little script to understand nominal-use-case histogram sizes

test_hist_size.py

from coffea import hist
import numpy as np
import random
import string
from tqdm import tqdm

N_ds = 45
N_reg = 3
N_ds_full_syst = 6
N_syst = 53
N_syst_reduced = 10

def get_rand(N=6):
    return ''.join(random.choices(string.ascii_uppercase + string.digits, k=N))

dsnms = [get_rand() for _ in range(N_ds)]
regnms = [get_rand(4) for _ in range(N_reg)]
systnms = [get_rand(10) for _ in range(N_syst)]

h = hist.Hist(
    'Events',
    hist.Cat('dataset', 'Dataset'),
    hist.Cat('region', 'Region'),
    hist.Cat('systematic', 'Systematic'),
    hist.Bin('genflavor', 'Gen. jet flavor', [0, 3, 4]),
    hist.Bin('pt1', r'Jet $p_{T}$ [GeV]', [400, 450, 500, 550, 600, 675, 800, 1200]),
    hist.Bin('msd1', r'Jet $m_{sd}$', 23, 40, 201),
    hist.Bin('ddb1', r'Jet ddb score', [0, 0.64, 1]),
    hist.Bin('mjj', r'$m_{jj}$ [GeV]',[500,1000,2000,4000,6000,10000]),
)

n_fill = 1_000_000

genflavor = np.random.randint(0,4, n_fill)
pt1 = 300 + np.random.exponential(1000, n_fill)
msd1 = np.random.normal(100, 30, n_fill)
ddb1 = np.random.uniform(0, 1, n_fill)
mjj = 400 + np.random.exponential(1000, n_fill)


for ids, ds in tqdm(enumerate(dsnms)):
    for reg in regnms:
        for syst in systnms if ids < N_ds_full_syst else systnms[:N_syst_reduced]:

            h.fill(
                dataset=ds,
                region=reg,
                systematic=syst,
                genflavor=genflavor,
                pt1=pt1,
                msd1=msd1,
                ddb1=ddb1,
                mjj=mjj,
            )

print(h)