commit 04070a934ef8318872144f6aa6331d56ed60236e
parent 8cc63dce4b0347d00ae1aeaec2be5d29e6d208d7
Author: NunoSempere <nuno.sempere@protonmail.com>
Date: Sat, 22 Jul 2023 19:21:20 +0200
give more expressive names to main functions
This bash function was helpful:
function replace(){
grep "$1" -rl .
grep "$1" -rl . | xargs sed -i "s/$1/$2/g";
}
Diffstat:
12 files changed, 38 insertions(+), 43 deletions(-)
diff --git a/README.md b/README.md
@@ -77,7 +77,7 @@ Behaviour on error can be toggled by the `EXIT_ON_ERROR` variable. This library
## To do list
-- [ ] Rename functions to something more self-explanatory, e.g,. sample_unit_normal.
+- [ ] Rename functions to something more self-explanatory, e.g,. sample_sample_sample_unit_normal.
- [ ] Have some more complicated & realistic example
- [ ] Add summarization functions, like mean, std, 90% ci (or all c.i.?)
- [ ] Publish online
@@ -90,7 +90,7 @@ Behaviour on error can be toggled by the `EXIT_ON_ERROR` variable. This library
- [x] Add example for many samples
- ~~[ ] Add a custom preprocessor to allow simple nested functions that don't rely on local scope?~~
- [x] Use gcc extension to define functions nested inside main.
-- [x] Chain various mixture functions
+- [x] Chain various sample_mixture functions
- [x] Add beta distribution
- See <https://stats.stackexchange.com/questions/502146/how-does-numpy-generate-samples-from-a-beta-distribution> for a faster method.
- ~~[-] Use OpenMP for acceleration~~
diff --git a/examples/01_one_sample/example b/examples/01_one_sample/example
Binary files differ.
diff --git a/examples/01_one_sample/example.c b/examples/01_one_sample/example.c
@@ -16,12 +16,12 @@ float sample_1(uint32_t* seed)
float sample_few(uint32_t* seed)
{
- return random_to(1, 3, seed);
+ return sample_to(1, 3, seed);
}
float sample_many(uint32_t* seed)
{
- return random_to(2, 10, seed);
+ return sample_to(2, 10, seed);
}
int main(){
@@ -37,7 +37,7 @@ int main(){
float weights[] = { 1 - p_c, p_c / 2, p_c / 4, p_c / 4 };
float (*samplers[])(uint32_t*) = { sample_0, sample_1, sample_few, sample_many };
- float result_one = mixture(samplers, weights, n_dists, seed);
+ float result_one = sample_mixture(samplers, weights, n_dists, seed);
printf("result_one: %f\n", result_one);
}
diff --git a/examples/02_many_samples/example b/examples/02_many_samples/example
Binary files differ.
diff --git a/examples/02_many_samples/example.c b/examples/02_many_samples/example.c
@@ -16,12 +16,12 @@ float sample_1(uint32_t* seed)
float sample_few(uint32_t* seed)
{
- return random_to(1, 3, seed);
+ return sample_to(1, 3, seed);
}
float sample_many(uint32_t* seed)
{
- return random_to(2, 10, seed);
+ return sample_to(2, 10, seed);
}
int main(){
@@ -40,7 +40,7 @@ int main(){
int n_samples = 1000000;
float* result_many = (float *) malloc(n_samples * sizeof(float));
for(int i=0; i<n_samples; i++){
- result_many[i] = mixture(samplers, weights, n_dists, seed);
+ result_many[i] = sample_mixture(samplers, weights, n_dists, seed);
}
printf("result_many: [");
diff --git a/examples/03_gcc_nested_function/example b/examples/03_gcc_nested_function/example
Binary files differ.
diff --git a/examples/03_gcc_nested_function/example.c b/examples/03_gcc_nested_function/example.c
@@ -16,8 +16,8 @@ int main(){
float sample_0(uint32_t* seed){ return 0; }
float sample_1(uint32_t* seed) { return 1; }
- float sample_few(uint32_t* seed){ return random_to(1, 3, seed); }
- float sample_many(uint32_t* seed){ return random_to(2, 10, seed); }
+ float sample_few(uint32_t* seed){ return sample_to(1, 3, seed); }
+ float sample_many(uint32_t* seed){ return sample_to(2, 10, seed); }
float (*samplers[])(uint32_t*) = { sample_0, sample_1, sample_few, sample_many };
float weights[] = { 1 - p_c, p_c / 2, p_c / 4, p_c / 4 };
@@ -25,7 +25,7 @@ int main(){
int n_samples = 1000000;
float* result_many = (float *) malloc(n_samples * sizeof(float));
for(int i=0; i<n_samples; i++){
- result_many[i] = mixture(samplers, weights, n_dists, seed);
+ result_many[i] = sample_mixture(samplers, weights, n_dists, seed);
}
printf("result_many: [");
diff --git a/examples/04_sample_from_cdf_simple/example b/examples/04_sample_from_cdf_simple/example
Binary files differ.
diff --git a/examples/04_sample_from_cdf_simple/example.c b/examples/04_sample_from_cdf_simple/example.c
@@ -84,12 +84,12 @@ int main()
test_and_time_sampler_float("cdf_normal_0_1", cdf_normal_0_1, seed);
// Get some normal samples using a previous approach
- printf("\nGetting some samples from unit_normal\n");
+ printf("\nGetting some samples from sample_unit_normal\n");
clock_t begin_2 = clock();
for (int i = 0; i < NUM_SAMPLES; i++) {
- float normal_sample = unit_normal(seed);
+ float normal_sample = sample_unit_normal(seed);
// printf("%f\n", normal_sample);
}
diff --git a/examples/05_sample_from_cdf_beta/example b/examples/05_sample_from_cdf_beta/example
Binary files differ.
