When a new node is added in teradata then the how the remapping will take place. It will calculate the hash for full data to allot data to new nodes and old nodes, OR Hashing will take place for new node only (something like Consistent Hashing which is there in Amazon Dynamo).
The hash stays the same, it is not recalculated. A subset of the bits of the hash are designated as the "hash bucket". In Teradata there is a map that assigns buckets to AMPs. When new AMPs are added to the system, a proportional subset of the buckets are reassigned to the new AMPs. Then the data from existing AMPs corresponding to the moved buckets is moved to the new AMPs. The bucket reassignment is done such that the minimum amount of data needs to be moved.
There is more detail in the documentation if one searches on "hash bucket".
Thanks Walter for answer...I think I need to read this concept (Hash Bucket) in deep to understand.