### 产品中心

### 主页 > >

GETTING computers to think like humans is tough but what about the opposite? A new mathematical framework gets humans to think like machines in order to harness the power of automated proof-checking. The framework provides the possibility of proofs that can’t be wrong and so wouldn’t need to be laboriously checked by people. It could also be the first step towards computers making their own mathematical leaps. In this new framework, proofs couldn’t be wrong and so wouldn’t need to be checked by people Today, mathematics is based on set theory, the study of collections of objects. But this doesn’t suit computers, because although all mathematical objects can be expressed as sets, logical statements cannot. Instead, automated proof-checking software uses type theory, in which objects and statements both count as types. To build this capability into the normal work of mathematicians, Vladimir Voevodsky of the Institute for Advanced Study in Princeton, New Jersey, and his colleagues have created a 600-page instruction manual on how to shift from sets to types (bit.ly/HoTTBook). Team member Andrej Bauer of the University of Ljubljana, Slovenia,