In the Dilithium draft, section 3.2.1, the table lists "memory in bits" (listing fractional bits such as 97.8) with no further explanation.
The table should either explain what this is (the log2 of the estimated amount of memory needed for the best known attack against MLWE - although, without the estimated work effort, this is less valuable than it could be), or omit it entirely.
My opinion: I suggest you omit it (I don't see how this would help an implementor)
In the Dilithium draft, section 3.2.1, the table lists "memory in bits" (listing fractional bits such as 97.8) with no further explanation.
The table should either explain what this is (the log2 of the estimated amount of memory needed for the best known attack against MLWE - although, without the estimated work effort, this is less valuable than it could be), or omit it entirely.
My opinion: I suggest you omit it (I don't see how this would help an implementor)