clean up monotonicity checks

[rnugent3]

use @HenryGeorgist 's rule stuff

[rnugent3]

william.lehman 11:31 AM i think we do it in the sample method directly so any time we produce a paired data from a sample of uncertain paired data we call validate

richard.nugent richard.nugent 11:32 AM Ok. I'll write an issue to clean up the monotonicity check

william.lehman 11:32 AM then we only have other cases where we dont interact with uncertain paired data itself we also should have validation on uncertain paired data itself that way when simulation calls validate we just validate all the curves too

[HenryGeorgist]

i found that we are producing non monotonic paired data all over the place

[HenryGeorgist]

        public IPairedData SamplePairedData(double probability)
        {
            double[] y = new double[_yvals.Length];
            for (int i = 0; i < _xvals.Length; i++)
            {
                y[i] = _yvals[i].InverseCDF(probability);
            }
            PairedData pd = new PairedData(_xvals, y, Category);//mutability leakage on xvals
            pd.Validate();
            if (pd.HasErrors){
                throw new Exception("the produced paired data is not monotonically increasing."); //TODO: be more intellegent abou this. possibly use force monotonic in range from go-consequences.paireddata
            }
            return pd;
        }

[HenryGeorgist]

this fails 12 of our tests. mostly levee tests i think. i can fix the issue i think for a temporary patch, but there is something wrong here.

[rnugent3]

The probability of failure is constant at and over top of levee elevation

Best,

R Nugent III

On Feb 2, 2022, at 7:25 PM, Will Lehman @.***> wrote:

 this fails 12 of our tests. mostly levee tests i think. i can fix the issue i think for a temporary patch, but there is something wrong here.

— Reply to this email directly, view it on GitHub, or unsubscribe. You are receiving this because you were assigned.

[HenryGeorgist]

yeah - so errors are existing even if we have increasing values. there is still some serious problems going on that we need to probably fix, but i have it passing all tests.

[HenryGeorgist]

implemented strict monotonic increasing and monotonic increasing. need to finish decreasing - it is a bit trickier

[HenryGeorgist]

@rnugent3 didnt we fix this with alex?

[rnugent3]

Yes, I agree. Closing issue.