Reducing computing time in floss

[NimaSarajpoor]

We can see the following lines in method update:

stumpy/stumpy/floss.py

Lines 560 to 565 in 59ee63a

	# Ingress
	if self._normalize:
	M_T, Σ_T = core.compute_mean_std(self._T, self._m)
	D = core.mass(self._finite_Q, self._finite_T, M_T, Σ_T)
	else:
	D = core.mass_absolute(self._T[-self._m :], self._T)

As shown above, we are computing the full arrays M_T and Σ_T in each call of this method. However, this might be time consuming, and a better way is to just update these arrays in each call. It seems that this part of code is similar to:

stumpy/stumpy/stumpi.py

Lines 231 to 242 in 59ee63a

	if np.any(~self._T_isfinite[-self._m :]):
	μ_Q = np.inf
	σ_Q = np.nan
	else:
	μ_Q, σ_Q = core.compute_mean_std(S, self._m)
	μ_Q = μ_Q[0]
	σ_Q = σ_Q[0]
	self._M_T[:-1] = self._M_T[1:]
	self._Σ_T[:-1] = self._Σ_T[1:]
	self._M_T[-1] = μ_Q
	self._Σ_T[-1] = σ_Q

which is a part of method _update_egress in class stumpi. Here, we take advantage of the previously-computed M_T and Σ_T, and we just update it.

In floss, we need to create attributes M_T and Σ_T when we create an instance of this class, and then update them.