I found that in the source file which addressed at ray-tracing-gems/Ch_11_Automatic_Handling_of_Materials_in_Nested_Volumes/push.cpp may have a mistake. The index variable used at line 14 as prev_same is an index of the stack, not an index of the scene materials. So, I think the variable prev_same should be change to stack[prev_same].material_idx.Just like the following codes:

line 14 if (material[material_idx] == material[prev_same]) { => if (material[material_idx] == material[stack[prev_same].material_idx]) {

It was also used at line 26 but in right form:
line 26 if ((material[stack[idx].material_idx] != material[material_idx]) &&

In section 16.4.2.3 Hierarchical Transformation (p233), the first code sample is missing a curly braces between line 8 and 9 to close off the else statement:

 1 node = root;
 2 while (!node.isLeaf) {
 3     if (u < node.probLeft) {
 4         u /= node.probLeft;
 5         node = node.left;
 6     } else {
 7         u /= (1.0 - node.probLeft);
 8         node = node.right;
 9 }
10 // Ok. We have found a leaf with the correct probability!

Should be:

 1 node = root;
 2 while (!node.isLeaf) {
 3     if (u < node.probLeft) {
 4         u /= node.probLeft;
 5         node = node.left;
 6     } else {
 7         u /= (1.0 - node.probLeft);
 8         node = node.right;
 9     }                                 <<<<<<<<<<<<<<<<<<<<< Missing this
10 }
11 // Ok. We have found a leaf with the correct probability!

The introduction of Listing 27-2:

Listing 27-2.This closest-hit shader code snippet skips shading of transparent surfaces when the incident ray has crossed through a user-defined maximum number of transparent surfaces, proceeding instead by shooting a transmission ray and continuing as though there had been no ray/surface intersection

is same as that of Listing 27-4:

Listing 27-4. This closest-hit shader code snippet skips the shading of transparent surfaces when the incident ray has crossed through a user-defned maximum number of transparent surfaces, proceeding instead by shooting a transmission ray and continuing as though there had been no ray/surface intersection.

and I believe the former is wrong.
I didn't see this mistake in the errata and I have checked version 1.5, 4/29/2019, it's still there.

Greetings I am interested in the implementation of Ch. 14 (Ch_14_Ray-Guided_Volumetric_Water_Caustics_in_Single_Scattering_Media_with_DXR), but it seems that the source code is missing. I can only find dll, exe & media files.

There is a potential GL texture memory leak in the code. It generates a new texture object each frame (line 459-462).

Solution:
Two additions in the code should do it :)

1. Add the following code after line 431

kernel_params.iteration = 0;

//allocate texture once
glBindTexture(GL_TEXTURE_2D, display_tex);
glTexImage2D(GL_TEXTURE_2D, 0, GL_RGBA8, width, height, 0, GL_BGRA, GL_UNSIGNED_BYTE, NULL);
glTexParameteri(GL_TEXTURE_2D, GL_TEXTURE_MIN_FILTER, GL_LINEAR);
glTexParameteri(GL_TEXTURE_2D, GL_TEXTURE_MAG_FILTER, GL_LINEAR);
}

2. Replace glTexImage2D and parameter setting calls (lines 459-462) with the following glTexSubImage2D call,
   glTexSubImage2D(GL_TEXTURE_2D, 0, 0,0, width, height, GL_BGRA, GL_UNSIGNED_BYTE, NULL);

Hi, very minor thing but I'm looking at http://raytracinggems.com/unofficial_RayTracingGems_v1.7.pdf and the last line code for estimate_sample_variance() seems incorrect.

The code is:

1 float estimate_sample_variance(float samples[], int n) {
2     float sum = 0, sum_sq = 0;
3     for (int i = 0; i < n; ++i) {
4         sum += samples[i];
5         sum_sq += samples[i] * samples[i];
6     }
7     return sum_sq / (n * (n - 1))) -
8     sum * sum / ((n - 1) * n * n);
9 }

in lines 7 & 8 there is an additional ')' but the formula also appears wrong - it should be return sum_sq / (n - 1) - sum * sum / ((n - 1) * n); ( or return (sum_sq - sum * sum / n) / (n - 1); )? (or I might be missing something :) )