Chapter 13: Breakthrough

"Graduated early?" Although he saw that Yan An's performance was a little cramped, Liu Xi was still surprised by what he said, "When?" ”

She didn't expect it to be so sudden, she didn't have a tendency to graduate early when she had hot pot with Yan An next year ago, how could she suddenly get such a news when she met again.

"You're here today to go through the graduation procedures, aren't you?" She then reacted that graduating early and withdrawing from school are completely different things, the latter can be done at any time, and the former can only be done if the graduation conditions are met.

In other words, Yan An has completed all the courses in his major in the past six months.

"I did it a few years ago, and it's been a while now."

"Then ...... you still come to school now" Liu Xi looked at him, and suddenly felt that the distance between the two was a little farther away, and it became unattainable.

"Part-time job." What research is being done in the project, what position he is in the project, these are all to be kept secret, and Yan An used such an answer to fool Liu Xi.

"Won't it be hard to see you again after that...... The canteen has opened a new window, and the yellow stewed chicken is spicy, delicious, and there is more chicken.

It's rare to meet you, so I'll invite you to eat. ”

didn't forget that Yan An was very cramped here, and he was not feeling good, Liu Xi quickly regained his strengths after a little loss, and pulled him towards the cafeteria.

Gao Shenghan silently followed behind him, keeping a certain distance from the two.

Protecting Yan An is not to monitor Yan An, and he is not a pervert, when there is no need for 24-hour personal protection, he will naturally leave enough private space for Yan An.

This is not only Yan An's needs, he also needs them.

No one can stand being invaded 24 hours a day, with no personal space to speak of, not even between husband and wife, let alone them.

Before eating, Yan An explained that he was usually on the side of the algorithm building, which made the senior sister's eyes shine, and she hurriedly fanned herself with her hand, pretending to be spicy to cover up the uncontrollable corners of her mouth.

The yellow stewed chicken she said was not authentic and put aside, anyway, it was really spicy, and even Yan An and Liu Xi, who had advanced anti-spicy ability, couldn't stand it.

But the taste is also very good, just for the appetite of the two, even if it is spicy, they are still persistently pouring rice with the thick soup underneath, and even making two large bowls.

Obviously, the two of them didn't speak during the meal, but the atmosphere made people feel very harmonious at a glance from a distance.

Liu Xi, who put down the bowl after eating, glanced at Gao Shenghan, who had already eaten and was sitting at the next table waiting for Yan An, she had no idea of pulling Yan An to enjoy the nightlife together, and she also knew that he was a person without nightlife.

"Then I'll go back first, and next time I find something delicious, I'll call you out to play." Liu Xi waved his hand generously and said.

The two separated, and Yan An, who returned to the algorithm building, was in a much better state after adjustment, and there was no barrier to condensed thoughts in front of him.

Even that film is gone.

It's just that this doesn't mean that he can easily solve this problem, it's just that Yan An will start all over again, thinking about the solution from one direction with a new thinking that is not framed by himself.

It's a pity that there was no casual mention by the senior sister during the meal, and then he had an epiphany.

The kind of thing that suddenly opens up the second pulse of Ren and Du is probably not applicable to this problem - this is a problem that is difficult to solve even if it is connected to the second line of Ren and Du.

It doesn't take so much inspiration to solve it, for Yan An it is like horizontal and vertical strokes, and the problem is how to use these limited strokes to form a word that does not exist in the dictionary, but has practical meaning when written.

"Continuing to study in this direction should solve the problem of ciphertext inflation.

The low ciphertext expansion means that the efficiency of the algorithm can be improved to a certain extent, and if you let them study it in this direction, they should be able to figure it out. ”

This was the only feasible direction that Yan An had found all night, and after informing Gao Shenghan of this, he fell asleep in a groggy state, and he didn't have time to go home, so he fell asleep as soon as he lay down on the marching bed in the algorithm building.

Don't look at the lack of physical work, mental activities are also extremely energy-consuming, especially the consumption of the whole concentration.

Yan An has been using his brain at such high intensity for two or three days in a row, and it is naturally difficult to keep up with his energy.

He slept until three o'clock in the morning, and no one woke him until he slept here.

When he woke up and found that it was half past nine, he didn't bother to brush his teeth and wash his face, he got up as soon as he turned over and hurriedly walked towards the research area, and couldn't wait to ask, "How is it?" Is there any progress? ”

"The problem of low ciphertext expansion has been solved, and the solution is very similar, we just need to transplant it." Ni Chuanjun replied immediately.

Compared with the fully homomorphic encryption algorithm that specifically solves the ciphertext expansion problem, the low ciphertext expansion effect of flawless is not bad, but it is only slightly inferior, and it can be called the ultimate for non-polynomial computing flawless.

"That's good." Yan An breathed a sigh of relief, he was worried that if this problem could not be ported and reproduced like the key problem, it would mean that they had one more job.

Fortunately, it was solved smoothly.

He then went to brush his teeth and wash his face with confidence, and when he came back, he was refreshed, "It's better to relax and think more, part of the LWE decryption of a batch of LWE ciphertext can be written as a linear transformation As+b, where A belongs to ......

The resulting elements are not individually encrypted, but encoded, and the repacked key is exported as a single RLWE ciphertext.

And since the key is usually a low normal vector, the scaling factor △r is used to save the precision......

In other words, we need to evaluate the ciphertext and 'plaintext' values A and b of the linear transformation As+b given the coding vector Ecd(s,△r), so that the matrix A can be a tall matrix.

A new algorithm is introduced to avoid the inadequacy of existing methods to cover the 'tall' and 'short' matrices.

The resulting override is enough to reduce the key by an order of magnitude without losing precision, even for non-polynomial calculations! ”

Yan An said excitedly, he was talking while writing down his thoughts on the pen whiteboard.

Until the last stroke was completed, the smile on his face could no longer hide the bright bloom, "This method should be feasible!" We solved the problem of inefficient non-polynomial calculations!

Next, you can officially enter the strong applicability study! ”

He has already given the method, how to achieve the application is Li Junwei's business, this is not difficult, even if you find a doctoral student, or even a graduate student, you may be able to do it.

Yan An's energy is now focused on tackling the third direction, and the applicability of the algorithm represents the breadth of the market.

Compared with the first generation of public key cryptography, the current network applications are completely different and more complex, including mobile Internet, satellite communications, Internet of Things, big data, cloud computing and other network applications.

The more applicable the algorithm, the larger the scope, which means that the market is also larger.

It's just that Gao Shenghan has different ideas about this, and there is no need to really make a perfect full homomorphic encryption.

It's not perfect, either.

It's obviously easier to specialize on top of existing flawless for different web applications.

Yan An, who was exhausted by the efficiency optimization problem, agreed with this approach, which at least made it much easier for him.

As long as the other aspects of the improvement are steadily advanced, the first phase of the Flawless project can be declared over.