Chinese entertainment for engineering students

In addition to arXiv, Wang Xu also built a CNKI, um, a Chinese version. The management of arXiv is also handed over to the GIT management committee. There are distributed scene stations all over the world. In fact, the backend support programs of the two websites are very similar, except for the display layer.

As for this kind of website that is open to the whole world, Wang Xu will not hold it in his own hands, nor will he let Chou State completely control it. Instead, he will make it look like the United Nations.

In general, basically no one can completely cut off arXiv's connection with the world, and arXiv has even specially opened an external access automation interface, and you can pay to perform some download tasks similar to crawler tasks.

Because it requires more machines and bandwidth, this service is charged.

As for CNKI, it is more troublesome, because it really needs to negotiate with major copyright agencies. Domestic ones are okay, but international ones are very troublesome, so this thing must be at least semi-commercially operated. Otherwise, the cost may really be endless.

The establishment of these two infrastructures is actually to prepare for the future era of artificial intelligence.

Wang Xu did not experience the complete era of artificial intelligence in his previous life, and in this life, he is using various means to promote the arrival of the era of artificial intelligence. This makes him a little worried. In the era of artificial intelligence in his previous life, China was lagging behind. Visible to the naked eye.

Of course, it may just be a dispute over routes, but most of the so-called artificial intelligence in China is actually large models that are open sourced by others. Wang Xu cannot say that he understands them all, but it is true that many of them are like this.

Wang Xu did not release the most popular models of later generations. Instead, he released relatively early content, such as the convolutional neural network model, which is now the main direction of Wang Xu's laboratory to publish papers.

In the final analysis, this thing is more suitable for the field of image recognition. This is the basis of license plate recognition, handwriting recognition, face recognition and other products. Now the theory is laid out, and after several years of promotion and fermentation , and slowly it can appear in practical application fields.

Of course, there is also a recurrent neural network model that is more suitable for human language. This thing is mainly used for semantic analysis and can analyze the context of the text.

These two types of neural networks were the most popular content in the first period of the 21st century. Wang Xu's laboratory is now slowly publishing papers in this area.

With one or two breakthrough papers and a few expansion papers every year, it can be considered a return to the normal state of a laboratory. Now in the field of artificial intelligence, it is considered a relatively fruitful laboratory and has gained some fame in the industry.

But it is still theoretical. Since China's hardware equipment, software equipment, and network equipment are still in their initial stages, Wang Xu consciously suppressed the output in this area. It is really unnecessary. Once it is made, China will not be able to use it. .

On the contrary, some are provided for the above to decide whether to use them. After all, whether it is neural network learning or training, the data resources required are massive, and the computer computing resources required are almost endless.

Wang Xu even reminded the higher authorities to pay attention to the collection of data. These things will become new gold mines in the future. Of course, there is no need to collect for the sake of collection, but it needs to be done during the process of digitization and informatization.

In fact, according to the development of science and technology in the world, from the 1980s to the future, it can be roughly divided into five stages: digitization, informatization, automation, unmanned, and intelligence.

By the way, this digitization is not the same concept as the industrial digitization advocated by later generations. It is the concept of comprehensive digitization in the Internet of Things era, which includes all-round digitization of factories, equipment, processing processes and products.

The digitization currently being carried out is only for products and design processes, and does not even have rigid requirements for the production process.

Just like the “off the drawing board” project proposed by the country a few years ago, it is an attempt to digitize industrial design in this era.

In fact, there is also a plan that has not been publicized on a large scale, and that is the "letter paper throwing away" project. This is the beginning of China's move towards office digitization.

In later generations, some people called these two projects "double dump projects", but because not many people used them, they were gradually forgotten.

The drawing board project actually started in 1991, but the progress has not been rapid. At the beginning, it was only promoted in key enterprises and large processing and manufacturing enterprises.

However, due to too many constraints, the progress of this project has always been satisfactory. In the previous life, it was probably not until around 1997 that this progress gradually accelerated.

There are many reasons, but the core ones are nothing more than three points: lack of money, lack of people, and lack of independent software.

The 1990s was an era of rapid and vigorous development of the domestic software industry. These software companies basically relied on one-to-one copying of similar overseas software functions for development, but it is true that most of them developed their own algorithms and wrote their own codes. , because there is no need to copy those things.

There are many things that I really don’t know how others implement, so I have to use decompilation to look at their assembly code, then deduce the algorithm, and finally implement it myself.

Wang Xu once worked in a company that developed CAD in his previous life. Their testers compared their products with AutoCAD. You should draw what others draw, otherwise it would be wrong.

There is indeed no creativity, but it was during this process that China's first batch of CAD software products were produced. In the early 1990s, relying on the progress of the "dumping the drawing board" project, they still made some money from selling them.

However, after 19596, optical disc media began to be popularized on a large scale. Various new versions of AutoCAD were all taken away on a disc, which even contained many drawn resource libraries. Since then, Huaxia CAD has plummeted and gradually faded out of the public eye. .

AutoCAD can almost be said to dominate the world. Therefore, Wang Xu dares to say that in fact, these companies must have deep thinking in allowing piracy to become popular in China. They do not rely on the Chinese market to survive. At this moment, China The market cannot support any large software company.

But this kind of laissez-faire, even intentional, can naturally eliminate almost all small competitors, killing almost all their budding possibilities.

Some people may say that they may not think so far and profoundly. This is a conspiracy theory. Wang Xu has no evidence, but this is the result of the facts. Wang Xu's open source plan is to reshuffle the entire software ecosystem from another level.

Use free and open source products to gather the power of the world, and then use these products to form countless domestic and foreign commercial companies. These companies, in turn, nourish the open source community and form a brand new software ecosystem.

