I am trying to model a process where I only get to see the historical maximums at any point in time. A simple generative setup in numpy for the data is as follows: import numpy as np import math # Model parameters mu = 10. sigma = 1. # Sample generation ts = np.arange(100) obss = [] obs = -math.…

I see! So to synthesize, the code would look like: # Sample parameters N = 10 mu = 10. sigma = 1.0 import numpy as np import pymc3 as pm # Sample generation ts = np.arange(N) obss = [] obs = -np.math.inf for t in ts: aux = mu + sigma*np.random.randn() obs = np.max([obs, aux]) obss.append(o…

Modelling a timeseries of cumulative maximum

Questions

Jaime_Sevilla August 2, 2021, 9:14am 3

Interesting! It renders correctly for me

FWIW the first time I loaded the page I got a similar error, but after I reloaded it once it never happened again. Very mysterious!

Topic		Replies	Views
Sampling MaxEnt Distribution Questions	20	2033	September 28, 2017
Sampling time is very long (minutes per sample) Questions	13	4561	November 2, 2017
Adding observed to result of maximum Questions	2	925	August 1, 2021
How to set up a custom likelihood function for two variables Questions	8	13306	March 9, 2018
Averaging over model output that sometimes returns null values? Questions	12	1458	November 16, 2017

Modelling a timeseries of cumulative maximum

Related topics