erikbern

import asyncio


class AsyncConstructorMeta(type):
    """Metaclass to support asynchronous constructors in Python.
    
    Basically we're exploiting the fact that __new__ can return anything in Python.
    So we're taking the old __init__ code, removing it from the class, and instead,
    we create a custom __new__ method that returns a coroutine wrapping the original
    constructor.

erikbern

class MyDict(dict):
    def __init__(self):
        self._dict = {}

    def __getitem__(self, k):
        print(f'Looking up {k}')
        return self._dict[k]

    def __setitem__(self, k, v):
        print(f'Assigning {k} to {v}')

erikbern

class MyDict(dict):
    def __init__(self):
        self._dict = {}

    def __getitem__(self, k):
        print(f'Looking up {k}')
        return self._dict[k]

    def __setitem__(self, k, v):
        print(f'Assigning {k} to {v}')

erikbern

import random
import numpy
from matplotlib import pyplot
rs = numpy.random.randn(1000)

xs = rs[1:-1] - rs[:-2]
ys = rs[2:] - rs[1:-1]
pyplot.scatter(xs, ys)
pyplot.show()

erikbern

import numpy
import scipy.optimize
from matplotlib import pyplot

cs = numpy.linspace(0.01, 0.99, 100)
ks = []

for c in cs:
    def f(log_k):
        k = numpy.exp(log_k)

erikbern

import numpy
import random
from matplotlib import pyplot

percentiles = [50, 75, 90, 95, 99]
latencies = [[] for p in percentiles]
loads = []

n = 100000
for k in numpy.linspace(0.01, 1.0, 100):

erikbern

# Uses American Community Survey data to estimate property taxes
# https://www.census.gov/programs-surveys/acs/
# The data is a f-ing PITA to parse, but here's an attempt

import bs4
import csv
import io
import os
import requests
import sys

erikbern

from num2words import num2words
from matplotlib import pyplot

lang = 'de'
words = [num2words(i, lang=lang) for i in range(1000000)]
fig = pyplot.figure()
ax = fig.add_subplot(111)

ax.semilogx([len(word) for word in words], color='green')

erikbern

import matplotlib.pyplot, numpy, scipy.stats, seaborn

for i in range(50):
    p, = scipy.stats.uniform.rvs(size=1)
    a, b = scipy.stats.geom.rvs(p, size=2)
    if a == b == 1:
        continue
    print(a, b)
    x = numpy.linspace(0, 1, 1000)
    y = scipy.stats.beta.pdf(x, a, b)

erikbern

def ll_to_3d(lat, lon):
    lat *= math.pi / 180
    lon *= math.pi / 180
    x = math.cos(lat) * math.cos(lon)
    z = math.cos(lat) * math.sin(lon)
    y = math.sin(lat)
    return numpy.array([x, y, z])