NEW TO MPSLIB

[jbensabat]

Hello
I am trying to use the library and I have two questions that I would be grateful if you could address.

1. the user has to provide a training image. Does it have to be a full image (nxny data points in 2D or nxny*nz data points in 3D) ? Is it possible to provide a set of data points only (data at sampling locations) ?
2. Is it possible to work with a continuous variable instead of a categorical variable ?

thanks
jac

[cultpenguin]

Hi Jac.
You can use a sparse training image with mps_snesim_tree and mps_snesim_list. See an example here:
https://github.com/ergosimulation/mpslib/blob/master/scikit-mps/examples/ex_sparse_ti.ipynb
I will test if this applies to mps_genesim as well.

You can use a continuous training image with mps_genesim.
You need to set

 O.par['distance_measure'] # Distance measure [1]: discrete, [2]: continous
 O.par['distance_max'] ; # Max distance
 O.par['distance_pow'] ; # Power

See an example here:
https://github.com/ergosimulation/mpslib/blob/master/scikit-mps/examples/ex_genesim_continuous.ipynb

[jbensabat]

hi
thanks a lot for your reply
I am using the c++ front end so this is why I missed the sparse example.
thanks
jac

[cultpenguin]

I only just added the example of using a sparse TI. If you make some nice examples of the use of MPSlib feel free to add them or send them to me and I will.

• Thomas