Closed tlnagy closed 8 years ago
The behavior of the current sigmoid function (https://github.com/tlnagy/pooled-screen-optimization/blob/9102f72f38466108ac4a182fa9ed1949f12f5a19/src/library.jl#L5) is not ideal. A 100% KD doesn't always correspond to the maximum phenotype so the tests fail.
I think something like this could work well.
The behavior of the current sigmoid function (https://github.com/tlnagy/pooled-screen-optimization/blob/9102f72f38466108ac4a182fa9ed1949f12f5a19/src/library.jl#L5) is not ideal. A 100% KD doesn't always correspond to the maximum phenotype so the tests fail.