Top academic

Chapter 281 Solution
At the same time, in the building of the Asian Situation Awareness Center, Qiao Yu was mobilizing the bigwigs of his project team.

Yes, now all the professors in Qiao Yu's project team are already Chinese academic leaders. And they are also the kind with a very high level of confidentiality.

Now that the Sky Shield exercise has been a success, it is foreseeable that the status of these professors and researchers in the international academic community will rapidly increase.

Especially the professors and researchers who have published some theoretical articles in the past four years.

As Qiao Yu said before, if there are not a few Nobel Prize winners among these people, then this award would have no need to exist.

As we all know, the most important criterion for the Nobel Prize is how much visible contribution one has made to the world.

Now the planetary defense system is there, which is an outstanding contribution to all mankind. From now on, we no longer have to worry about a civilization being extinct because of a meteorite hitting the earth as happened a few years ago.

If they still don't give the Nobel Prize, then the Nobel Prize Committee of the Royal Swedish Academy of Sciences had better not take planes in the future, because they have such evil hearts that they may be struck by lightning.

"Everyone, I know you have been working hard recently. The Sky Shield exercise has taken up a lot of your energy. Even on the day of the exercise, the head of the team had to come to the scene to provide guidance.

I also specifically suggested to my superiors that the exercise has been successful, the first phase of the Space-Based Denial Project has come to a successful conclusion, and it coincides with the National Day holiday, so we should give everyone a holiday. Unfortunately, it was not approved. "

Qiao Yu spread his hands, looking regretful. Anyway, he said this with a clear conscience.

At first, he did suggest to Chief Engineer Yu that everyone should be given a holiday, but Yu Yanjiang flatly refused.

The reason given was that the second phase of the plan was already being prepared. How could there be time to rest?

After all, it is not safe to just defend the entire Asia. We have to wait until the second phase of the project is completed and the entire northern hemisphere is within the defense range of space-based denial before we can breathe a sigh of relief.

As the old man said, only by doing things in one go can we get things done well...

Of course, Qiao Yu would definitely not say at this time that the holiday he suggested to Yu Yanjiang was for the engineers involved in the space-based denial project, and Yu Yanjiang was not considered the superior of their project team.

It is indeed up to Lao Yu to decide whether the engineers of the space-based project can take a holiday, but it is also up to him to decide whether the project team can take a holiday.

Fortunately, although the professors in the project were not so easy to fool and would not believe Qiao Yu's lies, they still knew how "ruthless" Qiao Yu was.

Although everyone knew that Qiao Yu was talking nonsense, no one looked surprised, let alone expressed dissatisfaction.

"So don't take a break. Although we don't need to solve those problems in the space-based project for the time being, I believe everyone knows that there are still many things to be done on the quantum simulation computer side.

We have been working together for such a long time, and I believe in your faith and passion for work. So I won't say more. Let's use this conference room to discuss the key issues that need to be resolved next.

After you go back, hurry up and finish your own business. I promise that after our Chinese quantum simulation computer is officially released, I will give everyone at least a week's vacation!"

Warm applause broke out instantly.

This is a seven-day holiday! Compared to the past four years when they only had three days off during the Spring Festival, Qiao Yu was really generous this time, which was very touching.

This really treats them as human beings!
……

In fact, Qiao Yu didn't want to be like a capitalist, forcing everyone to work every day. It was mainly because of the situation on the quantum simulation computer that made him ashamed.

In fact, the engineering prototype of the new generation of quantum simulation computers has reached the fifth generation.

The fifth generation of engineering prototypes has almost perfectly solved the problems of the previous two generations of engineering prototypes, such as the generation and maintenance of quantum states being affected by thermal noise, electromagnetic interference and other factors during simulation, which led to a decrease in simulation accuracy and problems such as too short decoherence time.

It can be roughly understood that a classic electronic computer will blue screen once every few hours of use due to various problems.

However, the specific manifestation of a quantum simulation computer is not a blue screen, but various strange problems.

For example, the system becomes extremely clunky, and all the ways to interact with the system become less accurate. I wanted to open this app, but ended up opening another app instead...

