Closed byorgey closed 6 years ago
Apparently deformEnvelope
has been buggy this whole time, but the bug was exactly canceled out by the particular way it was used in the examples in the user manual. In particular, the envelope that resulted from calling deformEnvelope
gave inconsistent results depending on the length of the input vector. In particular it only worked correctly if given a vector of length exactly 2. There were strange values of 0.5
in the code which really should be 1
, and another value was being divided by only the norm of the input vector where it should have been the squared norm.
From
benzrf
in IRC: .Or consider
which produces You can see that the squares are clearly at distance 1/2 from each other, but should be 1.