nucmednet Chapter 4: Seduction Success
Chapter 4: Seduction Success
2005年10月11日，周一，清晨7:23。
Jiang Yu opened his eyes in room 312 of the Gateway apartment building at USC.
After a three-second daze, consciousness returned like a receding tide.
2005, Los Angeles, USC, 21 years old, exchange student, $100,000, rebirth.
He sat up and looked at the bed opposite him.
Mike was still fast asleep, snoring evenly.
Jiang Yu quietly got out of bed and took a black notebook out of his backpack.
It wasn't one of those issued by the school; it was a hardcover notebook that he had prepared back in the country.
He turned to the first page and wrote in Chinese:
Target list:
1. Complete registration for all courses this semester (Priority A)
2. Open a US stock account and research leveraged trading (priority A+).
3. Establish contact and trust with Zhou Mu (Priority S)
4. Find Chen Wei (Priority A)
After finishing writing, he closed the notebook and stuffed it into the inner layer of his backpack.
In the bathroom, he looked at his younger self in the mirror, washed his face with cold water, and brushed his teeth.
A photo of Mike and his family is pasted on the edge of the mirror. It's a typical middle-class family: the parents are all smiles, there are two younger sisters, and a golden retriever.
He had never had a roommate in his previous life.
After starting my own business, I lived alone, and after turning forty, I became more accustomed to the solitude of hotels and apartments.
Listening to Mike's snoring outside the door at this moment, I felt a strange sense of vividness.
"Hey, Jiang!" Mike woke up, rubbing his eyes as he sat up. "Good morning. Are you going to the international student office to register today?"
"Nine o'clock." Jiang Yu wiped his face with a towel.
"I'm going to practice at eight, shall we go downstairs together?" Mike was already changing into his training clothes. "By the way, are you free tonight? There's a fraternity party, lots of girls from the film academy will be there. I can take you in."
"I might have something to do tonight," Jiang Yu thought for a moment, "but thank you."
"Anytime." Mike grinned, revealing his white teeth. "By the way, you really can cook, right? I wasn't joking yesterday. The cafeteria food is awful. If you could cook two Chinese meals a week, I'd do the shopping, how about it?"
"Deal." Jiang Yu nodded.
At 7:50, the two went downstairs together.
Mike slung his sports bag over his shoulder and headed towards the training field, while Jiang Yu walked toward the international student office.
Students passing by spoke in various languages, bicycle bells rang crisply, and in the distance, students were already sitting on the steps in front of the library, reading aloud.
Jiang Yu slowed his pace.
In his past life, he returned to USC countless times as an "industry leader" to give speeches, recruit, and participate in forums.
He was already in his thirties at the time, and looking at these young faces, he always felt a sense of alienation, as if he were in another world.
Now, he is one of them.
It doesn't feel bad...
.......
9:15 a.m., International Student Office.
More than thirty exchange students from all over the world were crammed into the conference room, where a blonde staff member named Lisa was explaining the rules for international students:
"...The F-1 visa requires full-time study, no more than 20 hours of on-campus work per week, and prohibits unauthorized off-campus work. Any academic misconduct or illegal activity may result in visa cancellation..."
Jiang Yu sat in the back row, listening while checking Apple's real-time stock price on his laptop—AAPL: $40.31, a slight increase of 0.03.
He opened ETRADE's website. In 2005, this online brokerage already supported opening international accounts. In his previous life, he had opened his first US stock account here.
Please fill in your personal information: Jiang Yu, Chinese national, F-1 visa, USC student, US address, China permanent address (Daye, Hubei), annual income (please fill in: $10/year financial support from parents).
The most crucial step: investment experience.
He checked "Experienced (3+ years)" and noted "Has participated in stock and futures trading in China".
This isn't entirely a lie; it did exist in a past life, but not in this one.
Then comes the risk tolerance questionnaire.
He selected the highest risk tolerance for all options: "Can accept a loss of more than 50% of principal", "Seeking high growth, not pursuing stable returns", and "Investment period of 3-5 years".
Finally, there's the leverage application.
In 2005, US stock margin accounts were relatively strict for international students, but not impossible.
He applied for 5x leverage, meaning that with a principal of $10, he could purchase up to $50 worth of stocks.
This means that if the stock price drops by 20%, he will be liquidated and lose all his principal.
High risk, high reward.
Submit an application.
The page indicates: "The review process typically takes 3-5 business days. During this time, you can transact using your cash account."
Jiang Yu closed the page.
"...Next is the course registration guide." Lisa switched to the next slide. "Exchange students must complete all course registrations by this Wednesday. This is the course list for the School of Film Arts..."
Jiang Yu looked at the screen.
"Fundamentals of Digital Media Technology," "History of Film Visual Effects," "Principles of Computer Animation," "Real-time Rendering Algorithms"... these are all subjects he had mastered in his previous life.
