Timeline for LinearSolve and badly conditioned matrix
Current License: CC BY-SA 4.0
8 events
| when toggle format | what | by | license | comment | |
|---|---|---|---|---|---|
| Aug 25, 2024 at 16:21 | comment | added | Henrik Schumacher | Also, it is not uncommon that the condition number increases with the size of the matrix. So this explain why Mathematica complained only about the larger problems. | |
| Aug 25, 2024 at 13:07 | comment | added | Henrik Schumacher | Ill-conditioning has not effect if you use exact arithmetic. (But for larger problems, exact arithmetic will be too expensive.) Ill-conditioning does not mean that the matrix equation is not solvable. It just means that floating-point errors are likely to accumulate during the computations. This is why Mathematica warns you: The solution might have an error that is acceptable to your application or not. You have to check the quality of the solution yourself. | |
| Aug 24, 2024 at 23:26 | comment | added | Pedro Henrique | @DanielLichtblau depends. When I utilize small matrixes the two results are identical. When I increase the size of the matrix it does not solve because “badly conditioning”. However, mathematica still solves the system of equation. I only got problem with matrix equation…. | |
| Aug 24, 2024 at 22:00 | comment | added | Daniel Lichtblau | Are the two results different? If so, that also would indicate bad conditioning. | |
| Aug 24, 2024 at 21:23 | comment | added | Pedro Henrique | @MichaelE2 well I’m not getting any closer to the correct output :/ | |
| Aug 24, 2024 at 19:55 | comment | added | Michael E2 | @PedroHenrique "I do not understand why the matrix version cannot" -- What makes you think it did not? | |
| Aug 24, 2024 at 17:33 | comment | added | Pedro Henrique | I do not know if that helps, but if I write the system of equations that results from m.x==b, extract the coefficients using CoefficientArrays[ ], and then use LinearSolve[ ], I get the correct answer without error messages. However, the method of creating the system of equations is quite slow, and I am trying to avoid it. Nevertheless, the system can be solved, so I do not understand why the matrix version cannot. | |
| Aug 24, 2024 at 15:44 | history | answered | Henrik Schumacher | CC BY-SA 4.0 |