def setUp(self):
random.seed(2)
self.DELTA = 1
self.names = [
"tests/eoinseed1.txt", "tests/eoinseed2.txt",
"tests/eoinseed3.txt", "tests/eoinseed4.txt",
"tests/eoinseed5.txt", "tests/eoinseed6.txt",
"tests/eoinseed7.txt", "tests/eoinseed8.txt",
"tests/eoinseed9.txt", "tests/eoinseed10.txt"
]
self.grn = grn.GRN()
def test_eoin(self):
sizes = []
final_concs = []
correct_sizes = [7, 8, 7, 7, 8, 9, 10, 10, 7, 8]
correct_concs = [
0.10536119562462604, 1.0785531576949638e-05, 0.96204392235109859,
0.036227250980492995, 0.14936626819446591, 0.0017180411368314544,
0.74527253618090816, 0.20216842837768173, 1.8597966179449688e-41,
9.8837209610607373e-11, 2.8433950003789704e-12,
3.7272818452249546e-19, 0.79783157161841423, 0.99999999990116273,
1.0605769176570928e-12, 0.43021628377545634,
1.8607359883012191e-14, 7.8687897480028881e-11,
0.10528617335121833, 0.99999999999998135, 1.1568225120729014e-75,
0.46449754279463745, 1.0, 1.0, 3.852020054491351e-50,
2.2676764109720152e-47, 1.2698664131258545e-50,
4.614598027941728e-51, 1.0230365114501557e-49, 0.96250388634805484,
0.10434327320808674, 9.8246257052620553e-11,
9.8246257052620553e-11, 0.89565672649717465, 0.015449167345545771,
9.8246257052620553e-11, 0.022046946306399435,
1.549088544201994e-51, 2.0653767133682643e-49,
3.1437911279343385e-51, 0.18145538980022971, 1.0,
5.5872364375449987e-50, 0.1892264857816604, 0.62931812441810986,
5.4289702573481519e-52, 0.00076655281898026532,
0.99999999960006059, 0.0021459130763556624, 9.9984846630215615e-11,
0.75450804442096187, 0.21977813174014549, 9.9984846630215615e-11,
9.9984846630215615e-11, 9.9984846630215615e-11,
0.022801357943556831, 0.016570752661809442, 0.006791886227358491,
0.012708024669514883, 0.0055977427316390221, 0.97053528744991513,
0.0022669597556181486, 9.3192992915059737e-11,
0.0061020387097514761, 3.4951165492265383e-08, 0.97942727275003505,
0.17202757039233571, 0.99999999989999022, 1.0000982013359345e-10,
0.00012387320936094268, 0.77259278064842241, 0.055248455240959803,
7.3205089211345301e-06, 4.4798299830043145e-05,
0.10714347255184113, 0.83990369807369525, 9.9719383174870601e-11,
0.16009630172686587, 5.2259906141012131e-126,
9.9719383174870601e-11, 0.89281172914832874
]
for name in self.names:
self.grn = grn.GRN(delta=self.DELTA)
self.grn.read_genome(name)
self.grn.build_genes()
self.grn.precalc_matrix()
self.grn.regulate_matrix(2000)
sizes.append(len(self.grn.conc_list))
for conc in self.grn.conc_list:
final_concs.append(conc[-1])
for i in range(len(sizes)):
self.assertEqual(sizes[i], correct_sizes[i])
for i in range(len(final_concs)):
self.assertEqual(final_concs[i], correct_concs[i])
def test_miguel(self):
concs = []
correct_concs = [
2.4694881581878343e-12, 1.1537420071432216e-12,
2.3929814982520794e-37, 0.79999999999637683, 0.05, 0.05, 0.05, 0.05
]
self.grn = grn.GRN(1)
self.grn.read_genome("tests/miguel_parsed.txt")
self.grn.build_genes()
self.grn.add_extra("EXTRA_A", 0.05, [0] * 32)
self.grn.add_extra("EXTRA_B", 0.05, [1] * 32)
self.grn.add_extra("EXTRA_C", 0.05, [0] * 16 + [1] * 16)
self.grn.add_extra("EXTRA_D", 0.05, [1] * 16 + [0] * 16)
self.grn.precalc_matrix()
size = len(self.grn.conc_list)
self.assertEqual(size, 8)
self.grn.regulate_matrix(1000)
for conc in self.grn.conc_list:
concs.append(conc[-1])
for i in range(len(concs)):
self.assertEquals(concs[i], correct_concs[i])
def run_grn(genome, delta):
