Agent-based model of recycling
Translation of agent-based model from Coco et al 2020 from R to Java. Replication analysis can be found in the analysis folder.
Agent-based model of recycling in extreme case of no geological processes with additional parameters and outputs
args[0] of (String) name of folder for output
args[1] name (String) name of model run
args[2] size (int) number of rows and number of columns
args[3] startYear (int) initial data for all layers
args[4] timestep (int) number of years each timestep represents
args[5] maxUI (int) maximum number of actions agents can take when manufacturing
args[6] maxAC (int) maximum number of artifacts agents can carry when moving
args[7] maxFS (int) maximum flake size that can be removed from nodules
args[8] maxNS (int) maximum nodule surface area
args[9] bProb (double) probability of agents removing new flakes from nodules
args[10] sProb (double) probability of scavenging previously discarded material
args[11] overlap (double) parameter determines order of agent appearance on landscape
args[12] mu (double) parameter for Levy walk function
args[13] sizePref (boolean) selection is based on size of objects
args[14] flakePref (boolean) preference of flakes over nodules
args[15] minFS (int) selection for flakes of at least this size
args[16] strict (boolean) possibility of scavenging items that do not match selection parameters
args[17] ED (double) probability of erosion unused in current model code
args[18] GF (int) frequency of geological events unused in current model code
args[19] totalSteps (int) total number of timesteps for each model run
args[20] totalAgents (int) total number of agents that will be used for each model run
- Agents look for items to collect from their current location. If there are objects available, they collect based on scavenging probability and selection criteria
- If agents have items in hand, they either create new flakes based on a probability of making blanks or retouch flakes
- If agents does not have items, they "find"" new nodules
- If agents have exhausted items, they discard those at the current location
- If agents are carrying too much, they discard unwanted items that do not fit the selection criteria at the current location
- Agents move to a new location
Must create an output folder in the base recycling-java directory for model outputs to be created properly!
-
Model can be run for testing purposes from a Java IDE using the TestExtendedModel.java script
-
Model can be run in a command line environment using the call-sbatch.sh script
- In this case, filepath names in call-sbatch.sh and run-param.sbatch need to be updated
[model name]_artifacts-data.csv
information for each artifact on the landscape at the end of model run
[model name]_layers-data.csv
information for each layer on the landscape at the end of model run
[model name]_model-data.csv
information for the whole model at each timestep of a model run
-
All output should be left in respective folders for R scripts in analysis folder to run properly
-
A CSV with all of the model-level results needs to be created with the following command:
find -name model* -exec cat {} \; > joined_model_data.csv -
Each line of all .sbatch scripts in the experiments/model-analysis, experiments/layer-analysis, and experiments/artifact-analysis folders should be run
- Any CSV outputs created should be moved to a results folder
-
All R scripts and .sbatch scripts in the experiments/results-analysis folder should be run
- This requires creating a figures and figures/supplementary-figures folder
In most cases, filepaths in the analysis scripts were hardcoded to avoid any issues with producing results and figures. These would need to be refactored to run properly.
Figure 1 - made separately
Figure 2 - mu-viz.tiff
source(../analysis/ExtendedModel-analysis/experiments/model-analysis/mu-visualization.R)
Figure 3 - odds-ratios-YOFD.tiff
source(../analysis/ExtendedModel-analysis/experiments/results-analysis/artifact-exposure-results.R)
Figure 4 - SKEW_age-of-discard.tiff
source(../analysis/ExtendedModel-analysis/experiments/results-analysis/exposure-skew-analysis.R)
Figure 5 - average-recycling-intensity_by-probs.tiff
source(../analysis/ExtendedModel-analysis/experiments/results-analysis/recycling-intensity-landscape-averages.R)
Figure 6 - recycling-trends.tiff
source(../analysis/ExtendedModel-analysis/experiments/results-analysis/recycling-trends-graphs.R)
Figure 7 - recycling-trends_by-probs-mu.tiff
source(../analysis/ExtendedModel-analysis/experiments/model-analysis/trends-by-blank-scavenging.R)
Figure 8 - behavior-trends-by-mu.tiff
source(../analysis/ExtendedModel-analysis/experiments/results-analysis/behavioral-trends-graphs.R)
Figure 9 - behavior-trends-by-selection.tiff
source(../analysis/ExtendedModel-analysis/experiments/results-analysis/behavioral-trends-graphs.R)
Figure 10 - recycling-intensity-variation_by-probs.tiff
source(../analysis/ExtendedModel-analysis/experiments/results-analysis/recycling-intensity-variation.R)
Figure 11 - average-correlations.tiff
source(../analysis/ExtendedModel-analysis/experiments/results-analysis/spat-layers-correlations_by-parameters.R)
Figure 12 - RI-correlations_by-movement.tiff
source(../analysis/ExtendedModel-analysis/experiments/results-analysis/spat-layers-correlations_by-parameters.R)
Figure 13 - ri_assemblage-density_probs.tiff
source(../analysis/ExtendedModel-analysis/experiments/results-analysis/all-object-count-results.R)
Figure 14 - ri-cr_plot.tiff
source(../analysis/ExtendedModel-analysis/experiments/results-analysis/cortex-ratio-results.R)
Figure 15 - overlapping-grid-squares-dist_V2.tiff
source(../analysis/ExtendedModel-analysis/experiments/results-analysis/hotspot-overlap-analysis.R)
Figure 16 - log-odds_no-overlap.tiff
source(../analysis/ExtendedModel-analysis/experiments/results-analysis/hotspot-overlap-analysis.R)
Figure 17 - IRR_overlap-counts.tiff
source(../analysis/ExtendedModel-analysis/experiments/results-analysis/hotspot-overlap-analysis.R)
Figure 18 - retouched-encounter_overlaps.tiff
source(../analysis/ExtendedModel-analysis/experiments/results-analysis/hotspot-overlap-analysis.R)