Skip to content

Instantly share code, notes, and snippets.

View andyljones's full-sized avatar
🤖
👩👨👩👨

Andy Jones andyljones

🤖
👩👨👩👨
177 followers · 14 following
View GitHub Profile
@andyljones
andyljones / Vagrantfile
Last active August 29, 2015 13:56
Vagrantfile and provisioning script for a Vagrant instance for Octopress development.
VAGRANTFILE_API_VERSION = "2"
Vagrant.configure(VAGRANTFILE_API_VERSION) do |config|
config.vm.box = "debian7"
config.vm.box_url = "https://dl.dropboxusercontent.com/s/xymcvez85i29lym/vagrant-debian-wheezy64.box"
config.vm.network :forwarded_port, host: 4000, guest: 4000
config.vm.provision :shell, :path => "bootstrap.sh"
config.ssh.forward_agent = true
@andyljones
andyljones / compensation_comparison.py
Last active September 20, 2015 15:23
StackOverflow Compensation Survey Plotting
# -*- coding: utf-8 -*-
"""
To use this, drop the file
'Full Results - Stack Overflow Developer Survey - 2015.csv'
from
https://drive.google.com/file/d/0Bzd_CzYvUxE5U1NSWnA2SFVKX00/view
@andyljones
andyljones / coroutine_break_example.py
Last active August 29, 2015 14:23
def work():
print 0; yield
print 1; yield
print 2; yield
worker = work()
for i in range(2+1):
next(worker)
if i == 1: break
@andyljones
andyljones / developer_compensations.py
Created August 1, 2015 10:22
# -*- coding: utf-8 -*-
"""
To use this, drop the file
'Full Results - Stack Overflow Developer Survey - 2015.csv'
from
https://drive.google.com/file/d/0Bzd_CzYvUxE5U1NSWnA2SFVKX00/view
@andyljones
andyljones / nk_border_checkpoint.py
Last active July 26, 2016 17:40
import scipy as sp
from scipy.special import gammaln
def log_marginal(p, n, alpha=2):
"""Log-marginal probability of `p` positive trials and `n` negative trials from a
beta-binomial model with prior strength `alpha`. See
http://www.cs.ubc.ca/~murphyk/Teaching/CS340-Fall06/reading/bernoulli.pdf
for details.
@andyljones
andyljones / hypernormalisation.rst
Last active May 23, 2020 06:49
hypernormalisation

EXPLOSION

We live in a strange time.

Extraordinary events keep happening that undermine the stability of our world.

Suicide bombs, waves of refugees,

Donald Trump, Vladimir Putin, even Brexit.

@andyljones
andyljones / bankruptcy.py
Last active February 5, 2017 16:46
Simple script for estimating bankruptcy probabilities during retirement
#!/usr/bin/env python3
# -*- coding: utf-8 -*-
"""
Created on Sun Feb 5 11:01:52 2017
@author: andyjones
"""
import scipy as sp
import matplotlib.pyplot as plt
@andyljones
andyljones / temperature.py
Last active August 26, 2018 18:23
Demonstrates the importance of perceptually uniform colormaps using temperature in Central England over the past 200 years
import requests
import pandas as pd
from io import BytesIO
import matplotlib as mpl
import matplotlib.pyplot as plt
url = 'https://www.metoffice.gov.uk/hadobs/hadcet/cetdl1772on.dat'
raw = pd.read_csv(BytesIO(requests.get(url).content), sep='\s+', header=None)
raw.columns = ['year', 'day'] + list(range(1, 13))
@andyljones
andyljones / standalone.py
Last active May 21, 2019 18:43
"""
This is a standalone script for demonstrating some memory leaks that're troubling me. It's a torn-down version of
the project I'm currently working on.
To run this, you'll need panda3d, pandas and tqdm. You should be able to install these with
```
pip install panda3d pandas tqdm
```
You'll **also need to enable memory tracking**. Do that by setting `track-memory-usage 1` in `panda3d.__file__`'s
`etc/Config.prc` file. Setting it anywhere else is too late! (It's a unique setting in that way - thanks rdb!)
@andyljones
andyljones / mpshang.py
Last active October 28, 2022 09:28
Demonstration of the MPS hang bug
"""This script should be run on a machine with at least 2 GPUs and an MPS server running. You can launch an MPS daemon with
```
nvidia-cuda-mps-control -d
```
The script first uses `test_cuda` to verify a CUDA context can be created on each GPU. It then spawns two workers; a
'good' worker and a 'bad' worker. The workers collaborate through Pytorch's DataDistributedParallel module to calculate
the gradient for a trivial computation. The 'good' worker carries out both the forward and backward pass, while the
bad worker carries out the forward pass and then exits. This seems to lock up the MPS server, and any subsequent
attempts to create CUDA contexts fail by hanging eternally.