Workaround for scipy.stats seeding issue

paltas/Sampling/sampler.py

@@ -7,6 +7,8 @@
 set from the input distributions.
 """
 import warnings
+from scipy._lib._util import check_random_state
+
 # Definte the components we need the sampler to consider.
 # TODO: add point source & lens light
 lensing_components = ['subhalo','los','main_deflector','source','lens_light',
@@ -56,6 +58,17 @@ def draw_from_dict(draw_dict):
 		for key in sorted(draw_dict):
 
 			value = draw_dict[key]
+
+			# Dirty workaround for https://github.com/scipy/scipy/issues/16998
+			# Explicitly ensure the distributions use the global random state
+			if hasattr(value, '__self__'):
+				# Scipy distribution
+				value.__self__.random_state = check_random_state(None)
+			elif hasattr(value, 'dist'):
+				# Duplicate or MultipleValues. 
+				# Other custom paltas dists not yet supported..
+				value.dist.__self__.random_state = check_random_state(None)
+
 			if callable(value):
 				# Sampling function: call to draw
 				draw = value()