There are also problems that need to be solved at the hardware level, such as timing mismatch caused by data transmission delays between quantum simulation units, modal memory, and classical computing modules, which have all been solved.

Therefore, the first generation of quantum simulation computers named Tai Chi series are actually ready for mass production. Even many components that have been confirmed to be problem-free have begun to be stocked in small quantities, such as graphics cards and CPUs.

The only issues left for the motherboard are the optical interconnect bus protocol and the probabilistic clock synchronizer. The main problem now is actually at the software level.

The principles of quantum simulation computers are closely related to the generalized modal axiom system.

Under the generalized modal axiom system, the dimensional expansion of the modal number N(α, β) will exponentially increase the consumption of hardware resources.

This requires optimizing the compression algorithm currently used based on tensor networks, combined with a dedicated memory architecture. The current problem is that this compression algorithm is always flawed.

The main problem at present is what they call "entanglement correlation truncation in high-dimensional modes."

This is why Qiao Yu had been staring at Zheng Zefeng. It was actually nothing when only running classical computing tasks.

However, as long as quantum computing tasks are involved, problems such as phase distortion, correlation breakdown, and error exponential accumulation will arise.

This was obviously unacceptable to Qiao Yu.

There must be something wrong with his theory, otherwise a quantum simulation computer would not be possible.

……

On the way back, Qiao Yu called three professors, Zheng Zefeng, Zhang Xiaoli, Luo Qixun and Hu Xingyao, into a car.

Three years ago, after the core members of the project team achieved some results, they were transferred from their respective units to Yanbei University. They were also known as high-level talents.

Afterwards, the Yanbei International Center for Mathematical Research specially prepared two floors for the project team as a place for daily work and office.

Since the two projects got on track, in addition to internal meetings, people from the laboratories below often come to attend meetings.

Most of them are technical meetings. Although each meeting is not long, they are very frequent.

The main reason is that there are too many research units that need to be connected to the two projects. They are from universities, research institutions, and enterprises. On average, each person in the project team needs to connect with more than ten or twenty units.

You can imagine how busy we are on a daily basis, and we all have to travel frequently. When things that are theoretically fine are actually implemented in the lab, all kinds of weird problems always arise. Some can be solved remotely, but those that cannot be solved will be brought together for meetings, or the relevant person in charge will go to the front line to handle them personally.

So Qiao Yu was actually right. If we really talk about how busy he has been in recent years, Qiao Yu is not as busy as these professors. But relatively speaking, his work is definitely more important.

After all, all problems are ultimately reported to him at weekly team meetings.

Although he doesn't need to study the details, he has to give a general solution direction.

It's a similar situation today.

Qiao Yu called together a professor who worked on mathematical theory, a professor who worked on quantum theory, and the professor in charge of algorithm engineering, which was basically equivalent to a small meeting.

"I just talked about the situation at the meeting. It's not that I'm in a hurry to do something. The key is that the Quantum Simulation Alliance must be in a hurry. In the past few years, more than 40 companies have invested more than 300 billion yuan.

Now our space-based denial project has passed the first phase of the project acceptance. It is really unacceptable that quantum simulation computers are not yet on the market.

So my idea is that we should seize the time to tackle technical problems in the next few months and try to solve the algorithm problems. The National Day gift has been successful, and we will try to give you another gift before New Year's Day.

So let's have a good chat today and try to sort out the current problems so that you can get on with your work. Professor Wang, please tell us about the recent feedback from the laboratory."

Sitting in the spacious business car, Qiao Yu started the meeting directly.

The work of other people in the project team is just to do some optimization, but their work is very important.

If the current engineering prototype is used as an ordinary computer, it is probably no different from a low-end or mid-range personal computer of six or seven thousand yuan.

Watch videos, play games, whatever.

But if a computer is going to cost 50,000 yuan, then naturally those additional functions have to come into play.

When using functions that are only available in quantum simulation, frequent errors are naturally unacceptable.

They have been working together for more than four years, and Wang Xingyao has long been accustomed to Qiao Yu's style. After hearing Qiao Yu's call, he immediately started to report.