diff --git a/squiggle.c b/squiggle.c
@@ -28,48 +28,44 @@ uint32_t xorshift32(uint32_t* seed)
}
// Distribution & sampling functions
-float rand_0_to_1(uint32_t* seed)
+float sample_unit_uniform(uint32_t* seed)
{
+ // samples uniform from [0,1] interval.
return ((float)xorshift32(seed)) / ((float)UINT32_MAX);
}
-float rand_float(float max, uint32_t* seed)
-{
- return rand_0_to_1(seed) * max;
-}
-
-float unit_normal(uint32_t* seed)
+float sample_unit_normal(uint32_t* seed)
{
// See: <https://en.wikipedia.org/wiki/Box%E2%80%93Muller_transform>
- float u1 = rand_0_to_1(seed);
- float u2 = rand_0_to_1(seed);
+ float u1 = sample_unit_uniform(seed);
+ float u2 = sample_unit_uniform(seed);
float z = sqrtf(-2.0 * log(u1)) * sin(2 * PI * u2);
return z;
}
-float random_uniform(float from, float to, uint32_t* seed)
+float sample_uniform(float from, float to, uint32_t* seed)
{
- return rand_0_to_1(seed) * (to - from) + from;
+ return sample_unit_uniform(seed) * (to - from) + from;
}
-float random_normal(float mean, float sigma, uint32_t* seed)
+float sample_normal(float mean, float sigma, uint32_t* seed)
{
- return (mean + sigma * unit_normal(seed));
+ return (mean + sigma * sample_unit_normal(seed));
}
-float random_lognormal(float logmean, float logsigma, uint32_t* seed)
+float sample_lognormal(float logmean, float logsigma, uint32_t* seed)
{
- return expf(random_normal(logmean, logsigma, seed));
+ return expf(sample_normal(logmean, logsigma, seed));
}
-float random_to(float low, float high, uint32_t* seed)
+float sample_to(float low, float high, uint32_t* seed)
{
const float NORMAL95CONFIDENCE = 1.6448536269514722;
float loglow = logf(low);
float loghigh = logf(high);
float logmean = (loglow + loghigh) / 2;
float logsigma = (loghigh - loglow) / (2.0 * NORMAL95CONFIDENCE);
- return random_lognormal(logmean, logsigma, seed);
+ return sample_lognormal(logmean, logsigma, seed);
}
// Array helpers
@@ -91,7 +87,7 @@ void array_cumsum(float* array_to_sum, float* array_cumsummed, int length)
}
// Mixture function
-float mixture(float (*samplers[])(uint32_t*), float* weights, int n_dists, uint32_t* seed)
+float sample_mixture(float (*samplers[])(uint32_t*), float* weights, int n_dists, uint32_t* seed)
{
// You can see a simpler version of this function in the git history
// or in C-02-better-algorithm-one-thread/
@@ -104,7 +100,7 @@ float mixture(float (*samplers[])(uint32_t*), float* weights, int n_dists, uint3
float result;
int result_set_flag = 0;
- float p = random_uniform(0, 1, seed);
+ float p = sample_uniform(0, 1, seed);
for (int k = 0; k < n_dists; k++) {
if (p < cumsummed_normalized_weights[k]) {
result = samplers[k](seed);
@@ -289,13 +285,13 @@ struct box inverse_cdf_box(struct box cdf_box(float), float p)
// Sampler based on inverse cdf and randomness function
struct box sampler_cdf_box(struct box cdf(float), uint32_t* seed)
{
- float p = rand_0_to_1(seed);
+ float p = sample_unit_uniform(seed);
struct box result = inverse_cdf_box(cdf, p);
return result;
}
struct box sampler_cdf_float(float cdf(float), uint32_t* seed)
{
- float p = rand_0_to_1(seed);
+ float p = sample_unit_uniform(seed);
struct box result = inverse_cdf_float(cdf, p);
return result;
}
@@ -303,7 +299,7 @@ struct box sampler_cdf_float(float cdf(float), uint32_t* seed)
/* Could also define other variations, e.g.,
float sampler_danger(struct box cdf(float), uint32_t* seed)
{
- float p = rand_0_to_1(seed);
+ float p = sample_unit_uniform(seed);
struct box result = inverse_cdf_box(cdf, p);
if(result.empty){
exit(1);
diff --git a/squiggle.h b/squiggle.h
@@ -8,22 +8,21 @@
uint32_t xorshift32(uint32_t* seed);
// Basic distribution sampling functions
-float rand_0_to_1(uint32_t* seed);
-float rand_float(float max, uint32_t* seed);
-float unit_normal(uint32_t* seed);
+float sample_unit_uniform(uint32_t* seed);
+float sample_unit_normal(uint32_t* seed);
// Composite distribution sampling functions
-float random_uniform(float from, float to, uint32_t* seed);
-float random_normal(float mean, float sigma, uint32_t* seed);
-float random_lognormal(float logmean, float logsigma, uint32_t* seed);
-float random_to(float low, float high, uint32_t* seed);
+float sample_uniform(float from, float to, uint32_t* seed);
+float sample_normal(float mean, float sigma, uint32_t* seed);
+float sample_lognormal(float logmean, float logsigma, uint32_t* seed);
+float sample_to(float low, float high, uint32_t* seed);
// Array helpers
float array_sum(float* array, int length);
void array_cumsum(float* array_to_sum, float* array_cumsummed, int length);
// Mixture function
-float mixture(float (*samplers[])(uint32_t*), float* weights, int n_dists, uint32_t* seed);
+float sample_mixture(float (*samplers[])(uint32_t*), float* weights, int n_dists, uint32_t* seed);
// Box
struct box {