He still carefully wrote down the course number and the professor's name.
.......
1:30 PM, third floor of the Computer Building, School of Engineering, laboratory.
Jiang Yu found the lab for "Real-time Rendering Algorithms" according to the course schedule.
Pushing open the door, I found only six or seven people in the room, all of them boys, and the air was filled with the smell of code and instant noodles.
Zhou Mu sat in the corner, staring at two monitors in front of him, his eyes filled with mathematical formulas and code.
"Hi." Jiang Yu walked over and sat down next to him.
Zhou Mu looked up and adjusted his glasses: "Jiang Yu? You chose this course too?"
"Hmm." Jiang Yu glanced at his screen. "Optimizing the memory structure of photon mapping?"
"Yes." Zhou Mu's eyes lit up. "I've found that traditional photon image storage methods are too wasteful; 80% of the photons ultimately don't make a significant contribution to the final image. I want to use importance sampling and adaptive storage..."
He started talking non-stop again.
Jiang Yu listened patiently for five minutes, then spoke up: "Why not try using machine learning to predict the importance of photons?"
Zhou Mu was stunned.
"In 2005, the application of machine learning in computer graphics was almost non-existent."
Jiang Yu continued, "There are already papers exploring the use of neural networks for noise reduction and super-resolution. If we train a network, taking the geometry, material, and light source information of the scene as input, and outputting the contribution weight of each photon to the final image... by removing low-weight photons in advance, storage overhead can be reduced by more than 70%."
"This..." Zhou Mu stared at him as if he were an alien. "This requires a huge amount of training data and computing resources. And how can we guarantee real-time performance? Neural network inference also takes time..."
"Pre-trained + lightweight model," Jiang Yu said quickly. "We train the model using a large number of offline rendered scenes to obtain a general-purpose model. During deployment, we use a simplified network structure, coupled with GPU acceleration. The GeForce 7800 GTX from 2005 already had programmable shader units, which could perform simple matrix operations."
He paused, then added, "I know this sounds ahead of its time. If we don't do it, someone else will in a year or two. Then we'll just have to catch up."
Zhou Mu remained silent for a long time.
Finally, he asked, "Do you have a concrete plan?"
"Yes." Jiang Yu took out a USB drive from his backpack. "I wrote an outline last night, including data acquisition methods, network structure design, training and deployment procedures. But I need someone with math skills to refine it."
Zhou Mu took the USB drive and plugged it into the computer.
A 28-page PDF document popped up on the screen, complete with pictures and text, from theoretical derivation to pseudocode, and even preliminary benchmark test predictions.
"You wrote this all night?" Zhou Mu's voice was somewhat surprised.
"I can't sleep," Jiang Yu said casually.
In fact, this is a technology that only matured in 2018 during the previous era of Light and Shadow, based on deep learning for predicting the importance of photons.
He reconstructed the core framework from memory; although details needed to be filled in, the direction and path were clear.
Zhou Mu turned the pages one by one, his breathing gradually becoming rapid.
"The design of this loss function... is ingenious." He pointed to one of the pages, "using the gradient of the rendering error as a supervision signal, instead of the direct photon contribution value..."
"Because the direct contribution value is unstable and greatly affected by noise," Jiang Yu added, "using gradients is more robust."
"Yes! Yes!" Zhou Mu suddenly turned to look at him, his eyes gleaming behind his glasses. "Jiang Yu, what exactly is your background? Which lab in China did you come from?"
"The School of Animation and Digital Arts at the Communication University of China," Jiang Yu said with a smile. "Several professors at our school have collaborations with the MIT Media Lab. I've had access to some cutting-edge materials."
A half-true, half-false explanation.
Zhou Mu believed it.
"I want to do this project," he said solemnly. "Even if it's just theoretical verification, it has the potential to be published in a top-tier paper. At the SIGGRAPH level."
"It's not just about academic papers." Jiang Yu lowered his voice. "If we can really make a prototype, or even a toolchain, we can sell it to Industrial Light & Magic, Weta Digital, or start our own business in the future."
"Start a business?" Zhou Mu was stunned.
"Why not?" Jiang Yu looked at him. "Hundreds of tech startups are born in the US every year. If we have core algorithms and patents, we have a chance."
Zhou Mu fell silent.
He looked down at the document on the screen, tapping his fingers on the desktop.
Jiang Yu wasn't in a hurry; he knew Zhou Mu needed time to process this.
This brother who accompanied him in starting his business for ten years in his previous life is essentially a tech purist.
What he loves is the problem-solving itself, not commercial success.
You need to attract him with "technological possibilities," not "profit potential."
"I need to think about it," Zhou Mu finally said, "but I'll join this project. Even if it's not for starting a business, it's to create something different."
"Okay." Jiang Yu extended his hand. "It's a pleasure doing business with you."