MilesCranmer

from IPython.display import set_matplotlib_formats
set_matplotlib_formats('pdf', 'svg')
import matplotlib
import seaborn as sns
  
sns.set(font_scale=1.3)
sns.set_style("darkgrid", {"axes.facecolor": ".95"})

import matplotlib.pyplot as plt

MilesCranmer

#!/bin/bash
##################################
#
# THE ARCHIVE TRACKER
#
# REF: https://gist.github.com/markwk/c85a8a72bc8c03d0f510262bb5219a34/
#
# INTRODUCTION: 
# Daily script to navigate to a directory of plain text files,
# add files to git repo, calculate key stats, store stats to csv

MilesCranmer

set nocompatible              " be iMproved, required
let g:python3_host_prog = '/usr/local/opt/python@3.8/bin/python3.8'

if empty(glob('~/.vim/autoload/plug.vim'))
  silent !curl -fLo ~/.vim/autoload/plug.vim --create-dirs
    \ https://raw.githubusercontent.com/junegunn/vim-plug/master/plug.vim
  autocmd VimEnter * PlugInstall --sync | source $MYVIMRC
endif

call plug#begin('~/.vim/plugged')

MilesCranmer

%Make sure to have \usepackage{tikz}


%https://tex.stackexchange.com/a/45815/140440 - for grid
%https://tex.stackexchange.com/a/381175/140440 - for alignment in equation

% This function draws a matrix.
\newcommand{\mat}[2]{% cols, rows
\vcenter{\hbox{ %Vertical alignment
\begin{tikzpicture}[scale=0.3, align=center]

MilesCranmer

import numpy as np
from mpmath import mp, mpmathify
from pysr import *

#Set precision to 200 decimal places:
mp.dps = 200

x = np.linspace(-10, -5, num=300)

#High precision calculation:

MilesCranmer

# Copy this into your code. Call with, e.g., einop(x, 'i j -> j', reduction='mean')
import functools
import einops as _einops
from einops.parsing import ParsedExpression

@functools.lru_cache(256)
def _match_einop(pattern: str, reduction=None, **axes_lengths: int):
    """Find the corresponding operation matching the pattern"""
    left, rght = pattern.split('->')
    left = ParsedExpression(left)

MilesCranmer

using SymbolicUtils

mutable struct Node
    #Holds operators, variables, constants in a tree
    degree::Integer #0 for constant/variable, 1 for cos/sin, 2 for +/* etc.
    val::Union{Float32, Integer, Nothing} #Either const value, or enumerates variable
    constant::Bool #false if variable
    op::Integer #enumerates operator (separately for degree=1,2)
    l::Union{Node, Nothing}
    r::Union{Node, Nothing}

MilesCranmer

Before training:

last = model.state_dict()

Inside training loop, after computing loss:

if torch.isnan(loss).sum().item():
    model.load_state_dict(last)
else:

MilesCranmer

def acos2(num, denom, disamb):
    cosine = num/denom
    
    return torch.where((cosine > -1) & (cosine < 1.),
        torch.acos(cosine) * torch.where(disamb < 0.0, -1, 1),
        torch.where(cosine <= -1, np.pi, 0.0)
    )

def coord_transform(x):
    # Assumes in CoM frame

MilesCranmer

def soft_clip(x, lo, hi, pct=0.1):
    range = hi - lo
    frac = (x - lo) / range

    normalization = F.softplus(torch.ones_like(x))

    for _ in ['lo', 'hi']:
        frac = torch.where(frac > pct,
                           frac,
                           pct * F.softplus(frac / pct) / normalization