By Josh Wasserman Advisors Prof Valerie Barr and
By Josh Wasserman Advisors: Prof. Valerie Barr and Prof. Hans-Friedrich Mueller ACCURATELY SIMULATING THE BATTLE OF THERMOPYLAE TO ANALYZE “WHAT-IF”
Battle of Thermopylae 480 BCE Second Persian War Stall the Persian army Artemisium Thermopylae Department of History, US Military Academy, West Point
History Greek army (~5, 200) v. Persian army (~80, 000 - 100, 000) Outcome: Greeks are defeated at the Battle of Thermopylae in 3 days Ultimately defeat the Persians as a result
Research Goal Develop an accurate simulation of the Battle of Thermopylae Accurate according to historical record Use this model to simulate “What-if” scenarios Compare the results of counter-factual simulations with the historical simulation Determine the importance of key factors of the battle
Developing the Simulation Programming in Java and Unity Focus on graphical representation Too many obstacles, inefficient Sim. Py “A process-based discrete-event simulation framework” (Sim. Py)
Simulation Design Hierarchical approach to daily events
Battle Events Day 1: Mountain defense Day 2: Fighting ends early Day 3: Ephialtes’s betrayal Greeks retreat Final stand
Combat System Attack and Defense values The “Roll” Generate a random number between 0 and the attack/defense value Highest value wins Phalanx Persian defense v. Greek attack Persian attack v. Greek defense
Historical Simulation: Greek and Persian Base Stats Greek Soldiers Attack Defense Spartan 9 9 Greek (other) 7 7 Persian Soldiers Attack Defense Immortal 4 2 Persian (other) 2 1 • Assigning values: • Skill • Quality of arms and armor
Historical Simulation: Greek and Persian Bonus Skills Greek Bonuses Attack Defense Phalanx +1 +2 Frenzy +2 -1 Broken Weapons -5 -5 Fake Retreat +2 N/A Persian Bonuses Attack Defense Strength in Number +2 N/A • Assigning values: • Strengths and weaknesses
Results of Historical Simulation Data in table gathered from 50 trials According to Scholars, ~ 20, 000 Persians were killed ~ 1, 500 Greeks were killed Greek Persian Day 1 Day 2 Day 3 Total Average 168. 48 42. 06 1345. 36 1555. 9 Standard Deviation 14. 32 8. 34 7. 38 14. 48 Average 9811. 72 3804. 34 7253. 4 20869. 46 Standard Deviation 125. 06 70. 88 18. 89 130. 32
What-if Ephialtes did not betray the Greeks? Battle did not occur during the Carnea Festival? The Greeks did not choose to make their stand at the “Middle Gate”?
What if Ephialtes never betrayed the Greeks? Large group of Greeks would not retreat No final stand Different strategy (randomized) after Day 2 Fatigue becomes a factor
What if Ephialtes never betrayed the Greeks? Data collected from 50 trials Battle lasted 17. 6 days on average Greek Persian Day 1 Day 2 Day 3 Total Average 170. 34 41. 68 178. 42 390. 44 Standard Deviation 13. 88 6. 72 17. 69 23. 15 Average 9823. 26 3820. 7 9443. 92 23087. 88 Standard Deviation 126. 76 75. 47 433. 03 446. 57
Historical Greek Persian Day 1 Day 2 Day 3 (Final Stand) Total Average 168. 48 42. 06 1345. 36 1555. 9 Standard Deviation 14. 32 8. 34 7. 38 14. 48 Average 9811. 72 3804. 34 7253. 4 20869. 46 Standard Deviation 125. 06 70. 88 18. 89 130. 32 Day 1 Day 2 Day 3 (No Final Stand) Total Average 170. 34 41. 68 178. 42 390. 44 Standard Deviation 13. 88 6. 72 17. 69 23. 15 Average 9823. 26 3820. 7 9443. 92 23087. 88 Standard Deviation 126. 76 75. 47 433. 03 446. 57 Counter-Factual Greek Persian
What if the battle was not fought during the Carnea Festival? Add 8, 000 – 10, 000 Spartans Remove other Greeks
What if the battle was not fought during the Carnea Festival? Data collected from 50 trials Battle always lasted 4 days, even with a final stand sequence All Greeks are killed by the 4 th day Greek Persian Day 1 Day 2 Day 3 Total Average 111. 32 27. 62 563. 22 702. 16 Standard Deviation 12. 13 4. 69 76. 19 79. 61 Average 9764. 68 3738. 78 42741. 36 56244. 82 Standard Deviation 115. 11 69. 17 450. 61 458. 46
Historical Greek Persian Day 1 Day 2 Day 3 (Final Stand) Total Average 168. 48 42. 06 1345. 36 1555. 9 Standard Deviation 14. 32 8. 34 7. 38 14. 48 Average 9811. 72 3804. 34 7253. 4 20869. 46 Standard Deviation 125. 06 70. 88 18. 89 130. 32 Day 1 Day 2 Day 3 (Final Stand) Total Average 111. 32 27. 62 563. 22 702. 16 Standard Deviation 12. 13 4. 69 76. 19 79. 61 Average 9764. 68 3738. 78 42741. 36 56244. 82 Standard Deviation 115. 11 69. 17 450. 61 458. 46 Counter-Factual Greek Persian
What if the battle was not fought at the “Middle Gate”? 50 m wide gap No betrayal Larger Phalanx Persians benefit from strength in number Fatigue becomes a factor
What if the battle was not fought at the “Middle Gate”? Data collected from 50 trials Battle lasted 14. 86 days on average Greek Persian Day 1 Day 2 Day 3 Total Average 313. 74 303. 4 172. 32 789. 46 Standard Deviation 36. 06 38. 72 41. 28 82. 21 Average 12469. 58 12436. 46 7979. 66 32885. 7 Standard Deviation 138. 99 284. 79 949. 88 1014. 21
Historical Greek Persian Day 1 Day 2 Day 3 (Final Stand) Total Average 168. 48 42. 06 1345. 36 1555. 9 Standard Deviation 14. 32 8. 34 7. 38 14. 48 Average 9811. 72 3804. 34 7253. 4 20869. 46 Standard Deviation 125. 06 70. 88 18. 89 130. 32 Day 1 Day 2 Day 3 (No Final Stand) Total Average 313. 74 303. 4 172. 32 789. 46 Standard Deviation 36. 06 38. 72 41. 28 82. 21 Average 12469. 58 12436. 46 7979. 66 32885. 7 Standard Deviation 138. 99 284. 79 949. 88 1014. 21 Counter-Factual Greek Persian
Conclusions Ephialtes decided the battle Professional army v. citizen army Position was essential
Future Work Persian retreat The effect of limited supplies Fixing “random” Persian strategy
Questions?
- Slides: 24