File size: 1,538 Bytes
5032722 | 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 | #!/usr/bin/env python3
# -*- coding: utf-8 -*-
"""
Created on Tue Jun 24 15:07:29 2025
"""
from jax import numpy as jnp
import numpy as np
def init_time_array(args):
### init from geometric grid, like in cherryML
if args.pred_config['times_from'] == 'geometric':
t_grid_center = args.pred_config['t_grid_center']
t_grid_step = args.pred_config['t_grid_step']
t_grid_num_steps = args.pred_config['t_grid_num_steps']
quantization_grid = range( -(t_grid_num_steps-1),
t_grid_num_steps,
1
)
t_array = [ (t_grid_center * t_grid_step**q_i) for q_i in quantization_grid ]
# make sure it's small times -> large times
t_array.sort(reverse=False)
return jnp.array(t_array)
### read grid of times from flat text file
elif args.pred_config['times_from'] == 't_array_from_file':
times_file = args.pred_config['filenames']['times']
# read file
t_array = []
with open(f'{times_file}','r') as f:
for line in f:
t_array.append( float( line.strip() ) )
# make sure it's small times -> large times
t_array.sort(reverse=False)
return jnp.array(t_array)
### figure out time quantization per sample... later
elif args.pred_config['times_from'] == 't_quantized_per_sample':
raise NotImplementedError |