bug in IRootFunctions double.Hypot

[c-ohle]

Description

The double.Hypot function returns negative results.

public static double Hypot(double x, double y)
It is in the line result = x + y; where the abs values should be used.

Just as note: I've found that the algorithm used is less precise and slower than a common implementation.

Reproduction Steps

  x = double.Hypot(1, +1e20); // 1E+20 ok
  x = double.Hypot(1, -1e10); // 1E+10 ok
  x = double.Hypot(1, -1e20); // -1E+20 wrong

Expected behavior

  x = double.Hypot(1, +1e20); // 1E+20 ok
  x = double.Hypot(1, -1e10); // 1E+10 ok
  x = double.Hypot(1, -1e20); // 1E+20 

Actual behavior

  x = double.Hypot(1, +1e20); // 1E+20 ok
  x = double.Hypot(1, -1e10); // 1E+10 ok
  x = double.Hypot(1, -1e20); // -1E+20 wrong

Regression?

No response

Known Workarounds

No response

Configuration

No response

Other information

No response

[ghost]

Tagging subscribers to this area: @dotnet/area-system-numerics
See info in area-owners.md if you want to be subscribed.

Issue Details

---

Description

The double.Hypot functions return negative results.

public static double Hypot(double x, double y)
It is in the line result = x + y; where the abs values should be used.

Just as a note: I've found that the algorithm used is less precise and slower than the simple calculation.

Reproduction Steps

  x = double.Hypot(1, +1e20); // 1E+20 ok
  x = double.Hypot(1, -1e10); // 1E+10 ok
  x = double.Hypot(1, -1e20); // -1E+20 wrong

Expected behavior

  x = double.Hypot(1, +1e20); // 1E+20 ok
  x = double.Hypot(1, -1e10); // 1E+10 ok
  x = double.Hypot(1, -1e20); // 1E+20 

Actual behavior

  x = double.Hypot(1, +1e20); // 1E+20 ok
  x = double.Hypot(1, -1e10); // 1E+10 ok
  x = double.Hypot(1, -1e20); // -1E+20 wrong

Regression?

No response

Known Workarounds

No response

Configuration

No response

Other information

No response

Author:	c-ohle
Assignees:	-
Labels:	area-System.Numerics, untriaged
Milestone:	-

[jeffhandley]

This will be considered for backporting to 7.0 since it's a bug in new 7.0 functionality.

[c-ohle]

Is it ok if I report the other NET7 numeric bugs I've found as a collection in a case list with no detailed explanation for each case?
It would save a lot of time.

[tannergooding]

We need reproduction steps for each of the cases that have bugs so that regression tests can be added.

Not providing them will cause it to take more time overall to get the fix in and risk the fix not making it into .NET 7 RTM.

[c-ohle]

@tannergooding ok, next week.