dognovel Chapter 47 Training Groups, this question isn't complicated.
Chapter 47 Training Groups, this question isn't complicated.
A week after that interview, the discussion about Lufeng on campus only intensified.
The CCTV special report has not yet aired, but all sorts of rumors are already circulating.
Lu Feng ignored all these rumors and continued his routine of going between home, work, and home.
Just after the last class on Monday morning ended, Ye Guodong sent me a text message.
[Meet in Computer Center Room 301 to get to know each other first.]
After receiving the message, Lu Feng walked alone through the tree-lined path toward the building filled with the hum of servers.
He pushed open the heavy soundproof door of the 301 computer room, and a unique smell, a mixture of air conditioning coolness and electronic equipment heat, rushed out.
In the huge server room, rows of computers were neatly arranged, with green code scrolling on the screens.
In the innermost teaching area, three teachers and five students were already sitting around a long table.
Upon seeing Lu Feng enter, all conversations ceased abruptly, and eight pairs of eyes turned towards him.
Ye Guodong was the first to stand up, a warm smile on his face.
"Come, come, Lu Feng, come and sit down."
He pulled out the chair next to him, acting as if Lu Feng wasn't a student participating in the selection process, but a leader coming to inspect the area.
Lu Feng walked over and nodded to the teachers.
"Teacher Ye, Teacher Xu."
His gaze fell on the third teacher, a middle-aged man with frameless glasses, thinning hair, but a vigorous spirit and a strong programmer's aura.
A typical Mediterranean.
"Let me introduce you," Ye Guodong said, pointing to the teacher. "This is Professor Sun Yi from the School of Computer Science, our programming advisor for this modeling competition."
He then pointed to the five students at the table.
"These three are third-year students from the School of Computer Science. This is Li Hao, this is Wang Zhe, and this is Liu Yu."
The three boys whose names were called nodded at Lu Feng, and each of their reactions was quite interesting.
Li Hao appeared very composed, and simply gave a polite smile.
Wang Zhe, on the other hand, looked Lu Feng up and down with obvious curiosity.
Liu Yu reacted the most directly. He pushed up his glasses, stood up while holding onto the table, and extended his hand: "Junior Brother Lu Feng, I've long admired your name."
Their team had a lot of respect for this person who could handcraft exoskeletons.
This display of force made Lu Feng a little embarrassed.
"These two are third-year math majors, Chen Jing and Zhou Xiaoxiao."
The two girls also stood up, their eyes filled with both curiosity and anticipation as they looked at Lu Feng.
"Alright, now that everyone knows each other, you're all a team from now on. Keep in touch and communicate in private." Ye Guodong clapped his hands, drawing everyone's attention.
"I won't waste any more time. Let me explain the competition process to you first."
He turned on the projector, and a PowerPoint presentation appeared on the screen.
"The competition will last for three days. From the moment the questions are distributed, you will have 72 hours to complete all the work, from modeling and solving to verification and paper writing."
"The general procedure is that the competition questions are distributed at 10:00 a.m. on the first day, and you must come up with a complete mathematical model and solution approach on the same day."
"The next day, the programming group students had to implement the model in code and run it to get results. The modeling group students, on the other hand, had to start writing the theoretical part of their papers."
"On the third day, we tested and optimized the model, and at the same time improved all the contents of the paper, including figures, abstracts, and references, and finally submitted it before the deadline."
Ye Guodong spoke very quickly, clearly having said these words many times before.
"Regarding the group assignments, Ms. Xu, Ms. Sun, and I have discussed them, and this is the preliminary arrangement."
He pointed at Lu Feng and the two female students from the mathematics department.
"The three of you will be in charge of the core modeling team, with Lu Feng as the team leader."
He then pointed to the three male students from the computer science department.
"The three of you will be responsible for solving the problem using programming."
"As for the final paper writing, Chen Jing will be the main writer for now, with Zhou Xiaoxiao assisting. Of course, the students in the modeling group will also participate throughout the process."
The arrangement is clear.
