Skip to content

fix: Fix the problems with float precisions in triangulation process #112

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
Czaki merged 2 commits into from
Sep 27, 2026
Merged

fix: Fix the problems with float precisions in triangulation process #112

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
2 commits
Select commit Hold shift + click to select a range
81f9b05
fix: Fix the problems with float precisions in triangulation process
Czaki Sep 25, 2026
d1573e3
[pre-commit.ci] auto fixes from pre-commit.com hooks
pre-commit-ci[bot] Sep 25, 2026
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
35 changes: 19 additions & 16 deletions src/PartSegCore_compiled_backend/triangulation/intersection.hpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -133,8 +133,12 @@ enum Orientation {
*/
inline Orientation _orientation(const point::Point& p, const point::Point& q,
const point::Point& r) {
double val1 = ((q.y - p.y) * (r.x - q.x));
double val2 = ((r.y - q.y) * (q.x - p.x));
// Promote to double before multiplying: the float products lose precision
// and can flip the result for nearly collinear points.
double val1 =
(static_cast<double>(q.y) - p.y) * (static_cast<double>(r.x) - q.x);
double val2 =
(static_cast<double>(r.y) - q.y) * (static_cast<double>(q.x) - p.x);
// This commented code if for debugging purposes of differences between macOS
// and linux double val = ((q.y - p.y) * (r.x - q.x)) - ((r.y - q.y) * (q.x -
// p.x)); if (val!= 0 && val1 == val2) {
Expand Down Expand Up @@ -337,15 +341,16 @@ inline std::unordered_set<OrderedPair> _find_intersections(
*/
inline std::vector<point::Point> _find_intersection(const point::Segment& s1,
const point::Segment& s2) {
// To reduce problems with float precision, the same pair of segments
// should always be calculated in the same order.
if (s2 < s1) return _find_intersection(s2, s1);
// ReSharper disable CppJoinDeclarationAndAssignment
point::Point::coordinate_t a1, b1, a2, b2, det, x, y, t;
// point::Point::coordinate_t c1, c2, u;
a1 = s1.top.y - s1.bottom.y;
b1 = s1.bottom.x - s1.top.x;
// c1 = a1 * s1.bottom.x + b1 * s1.bottom.y;
a2 = s2.top.y - s2.bottom.y;
b2 = s2.bottom.x - s2.top.x;
// c2 = a2 * s2.bottom.x + b2 * s2.bottom.y;
double a1, b1, a2, b2, det, t;
point::Point::coordinate_t x, y;
a1 = static_cast<double>(s1.top.y) - s1.bottom.y;
b1 = static_cast<double>(s1.bottom.x) - s1.top.x;
a2 = static_cast<double>(s2.top.y) - s2.bottom.y;
b2 = static_cast<double>(s2.bottom.x) - s2.top.x;
det = a1 * b2 - a2 * b1;
if (det == 0) {
// collinear case
Expand All @@ -356,15 +361,13 @@ inline std::vector<point::Point> _find_intersection(const point::Segment& s1,
if (s2.point_on_line(s1.top)) res.push_back(s1.top);
return res;
}
t = ((s2.top.x - s1.top.x) * (s2.bottom.y - s2.top.y) -
(s2.top.y - s1.top.y) * (s2.bottom.x - s2.top.x)) /
t = ((static_cast<double>(s2.top.x) - s1.top.x) * -a2 -
(static_cast<double>(s2.top.y) - s1.top.y) * b2) /
det;
// u = ((s2.top.x - s1.top.x) * (s1.bottom.y - s1.top.y) -
// (s2.top.y - s1.top.y) * (s2.bottom.x - s2.top.x)) / det;
if (t < 0) return {s1.top};
if (t > 1) return {s1.bottom};
x = s1.top.x + t * b1;
y = s1.top.y + t * (-a1);
x = s1.top.x + static_cast<point::Point::coordinate_t>(t * b1);
y = s1.top.y + static_cast<point::Point::coordinate_t>(t * (-a1));
return {{x, y}};
}
} // namespace partsegcore::intersection
Expand Down
Loading
Loading