Good afternoon. My name is Fumihiro Sakahira and I'm an engineer at Koso K-Cup Engineering Incorporated.The title of this presentation is Generating Falsifiable Hypothesis Through Agent-Based Simulation of Population ReplacementDuring Agricultural Spread in Ancient Japan.This is an outline of today's presentation.I will discuss the strong point of agent-based modeling with regard to falsifiable through hypothesis generation in ancient historical studies.I will consider a method for generating working hypothesis that leads to the discovery and analysis of new remain by consider what kind of archaeological evidence would verify the hypothesis.As an example, I will consider the spread of agriculture and its effect on population increase in ancient Japan.The hypothesis is that the major player in agriculture were native people.This hypothesis can be verified by discovering bone remains of people with native traits together with artifact establishing the existence of agriculture.So it is the background to the present today.There is a need for more feedback between archaeologists and modelists.Due to that, we need comprehensive textbooks and handbooks for archaeologists on stimulation techniques.We also need ABM software that can be easily used by archaeologists.In addition, we think there is a need to bi-directionally link the result of ABM and field research.Due to that, we propose the following method.First, we generate a hypothesis based on which social phenomena predicted by the simulation agree with historical fact.We then consider what kind of archaeological artifact from what time period can verify the hypothesis.In particular, an advantage of ABM is that we can attach attributes such as archaeological, anthropological morphology, DNA, and culture to the agent.We then observed the composition ratio of each agent attribute as the result of the scenario simulation in chronological order.This allows us to propose a working hypothesis that can read the discovery and analysis of new remain by consider what kind of archaeological artifact would verify the hypothesis.That simulation can be input for field research.I would like to give a brief historical context to present study.About 2000 years ago, rice agriculture began for the first time in Japan.The IOI culture was established following the integration of native hunter-gather traditional with agriculture introduced by Chinese-Korean immigrants. Anthropological morphology study have showed that the human born from the IOI period differ from those from previous period.Therefore, Chinese-Korean immigrants thought to have had a large genetic influence on the IOI people.So, the presence of immigrants are important in the formation of IOI farming culture.However, a large question remains concerning Japan Anthropology and Archaeology.At the end of the German period, which was just before the IOI period, the population of native people was larger than that of immigrants.However, 300 years later, the situation was reversed.So, the most controversial question is who plays the major role in establishing a farming culture during the IOI period.Native people or immigrants.In an attempt to answer this question, we use an ABM approach.This slide shows an outline of the ABM model.This model considers the fusion of genetic traits under an increasing population and the diffusion of agriculture and mitochondrial DNA.Agent were created in the two-dimensional outer-locked space to represent individuals in the IOI period.This table shows the variable parameters for agents, such as sex, life expectancy, food production methods, and so on.Trades in were inherited from both parents when a new agent was created.Mitochondrial DNA was inherited only from the mother.The simulation model includes three rules.In the agriculture diffusion rule, agriculture knowledge was obtained either from a neighbor or inherited from a parent.In the marriage rule, a male was married to a female randomly selected from neighboring females.In the movement rule, each agent moves in a random direction.Here you can see the parameter for our simulation model for all combinations of parameters.The total number of simulations was 441.Each case was run 10 times.This slide shows the framework.The simulation input was fact concerning late German period with regard to genetic traits native were majority and immigrant were minority.The frequency of type in mitochondrial DNA was low.The simulation constraints used were fact regarding the mid IOI period.At this time, native traits were minority and immigrant traits were majority.In mitochondrial DNA, frequency of type N was high.Diffusion rate for agriculture allows us to investigate who played the major role in IOI agriculture spread in northern Kyushu.We then generated a hypothesis and evaluated the evidence supporting it.So, I will now show you some simulation result.The left graph shows the ratio of people with the immigrant trait 300 years later.The simulation result are grouped into three categories based on a slow, medium and fast diffusion rate for agriculture.Each dot represents one line.Over 80% of these data points are consistent with the archaeological records.It can be seen that the agricultural diffusion speeds increase so does the percentage of people with immigrant trait.The right graph shows the time evolution of the composition ratio of farmers for the case of slow diffusion rate.This is easy to see that immigrant or mixed people were the primary farmers throughout all stages of the IOI period.In other words, immigrants played the major role in the spread of agriculture.However, the right graph in this slide shows the time evolution of the composition rate of farmers in the case of the first diffusion rate.In this case, you can see that the simulation predicts the native people who are in fact the primary farmers in the early stages of the IOI period and so played the major role in the spread of agriculture.As a verifiable material about two foreseeable hypothesis, our simulation presents the bone remains of people with native traits.Along with the artifact, it verifies the existence and agriculture in the early stages of the IOI period.This slide is the same, so omitted.In conclusion, by assigning attributes such as morphology, DNA, food production method, and culture to multiple agents and simulating the component ratio for each attribute as a function of time,we were able to present materials based on the observed pattern of combination that could verify the hypothesis.We show working hypothesis based on ABM results that may lead to the discovery new remain.By reanalyzing the remain based on the simulation result, we showed that by directional simulation and field research is possible.Future work, we need identify which rules and parameter are important for the classification of different scenarios and hypothesis.We need to identify which microscopic phenomena in any given simulation step can produce different result even under the same condition.For the factor based on approximation, decision tree for simulation log and random forest algorithm can be used to separate rule and parameter combination.Finally, using the method described in this presentation, we previously determined the effect of the spread of agriculture on people change in the medieval Okinawa is run in Japan.Thank you very much for your attention.Any comments and questions are welcome.But if possible, speak with very slowly and very simply because I'm not good at English. Sorry.