The GUDHI library is a generic open source C++ library, with a Python interface, for Topological Data Analysis (TDA) and Higher Dimensional Geometry Understanding.
In DTMDensity (and Tomato), when the points are in "high" dimension, we recommend using a small value for dim instead of the default: the ambient dimension. But if we recommend it, why not make it the default behavior? Something like
if dim=None and the deduction would choose a value larger than 10 (?), then just pick 10 (or even go down to 2 while we are at it?) and print a warning advertising the next point
if dim is specified by the user, they can use any value they like
DTMDensity (without normalization) ends with density = dtm ** (-dim / q). This exponentiation is pretty useless if we are only going to use it in a log in Tomato's logDTM. Without it, only q needs to be small enough, dim becomes unused, and the behavior matches Tomato's doc before #1074.
In DTMDensity (and Tomato), when the points are in "high" dimension, we recommend using a small value for
dim
instead of the default: the ambient dimension. But if we recommend it, why not make it the default behavior? Something likedim=None
and the deduction would choose a value larger than 10 (?), then just pick 10 (or even go down to 2 while we are at it?) and print a warning advertising the next pointdim
is specified by the user, they can use any value they likeDTMDensity (without normalization) ends with
density = dtm ** (-dim / q)
. This exponentiation is pretty useless if we are only going to use it in a log in Tomato's logDTM. Without it, onlyq
needs to be small enough,dim
becomes unused, and the behavior matches Tomato's doc before #1074.