Functions:
# encoding: ACTG
def nuc2int_nochecking(b):
return (ord(b) >> 1) & 3, True
def nuc2int_if(b):
if b == 'a' or b == 'c' or b == 'g' or b == 't' \
or b == 'A' or b == 'C' or b == 'G' or b == 'T':
import torch | |
# cupy fails to import when there are no GPUs. This check means that the module can | |
# still be imported in that case, but of course the classes won't work. | |
if torch.cuda.is_available(): | |
import cupy | |
from torch.utils.data import DataLoader, Dataset | |
from torch.utils.dlpack import from_dlpack |
({ lib, newScope, stdenv, pkgs }: let | |
# nicer aliases | |
derive = stdenv.mkDerivation; | |
concat = builtins.concatStringsSep " "; | |
# vendored libuvc: don't build, just make sources available | |
libuvc-src = derive { | |
name = "libuvc-src"; | |
# using fetchgit instead fetchFromGitHub because |
import asyncio | |
import time | |
from unsync import unsync | |
@unsync | |
async def heartbeat(): | |
while True: | |
start = time.time() |
name: Binder | |
on: | |
pull_request: | |
types: [opened, reopened] | |
jobs: | |
Create-Binder-Badge: | |
runs-on: ubuntu-latest | |
steps: |
SAT and Satisfiability Modulo Theories (SMT) solvers have many important applications in PL, including verification, testing, type checking and inference, and program analysis – but they are often a mysterious black box to their users, even when those users are PL researchers with lots of solver experience! This talk will be partly a tutorial introduction to the inner workings of SAT and SMT solvers, and partly an extended analogy to navigating life as a researcher: making decisions when you have only incomplete information to go on, learning from decisions that turned out to be bad, and determining when to give up and when to try again. I’ll also highlight a variety of papers in this year’s POPL program that make use of SAT and SMT solving, and discuss why I think it’s worthwhile to learn about solver internals.
- Introduction
- Self-intro
#!/bin/bash | |
# This script will install AMDGPU-PRO OpenCL and Vulkan support. | |
# | |
# For Ubuntu and it's flavor, just install the package using this command | |
# in extracted driver directory instread. | |
# | |
# ./amdgpu-pro-install --opencl=legacy,pal --headless --no-dkms | |
# | |
# For Arch Linux or Manjaro, use the opencl-amd or rocm-opencl-runtime on AUR instread. |
import dask | |
import dask.array as da | |
import dask.dataframe as dd | |
import sparse | |
@dask.delayed(pure=True) | |
def corr_on_chunked(chunk1, chunk2, corr_thresh=0.9): | |
return sparse.COO.from_numpy((np.dot(chunk1, chunk2.T) > corr_thresh)) | |
def chunked_corr_sparse_dask(data, chunksize=5000, corr_thresh=0.9): |
The official installation guide (https://wiki.archlinux.org/index.php/Installation_Guide) contains a more verbose description.
- Image from https://www.archlinux.org/