I think answer should be (B).. because we can derive low precision matrix from high precison by some loss of information. eg: high precision one return a. bcdefg ( a 6 bits decimal.) we can derive a low precision depends on how much we want a. bcd (by loss of some information, three bits here).. But on the other hand we can not derive a high precision from a low precision because once a calculation has been done that means once infomation has been lost we can't recover it back.. so deriving high precision from low precision is not possible as per my understanding......
Arjun Sir pls help!!