el-hult’s gists

el-hult / leg_jax.py

Created August 20, 2024 11:08

Jax program that computes the Legendre approximation for the Entropy of a Gaussian Mixture Model, according to Caleb Dahlke and Jason Pacheco https://papers.nips.cc/paper_files/paper/2023/hash/ee860a9fa65a55a335754c557a5211de-Abstract-Conference.html

	import jax
	import jax.numpy as jnp
	import functools


	def gmm_max(logpi, mu, cov):
	"""Return an upper bound the GMM density
	logpi: (K,)-array of the log-weights
	mu: (K, D)-array of the component means
	cov: (K,D,D)-array of the component covarainces

el-hult / MySet1.hs

Last active November 6, 2023 10:43

Two bad implementations of a Set in haskell

	{- A module implementing a simple (and quite dumb) set -}
	module MySet1 (Set, empty, toList, insert, contains) where
	import Data.List(nub)

	{- Set
	A set, backed by a regular list.
	If the list has duplicates, it is nothing the user should notice
	-}
	data Set a = SetC [a]

el-hult / faster.c

Last active February 23, 2023 10:40

See https://el-hult.github.io/2023/02/23/bno055.html

	/*
	This code reads the BNO055 fast using the RPi linux kernel support for I2C.

	for running this file as a program, compile and run!
	gcc -o faster faster.c -O3 && ./faster

	You can also compile as a shared lib, and then call it from python.
	gcc -Wall -shared -o faster.so faster.c && python faster.py

	Page number and table number references are to the BNO055 data sheet.

el-hult / 9_2.py

Last active May 13, 2022 15:57

see https://el-hult.github.io/2022/05/13/vi-example.html

	import numpy as np
	import scipy.stats as sps
	import matplotlib.pyplot as plt
	from scipy.special import digamma

	#
	# Config
	#
	np.random.seed(999)
	plt.rcParams['font.family'] = "monospace"

el-hult / voice_memo.py

Last active May 5, 2022 08:31

see https://el-hult.github.io/2022/05/05/recording-audio.html

	# nice library that gives us access to the microphone
	import pyaudio

	# python standard libraries that come built-in
	import threading
	import queue
	import wave

	# optional libraries
	import numpy as np

el-hult / Pointers.hs

Created March 23, 2022 08:58

A Hasekll module that can store vectors from hmatrix as double arrays, suitable for passing into Fortran and other FFI

	{-# LANGUAGE DataKinds #-}
	{-# LANGUAGE KindSignatures #-}
	{-# LANGUAGE ScopedTypeVariables #-}

	module Pointers where

	import Data.Proxy (Proxy (..))
	import Foreign.Marshal.Array (newArray, peekArray)
	import Foreign.Ptr (Ptr)
	import GHC.TypeNats (KnownNat, Nat, natVal)

el-hult / classical_ols.py

Last active March 21, 2022 11:06

Run a linear regression under the classical OLS assumptions: well specification, homoscedacity and normal errors. Then draw a confidence box and a confidence ellipse.

	import numpy as np
	import matplotlib.pyplot as plt
	from matplotlib.patches import Ellipse, Rectangle
	from matplotlib.transforms import Affine2D
	from scipy.stats import t, f
	from scipy.linalg import sqrtm
	import statsmodels.api as sm

el-hult / empbern.py

Last active April 5, 2022 08:17

Comparison between Empirical Bernstein, Hoeffding and CLT

	"""
	Comparison of empirical bernstein measures

	See https://el-hult.github.io/2022/03/18/empirical-bernstein-bounds.html

	"""
	import numpy as np
	import matplotlib.pyplot as plt
	import scipy.stats as sps
	import pandas as pd

el-hult / LazyRandom.hs

Created March 10, 2022 15:24

Example on how to generate a lazy stream of random numbers in haskell using System.IO.Unsafe

	import System.IO.Unsafe ( unsafeInterleaveIO )
	import System.Random ( randomRIO )
	import Text.Printf ( printf )

	-- \| An infinite list of random die rolls, lazily generated
	--
	-- This is dangerous since calling this function steps the global random generator
	-- not when running the action, but when accessing its result.
	diceRolls :: IO [Int]
	diceRolls = do

el-hult / quadplay.py

Created February 26, 2022 15:27

A code sample on how to create a interactive visualization in matplotlib with the matplotlib widgets. In this case, it shows how a quadratic with unknown location can be upper bounded.

	# %% set up the plot
	import matplotlib.pyplot as plt
	import numpy as np
	from matplotlib.widgets import Slider

	LOC_MAX = 2


	# The parametrized function to be plotted
	def f1(t, amplitude, location):

Ludvig Hult el-hult