Convergence
 

Dear Experts,

I need your help.

My target I could not get my result converged with the single precision (maximum 10^-5). However, the solution converged at each time-step, using the double-precision.

Could anyone please explain a reason to use the double-precision? Can you please refer any material about the single and double-precision so I can study it myself.

Regards,
Nurzhan

Numerical round-off. The double precision numeric resolves the small differences between numbers more accurately so the single precision simulation gets lost in noise generated by numerical round-off, but the double precision simulation has sufficient precision in the numbers to have the numerical noise at a low enough level that convergence is still possible.

Thank you, Glenn!