add some thoughts to scratchpad and readme

scratchpad/scratchpad.c

@@ -21,7 +21,7 @@
             return error;                                                               \
         }                                                                               \
     } while (0)
-#define NUM_SAMPLES 10
+#define NUM_SAMPLES 1000000
 
 struct box {
     int empty;
@@ -399,7 +399,7 @@ void test_and_time_sampler_float(char* cdf_name, float cdf_float(float), uint32_
         if (sample.empty) {
             printf("Error in sampler function for %s", cdf_name);
         } else {
-            printf("%f\n", sample.content);
+            // printf("%f\n", sample.content);
         }
     }
     clock_t end = clock();
@@ -416,7 +416,7 @@ void test_and_time_sampler_box(char* cdf_name, struct box cdf_box(float), uint32
         if (sample.empty) {
             printf("Error in sampler function for %s", cdf_name);
         } else {
-            printf("%f\n", sample.content);
+            // printf("%f\n", sample.content);
         }
     }
     clock_t end = clock();
@@ -451,7 +451,7 @@ int main()
 
     for (int i = 0; i < NUM_SAMPLES; i++) {
         float normal_sample = sampler_normal_0_1(seed);
-        printf("%f\n", normal_sample);
+        // printf("%f\n", normal_sample);
     }
 
     clock_t end_2 = clock();
@@ -460,6 +460,11 @@ int main()
 
     // Test box sampler
     test_and_time_sampler_box("cdf_beta", cdf_beta, seed);
+		// Ok, this is slower than python!!
+		// Partly this is because I am using a more general algorithm, 
+		//   which applies to any cdf
+		// But I am also using really anal convergence conditions.
+		// This could be optimized.
 
     free(seed);
     return 0;

to-do.md

@@ -6,6 +6,11 @@
 - [x] Chain various mixture functions
 - [ ] Have some more complicated & realistic example
 - [ ] Add summarization functions, like mean, std, 90% ci (or all c.i.?)
-- [ ] Add beta distribution
+- [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
-- [ ] Add function to get sample when given a cdf
+- [x] Add function to get sample when given a cdf
+- [ ] Don't have a single header file.
+- [ ] Structure project a bit better
+- [ ] Add README
+- [ ] Publish