Tolerance fix for interval solvers #77
Conversation
Codecov Report
@@ Coverage Diff @@
## main #77 +/- ##
========================================
+ Coverage 0.00% 3.14% +3.14%
========================================
Files 20 20
Lines 990 1017 +27
========================================
+ Hits 0 32 +32
+ Misses 990 985 -5
... and 3 files with indirect coverage changes 📣 We’re building smart automated test selection to slash your CI/CD build times. Learn more |
| @@ -31,6 +31,11 @@ function SciMLBase.solve(prob::IntervalNonlinearProblem, alg::Bisection, args... | |||
| retcode = ReturnCode.ExactSolutionLeft, left = left, | |||
| right = right) | |||
| end | |||
| if iszero(fr) | |||
There was a problem hiding this comment.
Also, i dont see RightExactSolution condition in any of these bracketing solvers, was that intentional? Otherwise, it's gonna calculate an inexact solution even if the right value is the exact zero?
There was a problem hiding this comment.
If there is a condition by which right solution is exact it should do that retcode yes.
changed default tol to machine epsilon
yash2798
changed the title
|
@ChrisRackauckas this can be reviewed now |
|
Why isn't ITP in this list? |
sorry, just added it. i was under the impression that it was already there as we had an epsilon in the algorithm, but the early exit condition wasnt there. |
fixes #57