search

libscran / qdtsne

Quick-and-dirty t-SNE in C++
https://libscran.github.io/qdtsne/
MIT License
9 stars 2 forks source link

Initialization parameters for qdtsne ? #7

Open shrit opened 1 year ago

shrit commented 1 year ago

Hi,

I would like to use your library but I am not able to understand the following parameters?

What is the matrix type in data.data()? is it just a a float*?

I suppose we should always do the initialize_random() with N, but what isN, is the number of dimensions after reduction ?

Also what is D ? I would suppose it is the number of end dimensions.

LTLA commented 1 year ago

What is the matrix type in data.data()? is it just a a float*?

It's templated, so it can be either a float or a double, see specific docs here.

I suppose we should always do the initialize_random() with N, but what isN, is the number of dimensions after reduction ?

No, it's the number of observations, see here.

Also what is D ? I would suppose it is the number of end dimensions.

No, it's the number of input dimensions, see the first link.