I know how Professor South China University of Technology worked it out!
Probability = (1138-14+1) * c (514-14,138-) 1 125)/c(5 14， 1 138)
c(m，n)=n！ /[m！ (n-m)！ ]
In molecules:
( 1 138- 14+ 1)* c(500， 1 124)= 6.477657657 187595447053 172330 130984 e+336
And its subtraction is based on the following sequence.
c(50 1， 1 125)= 1.292945546426236468 127 1922 18 1826 e+334
c(500， 1 125)= 1.036425 1500 1527 1528507572820949 e+334
...
c(495， 1 125)= 3.249998708 136956622063467350886 e+333
...
c(490， 1 125)= 9.3 13 1 1672993670845084586725397 e+332
...
c(400， 1 125)= 2.3589327467028 15332474544 1556945 e+3 16
...
c(250， 1 125)= 1.8234654 15745 1435860787249582267 e+257
...
c(2， 1 125)=632250
It can be seen that the maximum term c(50 1, 1 125) is greater than (1138-14+1) * c (500, Therefore, these subtractions have little influence on (1138-14+1) * c (500, 1 124), and the calculation accuracy is 0.01.
Then it is simplified to probability = (1138-14+1) * c (500,1124)/c (514,
= 1 125* 1 124! /(500! 624! )/ 1 138! /(5 14! 624! )
= 1 125* 1 124! 5 14! /( 1 138! 500! )= 1.4%
However, if you add exactly the same units, the probability will be further reduced. If each unit has 46 people, the probability is 0.76%, which is close to the warning line of 0.5%.