The three male students from the Computer Science Department exchanged glances but said nothing.
Although it sounds a bit absurd to have a freshman as the core modeler, since it was a joint decision made by the three professors, they had no choice but to accept it.
"The training camp starts today. You will gather in this computer lab every evening during self-study time."
"The three of us will issue the leave slips to you all at once to ensure you are not disturbed."
After Ye Guodong finished speaking, he turned off the PowerPoint presentation.
"Alright, we've covered the rules. Now, let's get practical."
He turned around, picked up a whiteboard marker, and wrote a question on the whiteboard behind him.
"This is Problem A from the national competition two years ago. The problem isn't long, let me read it to you."
"A certain city has a complex transportation network that includes thousands of intersections and tens of thousands of roads, and currently has 500 taxis operating in this network."
"Please develop a mathematical model to plan the optimal empty cruising route for each taxi, so as to minimize the average waiting time for passengers throughout the city, while ensuring that the taxi driver's income per unit time is maximized."
The moment the question was posed, the computer room fell silent.
The five third-year students all looked rather serious.
This is a typical multi-objective optimization problem, and it is dynamic with extremely complex constraints.
Passengers arrive randomly, road conditions change in real time, and the driver's earnings and the passenger's waiting time are two mutually constraining objectives.
Building a model that perfectly describes this system is extremely difficult.
"This problem...we need to use Monte Carlo simulation, right? First, we need to model the probability of passengers appearing." Chen Jing from the mathematics department spoke first, her brows furrowed.
"No way," Li Hao from the School of Computer Science immediately retorted. "The Monte Carlo simulation involves too much computation. There are tens of thousands of roads and 500 vehicles in the city. The complexity of the real-time simulation is astronomical. It's impossible to finish it in three days."
"How about using the ant colony algorithm? Treat each car as an ant, the passengers' needs as a food source, and find the optimal path through pheromone iterations," suggested Wang Zhe, a male student from another programming group.
"Ant colony algorithm is prone to getting stuck in local optima and its convergence speed is too slow."
The group discussed the matter, came up with several solutions, but quickly rejected them themselves.
The three professors stood quietly to the side, listening without interrupting.
Their gaze remained fixed on the person who hadn't uttered a single word from beginning to end.
Lufeng.
He leaned back in his chair, his hands clasped in front of him, listening quietly to the discussion.
He only spoke slowly and deliberately after everyone had finished speaking and fallen silent.
"Your thinking is too complicated."
With just one sentence, he drew everyone's attention to himself.
"The essence of this problem is not a pathfinding problem, but a resource allocation problem."
Lu Feng stood up, walked to the whiteboard, and took the pen from Ye Guodong.
"Taxi vehicles are a resource, and passenger needs are a task. Our goal is to achieve the best match between resources and tasks in a dynamically changing environment."
Instead of writing out complicated formulas, he drew a few circles on the whiteboard.
"The first step is to reduce the dimensionality."
He drew a grid on the whiteboard with a pen.
"We don't need to consider every single intersection; we can divide the entire city into several grid areas of the same size. This simplifies the problem from intersection-to-intersection to 'area-to-area'."
"The second step is prediction."
"By using historical data, we can build a time series prediction model, such as the ARIMA model, for the ride-hailing demand in each region at different times. This way, we can predict which region might experience a ride-hailing peak half an hour in advance."
"The third step is scheduling."
"This is the most crucial point." Lu Feng drew arrows between several grids with the tip of his pen.
"We can view the entire system as a Markov decision process, where the state of each taxi represents its location within a given area."
"Its action is to choose which adjacent area to go to."
"The reward function is the probability of it picking up a passenger in that area, and the expected revenue for that order."
"What we ultimately need to do is find the optimal strategy for this Markov decision process."
"This can be solved using the Q-learning algorithm in reinforcement learning. There's no need for global path planning; each vehicle only needs to make the optimal next decision based on its current state and the Q-table."
He finished speaking in one breath, and the entire computer room fell silent.