Currently, the hash in WeightMaps is computed from its position and weights. To compare two positions, its coordinates are rounded to 1e-6. We don't have this rounding for weights, which exposes two DetuningMaps to be identified as different due to floating point errors. Comparing weights with a precision of 1e-6 should be good enough.
Currently, the hash in WeightMaps is computed from its position and weights. To compare two positions, its coordinates are rounded to 1e-6. We don't have this rounding for weights, which exposes two DetuningMaps to be identified as different due to floating point errors. Comparing weights with a precision of 1e-6 should be good enough.