Solved! - Trouble creating design matrix cued switching task

favdberg

Problem has been solved (I asked the question too soon...).

favdberg

This is the code I have so far to generate the design matrix. This code runs smoothly on my computer for approx. 16 trials, but my computer can't handle the computation of a matrix with 144 trials...

# SET-UP
# Design matrix PRACTICE list with 8 circles/squares, 8 blue/orange items,
# 8 color/shape cues, and 8 switch/non-switch trials


prac_colors = 8 * ["blue"] + 8 * ["orange"] # [0]: 0 = blue, 1 = orange
prac_shapes = 4 * ["circle"] + 4 * ["square"] + 4 * ["circle"] + 4 * ["square"] # [1]: 0 = circle, 1 = square
prac_cue = 8 * ["color","shape"] # [2]: 0 = color, 1 = shape
prac_design_matrix = zip(prac_colors, prac_shapes, prac_cues)
# print prac_design_matrix


def chain_check(prac_switch_list):
	''' this counts consecutive indices, from the second element to the final one
		the longest chain is then selected and its length is computed'''


	indices = []
	for check_count,x in enumerate(prac_switch_list):
		if check_count == 0:
			pass
		else:
			previous = prac_switch_list[check_count -1]
			current = prac_switch_list[check_count]


			if previous == current:
				indices.append(check_count)
			else:
				pass
	
	longest_chain = len(max(np.split(indices, np.where(np.diff(indices) != 1)[0]+1), key=len).tolist()) + 1
	
	return longest_chain

def draw_permutation(prac_design_matrix):
	prac_number_of_switches = 0
	while prac_number_of_switches != 8:
		# generate a random permutation of the design matrix
		prac_mixed_list = np.random.permutation(prac_design_matrix).tolist()
		
		# generate switch information
		prac_switch_list = []


		for i,x in enumerate(prac_mixed_list):    
			if True:
				if i == 0:
					prac_switch_list.append("non-switch")
				
				if i > 0 and i < (len(prac_mixed_list)):
					previous = prac_mixed_list[i - 1]
					current = prac_mixed_list[i]
					if previous[2] == current[2]:
						prac_switch_list.append("non-switch")
					else:
						prac_switch_list.append("switch")
			# print longest_chain
		prac_number_of_switches = prac_switch_list.count('switch')


	# add switch information to permuted matrix 
	for trial in range(len(prac_mixed_list)):
		prac_mixed_list[trial].append(prac_switch_list[trial])


	return prac_mixed_list, chain_check(prac_switch_list)

longest_chain = 5
while longest_chain > 4:  
	final_prac_mixed_list = draw_permutation(prac_design_matrix)
	longest_chain = final_prac_mixed_list[1]
	print(longest_chain)
	
# final design matrix is below  
for trial in final_prac_mixed_list[0]:
	print(trial)

final_prac_mixed_list = final_prac_mixed_list[0]

Is there a more efficient or computationally less complicated way to generate such a matrix? Hope someone can help out! Let me know if you need any other info.

Floor

eduard

Hi Floor,

Is this problem solved or not? I am a bit confused with the order of your postings?

Eduard