I want to do SVM classification (i.e. OneClassSVM) with sklearn.svm.OneClassSVM on physical states that come from a different library (tenpy). I'd define a custom kernel
def overlap(X,Y):
return np.array([[x.overlap(y) for y in Y] for x in X])
where overlap() is a defined function in said library to calculate the overlap between states. When I try to fit with my data
clf = OneClassSVM(kernel=overlap)
clf.fit(states)
where states is a list of such state objects, I get the error
TypeError: float() argument must be a string or a number, not 'MPS'
Is there a way to tell sklearn to ignore this test (w/o editing the source code)?
To my understanding the nature of the data and how it's processed is in principal not essential to the algorithm as long as there is a well-defined kernel for the objects.
