provided in the associated GitHub repository. Large simulation outputs are archived here rather than tracked in Git.