Since this thing is highly related to the film and television industry, Wang Xu will naturally pay special attention to the development of these products.

Shuimu himself has a company that makes CAD products. Now that they have an open source bottom layer, they have fully migrated to a new architecture system.

In fact, the core of early CAD was drawing. The word CAD itself is the abbreviation of computer-aided design. It did not really look at product design from the perspective of product design, and it did not have much auxiliary capabilities for the design itself.

On the contrary, it has greatly improved the time and difficulty of drawing. For example, the most troublesome annotations can be completed with almost one click.

In particular, it can reduce the speed of drawing modifications. The electronic file can be modified directly on the original drawing and then re-produced. However, this is almost impossible for finished drawings, which saves a lot of time on repeated drawings.

This is the original intention of the so-called "off the drawing board" project. Freeing engineering and technical personnel from tedious drawings, they can truly spend their time and effort designing the product itself.

The three-dimensional product design brought by Wang Xu is completely different from the traditional design method in concept. He directly presents it through three-dimensional modeling, so that the product can first appear in the engineer's field of vision intuitively. Engineers can according to the function Make 3D modifications intuitively.

You can even do some animated working condition simulations, such as whether there is interference in gear rotation, whether the shaft and the housing interact with each other, and in depth, you can also define materials, automatically generate machining processes, tolerance requirements and other complex operations.

When drawings need to be processed, views in all directions are automatically generated, and complex process drawings such as cross-sections and assembly drawings can be freely generated.

This is digital design in the true sense. Even in the West, only the top aviation design fields now adopt this design model. For example, Dassault in Gaul. Their digital design software is the top in the industry, at least nine It was still the case in the 1900s.

In fact, Huaxia has never slowed down in this regard. The new Leopard recently built by XAC was designed and built using full digital design. In the previous life, the platform they used was CATIA, yes, a product of Dassault.

But in this life, it is said, it is said that they chose Wang Xu’s open source product. Although this product is not mature enough, it is more advanced in basic functions and will be significantly more scalable in the future.

Therefore, they teamed up with many universities in the Imperial Capital, Magic Capital, and Xi'an, including Shuimu, to jointly conduct research and development on many basic software to build a tool platform for digital Feibao.

Wang Xu would also get a thick collection of questions from time to time, and then Wang Xu would give answers based on his own understanding. If he couldn't answer, he would tell them directly, but he did not personally participate.

You must know that aircraft design is still very different from traditional CAD. To design an aircraft, especially a modern third-generation fighter-bomber, you need to use dozens of different professional software.

Most of these software are developed by companies in the aviation professional field, and then charge high fees to iterate to form better products. Similarly, after a period of market verification, these software become increasingly mature and eventually win the entire industry. market.

Of course, this is not a big problem in the West. Now these software can also be sold to China. On the one hand, their own software is not fully mature. On the other hand, they may feel that even if China designs it, it cannot build it.

And if you are used to using your own software, maybe China's security agency will no longer be so secretive. After all, it is much easier to embed something into software than a machine.

But Huaxia has come all the way, and the concept of buying is not as popular in those units. Although it is not non-existent, most military workers are really afraid of all kinds of troubles that may appear at every turn.

Therefore, in this life, they chose open source products that were immature or even incomplete, built their own core tools, and used the money spent on purchasing overseas software to develop them themselves.

This might even slow down the development of their fighter jets, but they still chose here.

Maybe it's a line like this: "I didn't have a choice before, but this time, I choose to do it myself."

This is a big project, and the help Wang Xu can provide is limited. After all, he has never even seen many software products.

But everything remains unchanged, which reflects the importance of open source basic libraries. Some basic libraries that are indispensable for any industrial software are provided in the open source community.

Basic libraries such as two-dimensional and three-dimensional graphics processing, image processing, finite element analysis, various mechanical calculations, various fluid calculations, and even many biological and chemical calculation simulations are now available in the open source community. .

Moreover, due to the adoption of a unified communication standard, the calling and connection of these libraries are very convenient and can be integrated from the bottom. There is no need to connect from a high level like using foreign molded products. Integration is very difficult and can only rely on finished product data. file to communicate.

In the unified design network that Wang Xu conceived for them, these underlying libraries were further encapsulated and called, connected to each other to form a larger function library, and then directly formed a networked work platform based on the function library. This is almost twenty Design concepts years later.

Wang Xu made plans for distributed computing, cloud storage, unified work platforms, and external display boards and gave design opinions.

In this way, emerging network technologies can be fully utilized to make coordination between departments easier, which actually exceeds all existing industrial software products.

Wang Xu even planned an automated manufacturing module, from CAD to CAM, which can save a lot of processing technology coding work in the future. Of course, most of the automated processing equipment is expensive and cannot be bought, and the use of it must be limited.

Of course, it is even more impossible to open the underlying control rules to you. You can only use the programming language he gave you to write automated manufacturing processes.

In the future, if you have your own automated machining center, something with four axes, five axes, or six, seven, eight, or nine axes, you can connect it and directly control production. Let’s take a look at the unmanned workshop.

In fact, this is also an innovation in the concept of military workers. Using a fully three-dimensional design concept to build a new aircraft is a completely new attempt. Even advanced Western companies have not fully realized this.

Most parts still rely on engineers to directly draw two-dimensional CAD drawings. This process relies on the engineers' spatial imagination rather than the capabilities of the software itself.

However, Wang Xu’s support for 3D here can be said to be beyond the times. It not only has a 3D bottom-level library, but also has a complete operating interface. This thing was prepared for his own movie making. Without an interface, how could a computer layman do it? use.

——

(I can’t tell if there is more, and I can’t tell for sure. I’m going out soon, and I don’t know when I’ll be back.)