def create_english_vec(N=N,k=k, cluster_sz = [2]):
print "generating english vector of cluster size", cluster_sz
total_eng = np.zeros((1,N))
# generate english vector
for cz in cluster_sz:
english_vector = random_idx.generate_RI_lang(N, RI_letters, cz, ordered, languages=['eng'])
total_eng += english_vector
normed_eng = total_eng/np.linalg.norm(total_eng)
return total_eng, normed_eng
def create_english_vec(N=N, k=k, cluster_sz=[2]):
print "generating english vector of cluster size", cluster_sz
total_eng = np.zeros((1, N))
# generate english vector
for cz in cluster_sz:
english_vector = random_idx.generate_RI_lang(N,
RI_letters,
cz,
ordered,
languages=['eng'])
total_eng += english_vector
normed_eng = total_eng / np.linalg.norm(total_eng)
return total_eng, normed_eng
def generate_RI_lang(N, RI_letters, cluster_sz, ordered, languages=None):
return random_idx.generate_RI_lang(N, RI_letters, cluster_sz, ordered, languages)
#sparsity = sparsities[j] #k = int(N*sparsity/2) k = ks[j] if k >= N: continue RI_letters = random_idx.generate_letter_id_vectors(N,k) total_vec = [] print N,k print '==========' # iterate over ordered vs unordered for cluster_sz in cluster_sizes: print "~~~~~~~~~~" print 'cz = ', cluster_sz # calculate language vectors lang_vectors = random_idx.generate_RI_lang(N, RI_letters, cluster_sz, ordered, languages=languages) total_vec.append(lang_vectors) # print cosine angles if ordered == 0: ord_str = 'unordered!' else: ord_str = 'ordered!' # calculate total vector final_lang = sum(total_vec) # calculate variance of cos angle distribution cosangles = utils.cosangles(final_lang, languages) vary = utils.var_measure(cosangles) V[i,j] = vary print 'N = ' + str(N) + '; k = ' + str(k) + '; letters clusters are ' + str(cluster_sizes) + ', ' + ord_str + '\n'
def generate_RI_lang(N,RI_letters, cluster_sz, ordered, languages=None): return random_idx.generate_RI_lang(N,RI_letters, cluster_sz, ordered, languages)