I am trying to model data which comes in the form of “two cones”. You can imagine this as two cones of different negative slopes and heights positioned on a x,y grid and at every x,y the observed value is the maximum between two cones + some noise (so kind of like maximum of two mountains over a gri…

There is a transformation, pm.distributions.transforms.univariate_ordered that I think is what you are looking for. You could use it as follows: with pm.Model() as m: x0 = pm.Uniform("x0", -7, 7, size=(1,)) offsets = pm.Uniform("offsets", 0, 12, size=(ncones-1,), …

Fitting multiple cones to 2D data, how to impose ordering constraint on parameters

Questions v5

jessegrabowski September 1, 2023, 8:18am 11

You often need to clip when sampling indicies anyway, see here.

Topic		Replies	Views
Complex ordering mixture models	7	237	April 25, 2024
Constrain one variable to be greater than another Questions	14	3481	January 30, 2018
Order statistics in PyMC3 Questions	17	3530	January 18, 2024
Error using Ordered() in Mixture of Uniform Distributions Questions	14	1844	February 7, 2019
Ordered transform internals development , bug	18	133	August 20, 2024

Fitting multiple cones to 2D data, how to impose ordering constraint on parameters

Related topics