j20232

For vacation 🏝

j20232

#!/bin/sh

/usr/local/bin/g++-9 $1 -o a.out
contest=$(cut -d '_' -f 1 <<< $1)
index=$(cut -d '_' -f 2 <<< $1)
type=$(cut -d '.' -f 1  <<< $(cut -d '_' -f 3 <<< $1))
url="https://${contest}${index}.contest.atcoder.jp/tasks/${contest}${index}_${type}"
rm -rf test/*
oj dl $url
oj test

j20232

import numpy as np
import cv2
from sklearn import preprocessing

def convert_to_dot(img, color_num=16, dot_size=10):
    lab = cv2.cvtColor(img, cv2.COLOR_RGB2LAB) / 255.0
    flat_lab = lab.reshape((img.shape[0] * img.shape[1], 3))
    Z = preprocessing.StandardScaler().fit_transform(flat_lab).astype(np.float32)
    criteria = (cv2.TERM_CRITERIA_EPS + cv2.TERM_CRITERIA_MAX_ITER, 1, 10.0)
    _,label,_= cv2.kmeans(Z, color_num, None, criteria, 20, cv2.KMEANS_PP_CENTERS)

j20232

import numpy as np


def first_order_cp(X, rank, la=0.01, eta=0.01, eps=1e-5):
    """L2-regularized cp decomposition with first order alternating gradient descent
    Args:
        X (np.ndarray): IxJxK matrix
        rank (int): rank for decomposed matrices ([I, J] -> [I, rank], [rank, J])
        la (float, optional): coefficient for regularization. Defaults to 0.01.
        eta (float, optional): learning rate. Defaults to 0.01.

j20232

import numpy as np
import scipy
import trimesh
import polyscope as ps
from pathlib import Path


def double_area(geom):
    i = geom.faces[:, 0]  # [num_faces]
    j = geom.faces[:, 1]  # [num_faces]

j20232

import numpy as np
import scipy
import trimesh
import polyscope as ps
from pathlib import Path
from tqdm import tqdm


def double_area(geom):
    i = geom.faces[:, 0]  # [num_faces]

j20232

import os
import random
from tqdm import tqdm
import numpy as np
import torch


def seed_everything(seed=1116):
    random.seed(seed)
    os.environ['PYTHONHASHSEED'] = str(seed)

j20232

import numpy as np
import matplotlib.pyplot as plt
import torch

def func(x):
    return 1e-4 * ((x - 6) ** 3) + 1e-4 * ((x - 5) ** 4) + 1e-2 * ((np.sin(x * 0.1) - 3) ** 2)

def torch_func(x):
    return 1e-4 * ((x - 6) ** 3) + 1e-4 * ((x - 5) ** 4) + 1e-2 * ((torch.sin(x * 0.1) - 3) ** 2)
class FiniteDiff(torch.autograd.Function):

j20232

import numpy as np
import argparse


def f(x, func_id):
    if func_id == 0:
        return x[0] * x[1] + np.log(x[0])
    elif func_id == 1:
        return np.sin(x[0]) + np.cos(x[1])
    else: