MetaModel Analysis for Agent-Based Models: Scales, Uncertainty, Data Analysis

Abstract

The public reportmg burd en for th1s collection of mform ation is estimated to average 1 hour per response. includmg the tim e for rev>ewing instruct>ons. searclhing exi sting data so urces. gathering and mmnta1mng the data needed . and co mpleting and rev1ew 1 ng the collection of 1nforma t1 on. Send co mments regarding th1s burden est1m ate or any other aspect o fth1s collection of mformat1on . 1nclud1 ng suggesti ons for reduc1 ng the burden. to the Department of De fense. ExecutiV e Serv1ce D>rectorate (0704 -0 188) Respo ndents should be aware that notwithstanding any other prov1 s1 on of law. no person shall be sub;ect to any penalty for falling to comply w1th a collect> on of 1 nform at1on if 1 t does not display a currently v alid OMB co ntrol number PLEASE DO NOT RETURN YOUR FORM TO THE ABOVE ORGANIZATION. REPORT DATE (DD-MM-YYYY)12. REPORT TYPE 3. DATES COVERED (From-To) SPONSOR/MONITOR'S REPORTArlington, VA 22203 NUMBER(S)AFRL-OSR-VA-TR-20 12-0257 DISTRIBUTION/AVAILABILITY STATEMENTDistribution A -Approved for Public Release SUPPLEMENTARY NOTES ABSTRACTAn agent-based model of civil violenc e/criminal activity was developed. The agent based model was found to be in good agreement with FBI data on crime and violence in 5660 U.S. cities. In particular, the proportion oflaw enforcement officers required to maintain a steady low level of criminal activity w as found to increase with the size of the population. Reducing the number oflaw enforcement officers below a critical level can rapidly increase the incidence of violent/criminal activity. Application of global sensitivity analysis to the agent-based model identified the law enforcement officer vision as the most significant parameter. A method to speed-up the model evaluation was developed by employing stochastic differential equations. Jailed citizens and active citizens served as two coarse variables in the reduced model. The effect of preferential gathering sites attracting active citizens was studied. It was found that increased density of citizens around the gathering cites leads to increased violence/criminal activity in the city. The effect of citizen's remote vision via a small world network was also studied. It w as found that local vision is more important in creating outbursts of activity than the possibility of being connected to long-distance neighbors. 1S. SUBJECT TERMS AbstractAn agent-based model of civil violence / criminal activity was proposed and validated. The analysis of data on crime and violence for 5,660 U.S. cities over the period of 2005-2009 was performed and the following trends were uncovered. The proportion of law enforcement officers required to maintain a steady low level of criminal activity increases with the size of the population of the city. The number of criminal/violent events per 1,000 inhabitants of a city shows non-monotonic behavior with size of the population. The strong dependence on size of the population was found, which leads to partially irrational behavior on the part of citizens. The nature of v...