Boris Reuderink breuderink

## lagged.py
import numpy as np


def lag(x, n=0):
    observed = np.isnan(x) == False
    observed[0] = True
    offsets = np.flatnonzero(observed)
    return offsets[np.maximum(np.cumsum(observed) - 1 - n, 0)]


## Dockerfile
FROM jupyter/datascience-notebook
USER root
RUN apt-get update && apt-get install -yq graphviz
USER $NB_USER
RUN pip install lingam graphviz

## fx.c
#include "fx.h"
#include <stddef.h>
#include <stdint.h>
#include <math.h>
#include <assert.h>

void static inline wht_butterfly(float * const s, float * const d) {
	float temp = *s;
	*s += *d;
	*d = temp - *d;

## expander.c
// Generate expander graph. See https://mathoverflow.net/q/124714.
// cc expander.c -o expander && ./expander | \
// 	neato -Tpng -o expander.png -Goverlap=false -Gsplines=true

// From https://rosettacode.org/wiki/Modular_inverse#C
int mod_inv(int a, int b) {
	int b0 = b, t, q;
	int x0 = 0, x1 = 1;
	if (b == 1) return 1;
	while (a > 1) {

## idw.c
#include <assert.h>
#include <stdio.h>
#include <math.h>
#include <stdlib.h>
#include <time.h>

typedef struct {
	size_t size[2];
	size_t stride[2];
	float *elements;

## cartpolev1.tsv

          
            -0.044
            -0.15
            0.013
            0.28
            1
            1
            -0.047
            0.046
            0.018
            -0.013

            
              -0.047
              0.046
              0.018
              -0.013
              1
              1
              -0.046
              0.24
              0.018
              -0.3

            
              -0.046
              0.24
              0.018
              -0.3
              0
              1
              -0.041
              0.046
              0.012
              -0.0013

            
              -0.041
              0.046
              0.012
              -0.0013
              1
              1
              -0.04
              0.24
              0.012
              -0.29

            
              -0.04
              0.24
              0.012
              -0.29
              1
              1
              -0.035
              0.44
              0.0063
              -0.58

            
              -0.035
              0.44
              0.0063
              -0.58
              1
              1
              -0.026
              0.63
              -0.0053
              -0.87

            
              -0.026
              0.63
              -0.0053
              -0.87
              1
              1
              -0.014
              0.83
              -0.023
              -1.2

            
              -0.014
              0.83
              -0.023
              -1.2
              1
              1
              0.0027
              1
              -0.046
              -1.5

            
              0.0027
              1
              -0.046
              -1.5
              1
              1
              0.023
              1.2
              -0.075
              -1.8

            
              0.023
              1.2
              -0.075
              -1.8
              1
              1
              0.047
              1.4
              -0.11
              -2.1

## Makefile
libsum.so: sum.c
	cc -shared -o libsum.so sum.c

## tdc.c
// [1] Sutton, Richard S., et al. "Fast gradient-descent methods for
// temporal-difference learning with linear function approximation."
// Proceedings of the 26th Annual International Conference on Machine
// Learning. ACM, 2009.

#define TDC_NFEAT 4

typedef struct {
    float gamma;
    float theta[TDC_NFEAT], w[TDC_NFEAT];

## most_normal.py
from scipy import linalg, stats
import numpy as np
from functools import reduce

N = 16

def eval_scrambler(pat, reps=1):
    pat = np.atleast_1d(pat)
    D = np.diag(np.where(pat==1, -1, 1))
    H = linalg.hadamard(pat.size)

## halton.go
package main

import (
	"fmt"
)

func main() {
	// Implementation of baseline fixed-point algorithm in [1].
	//
	// [1] Kolář, Miroslav, and Seamus F. O'Shea. "Fast, portable, and
	import numpy as np


	def lag(x, n=0):
	observed = np.isnan(x) == False
	observed[0] = True
	offsets = np.flatnonzero(observed)
	return offsets[np.maximum(np.cumsum(observed) - 1 - n, 0)]
	FROM jupyter/datascience-notebook
	USER root
	RUN apt-get update && apt-get install -yq graphviz
	USER $NB_USER
	RUN pip install lingam graphviz
	#include "fx.h"
	#include <stddef.h>
	#include <stdint.h>
	#include <math.h>
	#include <assert.h>

	void static inline wht_butterfly(float * const s, float * const d) {
	float temp = *s;
	s += d;
	d = temp - d;
	// Generate expander graph. See https://mathoverflow.net/q/124714.
	// cc expander.c -o expander && ./expander \| \
	// neato -Tpng -o expander.png -Goverlap=false -Gsplines=true

	// From https://rosettacode.org/wiki/Modular_inverse#C
	int mod_inv(int a, int b) {
	int b0 = b, t, q;
	int x0 = 0, x1 = 1;
	if (b == 1) return 1;
	while (a > 1) {
	#include <assert.h>
	#include <stdio.h>
	#include <math.h>
	#include <stdlib.h>
	#include <time.h>

	typedef struct {
	size_t size[2];
	size_t stride[2];
	float *elements;
-0.044	-0.15	0.013	0.28	1	1	-0.047	0.046	0.018	-0.013
-0.047	0.046	0.018	-0.013	1	1	-0.046	0.24	0.018	-0.3
-0.046	0.24	0.018	-0.3	0	1	-0.041	0.046	0.012	-0.0013
-0.041	0.046	0.012	-0.0013	1	1	-0.04	0.24	0.012	-0.29
-0.04	0.24	0.012	-0.29	1	1	-0.035	0.44	0.0063	-0.58
-0.035	0.44	0.0063	-0.58	1	1	-0.026	0.63	-0.0053	-0.87
-0.026	0.63	-0.0053	-0.87	1	1	-0.014	0.83	-0.023	-1.2
-0.014	0.83	-0.023	-1.2	1	1	0.0027	1	-0.046	-1.5
0.0027	1	-0.046	-1.5	1	1	0.023	1.2	-0.075	-1.8
0.023	1.2	-0.075	-1.8	1	1	0.047	1.4	-0.11	-2.1
	// [1] Sutton, Richard S., et al. "Fast gradient-descent methods for
	// temporal-difference learning with linear function approximation."
	// Proceedings of the 26th Annual International Conference on Machine
	// Learning. ACM, 2009.

	#define TDC_NFEAT 4

	typedef struct {
	float gamma;
	float theta[TDC_NFEAT], w[TDC_NFEAT];
	from scipy import linalg, stats
	import numpy as np
	from functools import reduce

	N = 16

	def eval_scrambler(pat, reps=1):
	pat = np.atleast_1d(pat)
	D = np.diag(np.where(pat==1, -1, 1))
	H = linalg.hadamard(pat.size)
	package main

	import (
	"fmt"
	)

	func main() {
	// Implementation of baseline fixed-point algorithm in [1].
	//
	// [1] Kolář, Miroslav, and Seamus F. O'Shea. "Fast, portable, and