A question about using Herbgrind to dectect floating-point expression's error

[hpcyhr]

Hello, our team specializes in floating-point program optimization and analysis. Herbgrind is an excellent job, and recently we have been trying to use it to detect errors in floating-point arithmetic expressions, but we don't know how to modify the source code. For example, we want to detect the maximum error of x1* x2/(x1 * x1 * x2 * x2-1), x1, x2 ∈ [-100000,101] . Can you write a demo for me, telling us how to use Herbgrind to measure the error of floating-point arithmetic expressions? Thank you very much.Best wishs to you!

[HazardousPeach]

This isn't really an "issue", and the author emailed me separately about their request, so I'm closing this.

[HazardousPeach]

Oh wait, I can't close it... Oh well