anhqle

class Solution:
    def numIslands(self, grid: List[List[str]]) -> int:
        
        num_island = 0
        queue = deque()
        for x in range(len(grid)):
            for y in range(len(grid[0])):
                if grid[x][y] == "1":
                    queue.append((x, y))
                    while queue:

anhqle

# start container with correct port and bind mount
docker run -d -it -p 8788:8787 -v ~/two_sided_matching_empirical:/home/rstudio/two_sided_matching_empirical rocker/rstudio
# 8788 is the host port, to be access from browser

docker container ps # list containers

# Log into docker terminal as root (to install Miniconda)
docker exec -ti -u root container_name bash
apt update && apt install bzip2 # always update first so that the image has package cache
wget https://repo.continuum.io/miniconda/Miniconda3-latest-Linux-x86_64.sh

anhqle

#include <Rcpp.h>
using namespace Rcpp;

// [[Rcpp::export]]
NumericVector f_MHC(NumericVector y, double delta2, double S) {
  double s2 = 1; // known data variance
  double mu = 5; // prior mean
  double t2 = 10; // prior variance

  // Initialize storage

anhqle

expr() -- creates an expression expr(y <- 1)

"Parsing" = convert string to AST parse_expr("y <- 1") parse_exprs("y <- 1; x <- 3")

"Deparsing" = convert AST to string

anhqle

# Conditional logit model where the covariates only vary across choice, NOT across respondent
ww2 <- matrix(c(18, 40.9, 33.5, 0.256, 0.065, 0.123), nrow = 3) # Covariates, taken from real data
true_alpha <- c(0.07, 1) # Assuming we know the true preference parameter
num_obs <- 5000 # Number of observations
num_choices <- 3 # Number of choices
S <- 1000 # Number of simulations
res <- matrix(NA, nrow = S, ncol = length(true_alpha)) # Storage for simulation result
for (s in 1:S) {
  # Calculate the utility of respondent i for choice j
  Uij <- matrix(NA, nrow = num_obs, ncol = num_choices)

anhqle

# Simulations to understand the standard error of conditional logit
# In this particular case, my covariates are alternative specific,
# and do NOT vary across respondents
# (in Cameron and Triverdi, even alternative-specific covariates vary
# across respondents)

negloglik <- function(beta) {
  xb <- xx %*% beta
  - sum(y * (xb - log(1 + exp(xb))) + (1 - y) * (-log(1 + exp(xb))))
}

anhqle

If you have full lane control: (can't do this anymore with patch 7.09)

• Pull side camp into your creep
• Walk ahead and pull enemy creep into the camp too
• Stand outside to block the offlaner from contesting the camp

Always start attacking and denying as much as possible so that the two creep waves reset every time.

Deny your ranged creep sooner let the enemy creeps stay alive longer. This gives you more time to harass the offlaner without losing CS. (BSJ's concept of obligations)

The impact of tower on farming pattern

anhqle

git reset

https://git-scm.com/blog/2011/07/11/reset.html

anhqle

# Unix tool
brew install wget # needed downstream

# R and RStudio
brew tap homebrew/science
brew install --with-x11 r
brew cask install --appdir=/Applications rstudio

# Miniconda
wget https://repo.continuum.io/miniconda/Miniconda3-latest-MacOSX-x86_64.sh

anhqle

#!/usr/bin/env python
# coding=utf-8
from __future__ import print_function

import functools
import json
import os
import re
import sys