羽藤先生
●	overlapping in network assignment model is important
○	hamiltonian,  yacobian e.t.c
大山先生
●	unrealistic passes are usually removed, related to choice set generation 
●	for pedestrian routes, it might be useful
●	case studyではunrealistic route を含んでる？
○	yes
●	any potential numerical issues in this model?
○	 there are some limits to include many conditions of this problems, but they have some technique
渡邉先生
●	interpretation of beta? can we do that? 
○	yes
○	estimation beta in different city is interesting, e.g. in the big city ppl are more sensitive to beta 

Discussion

Group1 B4
●	歩行者の経路選択に使える。距離ベースだと極端な例は消えてしまう。
歩行者だと混雑や泥道などを考慮できるから非現実的なルートも選択肢として生成できるかも
●	large scale and simple utility function in this model, so different?

Group2 
●	How can we show the uniqueness? strong monotonicity 


Group3 M2
●	fixed-point problem is interesting, simultaneously estimation and observation pass based is our future work

Group4 Teachers
●	for osadasan thank you for the good material to learn what is behavior model

Group5 International
●	objective of route choice model, utility function is simple 
●	AI base route choice model 

井料先生
●	経路選択の目的とは？モデルとして何を目指してるかが重要だが経路選択だとそれが錯綜しやすいから、自分の研究でも気をつけた方がいい（Group2）

SU 
●	AI or machine learning in traditional route choice model
●	integrate some AI driven method in the model, I use them separately before