version_timings.py

import inspect

import matplotlib.pyplot as plt
import numpy as np
import pandas as pd
import perfplot
import scipy.optimize

import flux_correlation_functions

HOURS_PER_DAY = 24
DAYS_PER_YEAR = 365.2425
HOURS_PER_YEAR = HOURS_PER_DAY * DAYS_PER_YEAR
N_YEARS_DATA = 4

STARTING_PARAMS = dict(
    daily_coef=0.2,
    daily_coef1=0.6,
    daily_coef2=0.5,
    daily_width=0.4,
    Td=30,  # weeks
    dm_width=0.3,
    dm_coef1=0.1,
    dm_coef2=+0.3,
    ann_coef1=+0.3,
    ann_coef2=+0.3,
    ann_coef=0.1,
    ann_width=0.3,
    Ta=30.0,  # years
    resid_coef=0.1,
    To=2.0,  # weeks
    ec_coef=0.7,
    Tec=2.0,  # hours
)

for coef, val in STARTING_PARAMS.items():
    STARTING_PARAMS[coef] = np.float32(val)

LOOP_NAMES = [
    corr_name
    for corr_name in dir(flux_correlation_functions)
    if corr_name.endswith("loop")
]
LOOP_FUNCTIONS = [
    getattr(flux_correlation_functions, corr_name)
    for corr_name in CORRELATION_FUNCTION_NAMES
]

NUMPY_NAMES = [
    corr_name
    for corr_name in dir(flux_correlation_functions)
    if corr_name.endswith("numpy")
]
NUMPY_FUNCTIONS = [
    getattr(flux_correlation_functions, corr_name)
    for corr_name in CORRELATION_FUNCTION_NAMES
]

NUMEXPR_NAMES = [
    corr_name
    for corr_name in dir(flux_correlation_functions)
    if corr_name.endswith("numexpr")
]
NUMEXPR_EXPRESSIONS = [
    getattr(flux_correlation_functions, corr_name)
    for corr_name in CORRELATION_FUNCTION_NAMES
]