 So, the next talk is about good randomness tests, and by Zhuang Ji Zhu, Yuan Ma, Jiang Lin, Jia Zhuang, and Zhu Wu Yin, and Zhu will give the talk, please. Hello, everyone. As we all know, RMB, they are doing many cryptographic applications, and these people have a lot of thoughts on such a quality of the data, and under the non-heritage, and finally, we've got a conclusion to this text, and for these tests, and there are 15 test animals to verify the hypothesis from different aspects of the everyday pollution. And then I will give an introduction to the whole test procedure. First, the whole sequence is divided into 10 sequence blocks, here and once on, and then we will do the test at her, and come to the left side of the vitals, and do the first level test, and the second level test. And the first level test has the capacity of having a new quantity of the left side of the vitals, and I come to the new vitals called RMB. The 15 test animals are divided into two blocks. Here I will talk about our contribution. We found out that the p-value of the phenomenal vitals are anchored by the second level test, although it is proper to do since the stated p-value p of phenomenal vitals is a standard of terminal value of v, and the problem comes from the absolute value of p-value of this animal from the terminal value of v of this abnormal vitals. We can see that there is one and two anchored, and for stated p-value test, we can see that there are three and four anchored, but since two and three are not anchored, this is the problem. Here I will give an example to show how this problem adapts to testing. To construct, for example, we do the testing if for each block it is more than that of the block, otherwise it keeps us locked and changed. The process sequence is significantly faster, and the number of one is larger than that of zero in all blocks. However, the process sequence still has a very high probability to pass the country. If the zero and one have equal rules for each block, all the p-p are greater than what is thought of. For both of the new match p-values, p-value is very nature, since the problem comes from the abandonment after the value, and we check the corresponding p-value instead of p-value. As the other cataclyses your p-value and change, p-value also includes the test parameters. Sometimes the observe is inconsistent with the zero, sometimes there is a difference. As the p-value has p-value and p-value test, it is more powerful with sensitivity. Here is the definition. So actually, that p-value is also different to the p-value we have to read in theory. Here I just gave the details. Another improvement is about accurate distribution. As we all know, with any infinite p-values available, and our Taylor physics, all the figures only have problem with the p-value, and the dashed-land with no symbol is... p-value also has p-value and p-value, and the dashed-land represents the... You can see the p-value line is closer to the unit, and the cumulative function between p-value and p-value is the parameter in theory, that p-value is the p-value we have on the following yarn days of p-value. So are there any questions? I have one question myself. So as I understand you, in the beginning you had this artificial distribution that tricks the NIST suite, and now you have a better test that is not tricked by it. But could it be that there is some other artificial distribution with different ideas that would actually then... that would be biased and pass the q-value-based test? So you have a distribution that passes the p-value test, but it's biased. Can it be that there's another distribution, a different one, that is biased, and passes your new test, the q-value-based test? Okay, thank you. Any other questions? Then let's thank the speaker again. And that was the end of this session, everyone have a nice evening.