stabiter = 10000 / delta
runiter = 1000 / delta
completed = False
results = None
bh_total = 0
#run until delta stays above zero
while not completed:
regnet = grn.GRN(genome, delta)
regnet.build_genes()
regnet.add_extra("EXTRA_sineval", 0.0, [0]*32)
results = [0] * len(regnet.genes)
regnet.precalc_matrix()
#build list of p-genes, quit if only 1 p gene
p_genes = [ idx for idx, gn in enumerate(regnet.genes)
if gn.gene_type == "P"]
if len(p_genes) < 2: return [0],0
regnet.regulate_matrix(stabiter) # stabilise the grn
# use sine as input value for grn
onff = 0
for i in range(0, 50):
if i % 10 == 0:
if onff == 1:
onff = 0
else:
onff = 1
inputval = onff
extra_vals = {'sineval':inputval}
regnet.set_extras(extra_vals)
regnet.regulate_matrix(runiter)
samples = range(stabiter, (runiter*50)+stabiter, runiter)
#only finish if it doesn't go below zero
if regnet.below_zero == False:
completed = True
else:
delta = delta - 1
stabiter = 10000 / delta
runiter = 1000 / delta
# read results from grn to calc fitness
for p_gene in p_genes:
fitness = 0
onff = 0
for j in range(0, 50):
# if j % 10 == 0:
# if onff == 1:
# onff = 0
# else:
# onff = 1
# target = onff
idx, idy = p_gene, samples[int(j)]
target = regnet.conc_list[-1][idy]
signal = regnet.conc_list[idx][idy]
# print "T", target, "S",regnet.conc_list[-1][idy]
fitness += 1 - abs(target - signal)
results[p_gene] += fitness
return results, regnet.conc_list
def run_grn(genome, companies, delta):
stabiter = 10000 / delta
runiter = 1000 / delta
completed = False
grn = regnet.GRN(genome)
grn.build_genes()
train_results = [0] * grn.p_genes
test_results = [0] * grn.p_genes
bh_total = 0
for compid, data in enumerate(companies):
while not completed:
cash_total = 1000
cash_list = []
signal = []
buy_hold = 1000 / data['shareprice'][0]
grn = regnet.GRN(genome, delta)
grn.build_genes()
grn.add_extra("EXTRA_priceearn", 0.0, [0]*32)
grn.add_extra("EXTRA_pricebook", 0.0, [1]*32)
grn.add_extra("EXTRA_beta", 0.0, [0]*16 +[1]*16)
grn.add_extra("EXTRA_cashflow", 0.0, [1]*16 +[0]*16)
grn.precalc_matrix()
#build list of p-genes, quit if only 1 p gene
p_genes = [ idx for idx, gn in enumerate(grn.genes)
if gn.gene_type == "P"]
if len(p_genes) < 2: return 0,0
grn.regulate_matrix(stabiter) # stabilise the grn
# use compustat as GRN input for 60 months
for month in range(60):
extra_vals = {}
for key in data:
if not key == 'shareprice':
extra_vals[key] = data[key][month]
grn.set_extras(extra_vals)
grn.regulate_matrix(runiter)
samples = range(stabiter, (runiter*60)+stabiter, runiter)
if grn.below_zero == False:
completed = True
else:
delta = delta - 1
stabiter = 10000 / delta
runiter = 1000 / delta
print "lowering delta to", delta, "r",runiter,"s",stabiter
for i, p_gene in enumerate(p_genes):
cash_total = 1000
share_count = 0
invest = 0
for month in range(60):
share_price = data['shareprice'][month]
cash_total = (share_price*share_count) + cash_total
cash_list.append(cash_total)
idx, idy = p_gene, samples[month]
signal.append(grn.conc_list[idx][idy])
invest = grn.conc_list[idx][idy] * cash_total
cash_total = cash_total - invest
share_count = invest / share_price
final = (share_price*share_count) + cash_total
if compid < 5:
train_results[i] += final
else:
test_results[i] += final
buy_hold = buy_hold * share_price
bh_total += buy_hold
bestidx = train_results.index(max(train_results))
return train_results[bestidx], test_results[bestidx]