negative cross mate correlations break here

[sashagusev]

When the cross-mate correlation is set to be negative, this line breaks and produces no realized correlation. A simple fix is to correct for the correlation sign:

signal_scale = np.sqrt(abs(R))
noise_scale = np.sqrt(max(0.0, 1 - abs(R)))
sign = 1.0 if R >= 0 else -1.0

mating_score1 = sum_scaled_mate1 * signal_scale + noise_scale * np.random.normal(0, np.std(sum_scaled_mate1), n)
mating_score2 = sign * sum_scaled_mate2 * signal_scale + noise_scale * np.random.normal(0, np.std(sum_scaled_mate2), n)

xftsim/xftsim/mate.py

Lines 872 to 873 in 00080b3

	mating_score1 = sum_scaled_mate1 * np.sqrt(R) + np.sqrt(np.abs(1-R))*np.random.normal(0,np.std(sum_scaled_mate1),n)#sum_scaled_mate1.shape[0])
	mating_score2 = sum_scaled_mate2 * np.sqrt(R) + np.sqrt(np.abs(1-R))*np.random.normal(0,np.std(sum_scaled_mate2),n)#sum_scaled_mate2.shape[0])