Chapter 363 The idea you mentioned was proposed by me.

After confirming that everyone had no other questions or concerns, Chang Haonan returned to the topic of the meeting again.

"Since there is no problem, let's follow the reverse order on this PPT and first introduce a new parametric generation method of spatial grid structure."

When he said this, the scientific research assistant next to him also pulled in a blackboard for him from outside.

Although most of the content has been put on the PPT, considering that many of the people sitting below are just lecturers who have just joined the job, Chang Haonan felt that it was necessary to add some details in some places.

"Since the current workload is basically saturated (and the scientific research points have been exhausted), we will not add graphics drawing functions to the software for the time being, and considering the difficulty of promoting the use of the software, even if it is added in the future, we will definitely retain the same The interfaces of common software on the market such as CAD, CATIA, and 3DSMax do not leave us much room for manipulation in this aspect, so we need to use boundary representation as a geometric model expression."

"In terms of topological structure, boundary representation consists of vertices, edges, rings, surfaces, shells and bodies. Vertices correspond to points in geometry, edges correspond to curves in geometry, and surfaces correspond to surfaces in geometry. The simplest CAD geometry The topology of the model is like what I am drawing on the blackboard..."

Chang Haonan drew a simple rectangular diagram on the blackboard while introducing some basic principles related to geometric models.

"For some projects with relatively simple models, in order to continue the usage habits and ensure calculation efficiency, we will use delaunay triangulation theory to generate triangular or tetrahedral meshes. Delaunay triangulation is the dual diagram of the Vorinoi diagram."

"We assume that P={Pi}(n, i=1) is a set of non-overlapping points in a given d-dimensional Euclidean space Ed. Then for each point pi∈P, its corresponding Vorinoi element V can be defined (Pi), V(Pi) is a region near pi that satisfies the following conditions..."

V(Pi)={x∈Ed：||x-pi||≤||x-pj||, j≠i}……

"..."

In the huge conference room, except for Chang Haonan's voice and the occasional sound of chalk rubbing against the blackboard, there was almost no noise.

Everyone below, from the young lecturer who had just joined the job to Song Jian sitting in the middle of the last row, was either writing hard or thinking.

Since the system has provided all the knowledge needed to build the entire software framework, Chang Haonan's teaching method can be said to be from a high level, starting from the most basic but not simple basic theory, gradually rising to expansion, and even through Improved algorithm.

At the end of the 90s, computer-aided design was still in its infancy in China.

Even the so-called cutting-edge topics are mostly focused on the application level, and few people pay attention to the underlying theory of simulation tools.

This can be seen from last year when Chang Haonan and Yao Mengna first arrived at the 601 Institute, Lin Shikuan and others did not even fully understand the art of parametric design.

On the one hand, the current software tools are indeed not very easy to use. The simulation results for many practical problems are often extremely outrageous. It is often necessary to make judgments based on physical models or experience. For some experienced personnel, they may not be able to intuitively understand what is happening. superiority.

On the other hand, because it missed almost all three technological revolutions, China is indeed lagging far behind in relevant theoretical foundations, and it urgently needs to find out and fill in the gaps.

In this case, even if some people think of going deeper into research, there are simply not enough objective conditions.

Scientific research ultimately requires the support of the general environment.

But Chang Haonan's lecture was like a key, directly opening a new door to the audience sitting below.

This is true both at the theoretical level and at the applied level.

“For tetrahedral meshes, traditional regional decomposition methods mostly introduce additional constraints to the mesh generation of sub-regions. Take the classic method of using triangular meshes as sub-region interface meshes as an example. When the interface is If the angle formed by the model surface is too small, it will easily lead to the failure of the interface mesh generation and the deterioration of the sub-region mesh quality, so this problem should be avoided as much as possible..."

Having said this, Chang Haonan picked up the cup and took a sip of water.

And in this gap, someone below raised his hand to indicate that he wanted to ask a question.

"Hello, Mr. Chang, I am Wei Fangchao, a researcher from the Institute of Computing Technology, Chinese Academy of Sciences."

"As for the problem you just mentioned that the angle between the interface and the model surface cannot be too small, we have come across a new method in actual work, which is to convert the original triangular mesh into a pseudo-quadrilateral mesh. , that is, a quadrilateral with an infinitesimal length of one side. In actual operation, this situation can be completely avoided. What do you think of this idea?"

When Chang Haonan heard that the other party said there was a new method, his whole body became energetic almost instantly.

What matters in class is also positive feedback. If someone is willing to interact, especially high-level interaction, it is actually a very effective incentive for those on the stage.

However, after hearing what the other party said in detail, Chang Haonan almost choked on the water he had not yet swallowed.

Even Yao Mengna, who was sitting on the side of the first row, lowered her head and covered her forehead.

Judging from her shaking shoulders, she must be suppressing a smile.

"Cough cough..."

Chang Haonan calmed down a little.

"To be honest, I can't comment very well."

"Because this idea happened to be what I proposed last year..."

After all, invisible pretense is the most deadly.

The researcher who asked the question looked at Chang Haonan on the podium with his mouth half open, his face full of shock.

There was also a burst of laughter.

In fact, this was one of the many problems that the digital design team faced when they first arrived at the 601 Institute during the parametric modeling process of the Bayan San Project design plan. Chang Haonan solved it by relying on this somewhat tricky method. question.

But even he didn't expect that this idea, which was only a temporary means, would be promoted through unknown channels.

It has to be said that simulation methods in this era are really lacking.

"Getting back to the subject, in fact, this problem should be more professionally called the parallel unconstrained tetrahedral mesh generation method. Although the pseudo-quadrilateral idea I proposed at the time can ensure the mesh quality and the success rate of mesh generation, the application object The solution threshold cannot be too large, otherwise it will easily affect the numerical stability and ultimately reduce the robustness of the method and program."

"On the other hand, that method is also difficult to operate. If manual operation is used, it is quite a tedious task to convert a large number of triangles into pseudo-tetrahedral meshes. If automatic generation is used, then a pseudo-tetrahedral mesh with a side length of 0 The current generation method is very difficult to calculate and requires a huge amount of memory. The time consumed is often beyond our ability to afford. Regarding this point, if there are comrades here who are engaged in architectural design, they will be deeply aware of this. Experience.”

When he said this, Chang Haonan noticed that many people below nodded, obviously they had been troubled by this for a long time.

"I wonder if I explained it clearly?"

Chang Haonan looked again at the questioner.

That is of course clear.

The latter nodded quickly and sat back.

Thanks to book friend [775544] for the 10000 starting coins~