trappmartin

THIS Guide is all you need - (Vienna ed.)

Some general remarks, the list below is a collection of suggestions where to go/what to do when visiting Vienna. Consider this list to be incomplete!

Money: Make sure to bring some cash with you. Vienna is increasingly more cash-free but it is still far away from that.

Safety: Vienna is generally safe. However, there are regular cases of pickpocketing in the city center, so be aware. Some districts further outside the city can be unsafe at night (e.g., Favoriten and Floridsdorf).

Commute: If you commute for a couple of days, make sure to check "Wiener Linien" for their options on tickets. The cheapest option for a longer stay is the "8-days climate ticket" (8-Tage-Klimakarte) which is essentially a collection of 8 independent one-day tickets that you can validate as you like, e.g., share with others or reuse during a second visit.

trappmartin

import matplotlib.pyplot as plt
import numpy as np
from matplotlib.colors import ListedColormap

from sklearn.svm import SVC

classifier = SVC(gamma=2, C=1, random_state=42, probability=True)
dataset = make_moons(noise=0.3, random_state=0)

figure = plt.figure()

trappmartin

julia> function decToBin(n)
           if n > 0
               x = decToBin(n ÷ 2)
               return x === nothing ? [n % 2] : append!(x, n % 2)
           end
       end
decToBin (generic function with 1 method)

julia> decToBin(4)
3-element Vector{Int64}:

trappmartin

### A Pluto.jl notebook ###
# v0.19.9

using Markdown
using InteractiveUtils

# ╔═╡ 07034ef4-037e-11ee-30e6-37d72cee1635
using Plots, Distances, Statistics, Random, LaTeXStrings

# ╔═╡ dd352de2-0119-4562-b22c-39c2575bf721

trappmartin

# Collect all sum nodes.
snodes = filter(n -> isa(n, SumNode), values(spn))

# option 1

# Create an initial values.
q1 = mapreduce(n -> n.logweights[:], vcat, snodes)

# Helper function used by ForwardDiff.
function f1(θ)

trappmartin

using Turing
using Turing.RandomMeasures

# Implementation of infinite mixture model
@model function imm(y, α, ::Type{T}=Vector{Float64}) where {T}
    N = length(y)
    nk = tzeros(Int, N)
    z = tzeros(Int, N)

    for i in 1:N

trappmartin

using Turing

# This is a crappy implementation of Isabel's model.
# For simplicity, I assume all data to be positiv-real valued.
# Isabel's paper: http://proceedings.mlr.press/v70/valera17a/valera17a.pdf

@model function adst(y, ::Type{TV}=Vector{Float64}) where {TV}
  
  N,D = size(y)
  

trappmartin

using Turing

@model function ibp(y, α, kmax, ::Type{MV}=Vector{Float64}) where {MV}
   N = length(y)
   
   ks = tzeros(Int, N)
   ks[1] ~ Poisson(α)
   ks[1] = ks[1] <= kmax ? ks[1] : kmax
   
   z = tzeros(Int, N, kmax)

trappmartin

using Turing
using Bijectors
using LinearAlgebra

Bijectors.bijector(d::LKJ) = PDBijector()

@model function correlation(y)
    N,D = size(y)
    
    mu ~ filldist(truncated(Normal(0, 100), -10, 10), D)

trappmartin

using DrWatson
@quickactivate "DynamicPPL_NeurIPS"

using Turing
using LinearAlgebra
using Random: seed!
seed!(1)

@model testmodel(p, O) = begin
    x ~ Categorical(p)