Most importantly, fixes a bug in partition_nnls() which rendered that function useless.
Secondly, optimises partition_nnls() by using a QR factorisation of the problem before calling nnls() on an equivalent but much smaller problem.
Finally, adds a new integral constraint for both nnls and lstsq which ensures that the sum of the partition recovers the integral of the input function.
Fixes: #144
Fixed: A bug in partition() with the default method.
Changed: Both partition(method="nnls") and partition(method="lstsq") now have an additional integral constraint so that the sum of the partition recovers the integral of the input function.
Most importantly, fixes a bug in
partition_nnls()which rendered that function useless.Secondly, optimises
partition_nnls()by using a QR factorisation of the problem before callingnnls()on an equivalent but much smaller problem.Finally, adds a new integral constraint for both
nnlsandlstsqwhich ensures that the sum of the partition recovers the integral of the input function.Fixes: #144 Fixed: A bug in
partition()with the default method. Changed: Bothpartition(method="nnls")andpartition(method="lstsq")now have an additional integral constraint so that the sum of the partition recovers the integral of the input function.