I have got the answer from the Stata.com.
We could type "help datatypes" and read the section "Precision of numeric
storage types". For additional details, we could read the Stata Blog
posting "The Penultimate Guide to Precision", written by William Gould,
StataCorp's President.
https://blog.stata.com/2012/04/0 ... guide-to-precision/
In summary, the numeric numbers could not be precise in format float.
You could type x=float(2.6) or ,make the variable in format double to resolve the question.