"Based on the updated algorithm, the latest simulation of the Hubbard model ground state energy data of 100 qubits yesterday showed that when the modal dimension base point is less than 30, the compression rate can reach 80%, the entanglement fidelity is 99.2371%, and the theoretical value of the calculation error is less than 0.5%.

模态基点在30至70之间时，压缩率为67%，纠缠保真度为87.637%，计算误差在8%至12%之间。

When the modal base point exceeds 70, the compression rate is only 52%, the entanglement fidelity drops to less than 43%, and the error value is less than %.

From this, we can conclude that the updated algorithm is at least 30% better than the previous generation algorithm when running in low quantum mode, but when it starts running in high mode, the gap with the previous algorithm is not large. "

After finishing the report, Wang Xingyao glanced at Zheng Zefeng.

The new version can be said to be completed by the cooperation of the two. Although the result did not meet expectations, both of them tried their best, so now Lao Zheng has the say.

I have to say that Wang Xingyao actually doesn’t like holding such impromptu meetings.

Being too close to Qiao Yu, the pressure is a bit too much.

Although Qiao Yu was still young, he always felt creepy when those eyes stared at him.

"Ahem, well, let me say a few words. I've also read this lab report and analyzed it. I think the main problem is that the compressed tensor network structure is not compatible with the current Q-RAM fractal storage physical addressing mode, which causes memory access conflicts.

I haven't had time to raise this question because I was busy with the Sky Shield exercise during this period. So I am also considering talking to the engineers from HiSilicon, Changjiang, and Changxin to update Q-MemFlex synchronously.

Especially for fractal 3D stacked resistive random access memory, I think the current Q-RAM Core still has some problems in the original design, and the dynamic mapping of modal parameters is not supported well enough. "

Zheng Zefeng understood Wang Xingyao's look and took the initiative to take over the conversation.

In fact, he didn't want to jump out at this time, but there was no way. He was under more pressure than Wang Xingyao.

After all, he was the only number theory expert in the group and had direct contact with Qiao Yu. Whenever something bad happened, Qiao Yu would always think of him first.

"No, Lao Zheng, we can't pass the buck! Don't change the hardware design all the time? It's not easy to design chips!
It sounds simple as you say. The Q-MemFlex structure has a total of four layers. If the Q-RAM Core is changed, the other three layers will have to be redesigned!
If we start designing hardware now, I don’t think it will be finished by the New Year next year, let alone three months. Anyway, don’t even think about changing the memory architecture now.

If we really want to make a move, we have to wait until the second generation. Now my requirement is to solve the problem through core algorithm design. It is true that we are engaged in scientific research, but we cannot think about problems completely from the perspective of scientific research.

We also have to think from the perspective of the enterprise. The cost of moving hardware is too high now. If we start all over again and have to re-tape out and verify, how much money can you afford?"

Zheng Zefeng smiled awkwardly.

Although he has no shortage of money now, how can he say about researching chips...

On average, a company has to spend at least 600,000 yuan on salary, various benefits, social security, provident fund, etc. for a chip R&D engineer.

It is not easy to find people who are researching and developing analog quantum chips. The companies that are mainly responsible for memory design and manufacturing have at least 4,000 to 5,000 engineers.

Just to support people costs nearly three billion a year...

He certainly can't afford it.

“We should quickly launch the first generation of products to help upstream and downstream companies recover their losses and also teach foreign technology companies a lesson.

You should know that the quantum simulation computer project is in conjunction with the Sky Shield exercise, which is designed to let foreign investors know where the safest place in the world is.

Quantum simulation computers represent the most advanced productivity in the world today, allowing them to know where the most valuable investments are.

The two complement each other! So as long as we can solve 90% of the problems without changing the hardware, we will win! It’s time for us to step up!

In fact, I have carefully read the laboratory's analysis report before, and I have an idea: insert a virtual mapping layer between the physical fractal address of the Q-RAM Core and the compressed tensor logical address.

For example, the EU_xyz coordinates are dynamically adjusted through the quantum state topology perception algorithm. At the same time, the physical banks of Q-RAM are partitioned asymmetrically, and the heads of compressed tensors with high frequency access are assigned to the low-conflict bank group. What do you think? "

(End of this chapter)