kstoneriv3’s gists

kstoneriv3 / brownian_bridge_construction.py

Created February 1, 2024 14:37

Visualization of Brownian bridge construction of a Brownian motion


	"""Visualization of Brownian bridge construction of a Brownian motion."""

	import numpy as np
	from matplotlib import pyplot as plt


	T = 1
	TAB10_CMAP = plt.get_cmap("tab10")
	DEFAULT_COLOR = TAB10_CMAP(0)

kstoneriv3 / benchmark_batched_botorch.py

Created April 9, 2023 22:20

Benchmark script of batched sampling with botorch

	from concurrent.futures import ProcessPoolExecutor, wait
	import fire
	from matplotlib import pyplot as plt
	import numpy as np
	import optuna
	import os
	import time
	import torch

	N_WORKERS = 10

kstoneriv3 / halton.py

Last active January 25, 2022 15:01

`run_all.sh` runs the benchmark of `QMCSamplers` by kurobako. The performance of `QMCSampler` differs significantly when called from kurobako CUI vs when called using python script.

	from kurobako import solver
	from kurobako.solver.optuna import OptunaSolverFactory
	import optuna
	from optuna.samplers import QMCSampler


	def create_study(seed):
	sampler = QMCSampler(seed=seed, scramble=True, qmc_type="halton", warn_independent_sampling=False)
	return optuna.create_study(sampler=sampler)

kstoneriv3 / denoising.py

Last active April 17, 2021 14:09

(Incomplete) Reproduction of image denoising using kernel PCA in "Learning to Find Pre-Images"

	import numpy
	import scipy.ndimage
	import matplotlib.pyplot as plt

	from sklearn import datasets
	from sklearn import decomposition

	digits = datasets.load_digits()
	images = digits.images
	images = map(lambda x: scipy.ndimage.zoom(x, 2, order=1), images)

kstoneriv3 / RGS.py

Created March 9, 2021 08:13

A possibly incorrect implementation of RGS algorithm discussed in Owen, A. B. (1994). Controlling correlations in Latin hypercube samples. Journal of the American Statistical Association, 89(428), 1517-1522.

	#!/usr/bin/env python
	# coding: utf-8

	import numpy as np
	import matplotlib.pyplot as plt


	# centered case
	def LHS(n, d):
	samples = np.tile(np.arange(n, dtype=np.float64), (d, 1)).reshape(d, n)

kstoneriv3 / RGS.py

Last active March 9, 2021 08:42

A possibly incorrect implementation of RGS algorithm discussed in Owen, A. B. (1994). Controlling correlations in Latin hypercube samples. Journal of the American Statistical Association, 89(428), 1517-1522.

	#!/usr/bin/env python
	# coding: utf-8

	import numpy as np
	import matplotlib.pyplot as plt


	# centered case
	def LHS(n, d):
	samples = np.tile(np.arange(n, dtype=np.float64), (d, 1)).reshape(d, n)

kstoneriv3 / hpi_evaluatior_comparison_lightgbm.py

Last active September 21, 2020 18:52

benchmark code for `MutualInformationImportanceEvaluator`

	import lightgbm as lgb
	import numpy as np

	import pandas as pd
	import sklearn.datasets
	from sklearn.datasets import fetch_openml
	import sklearn.metrics
	from sklearn.model_selection import train_test_split

	import optuna

Kei Ishikawa kstoneriv3