def create_df_preflop_hand_distrib(save=False):
"""returns df containing the hand rank distribution for all combinations of preflop hands"""
global df_preflop_hand_distrib
print "\n--------------- start create_df_preflop_hand_distrib"
print 'all preflop hands = \nstart = {}\nend = {}\nnb of elements = {}'.format(
all_preflop_hands[:10], all_preflop_hands[-10:],
len(all_preflop_hands))
print "\nhand rank distribution for all preflop hands"
t0 = timer()
preflop_hand_rank_distrib = np.zeros([1 + len(all_preflop_hands), 7462],
dtype=np.int32)
sys.stdout.write('\rk=%3d / %3d' % (1, 1 + len(all_preflop_hands)))
sys.stdout.flush()
N = C(52, 7)
h1_array_all_seven_cards = create_array_all_seven_fast()
h1_rank7 = EvalSeven.get_seven_rank_fast(
h1_array_all_seven_cards, keys.CARD_FLUSH_KEY, keys.CARD_FACE_KEY,
keys.CARD_SUIT, keys.SUIT_MASK, keys.SUIT_BIT_SHIFT,
EvalSeven.flush_rank, EvalSeven.face_rank, EvalSeven.flush_suit)
preflop_hand_rank_distrib[0, :] = np.bincount(h1_rank7)
for k, hand in enumerate(all_preflop_hands):
sys.stdout.write('\rk = %3d / %3d' %
(2 + k, 1 + len(all_preflop_hands)))
sys.stdout.flush()
hand_no = hand_str_to_no(hand)
N = C(52 - 2, 5)
array_five_cards = all_n_cards_given_m_cards_fast(5, hand_no, N)
h1_cards = np.ones([N, 2], dtype=np.int32) * hand_no
h1_array_all_seven_cards = np.concatenate((h1_cards, array_five_cards),
axis=1)
h1_array_all_seven_cards = np.array(h1_array_all_seven_cards,
dtype=np.int32)
h1_rank7 = EvalSeven.get_seven_rank_fast(
h1_array_all_seven_cards, keys.CARD_FLUSH_KEY, keys.CARD_FACE_KEY,
keys.CARD_SUIT, keys.SUIT_MASK, keys.SUIT_BIT_SHIFT,
EvalSeven.flush_rank, EvalSeven.face_rank, EvalSeven.flush_suit)
preflop_hand_rank_distrib[1 + k, :] = np.bincount(h1_rank7)
df = pd.DataFrame(data=preflop_hand_rank_distrib.T,
columns=['NoHand'] + all_preflop_hands,
index=np.arange(7462))
t1 = timer()
print '\ndf_preflop_hand_distrib time = {:8.4f} s'.format(t1 - t0)
if (save):
df.to_pickle(os.path.join('Tables', 'df_preflop_hand_distrib.pd'))
print '{} saved to disk as {}'.format(
'df_preflop_hand_distrib',
os.path.join('Tables', 'df_preflop_hand_distrib.pd'))
df.to_csv(os.path.join('Tables', 'df_preflop_hand_distrib.csv'),
index=False)
print '{} saved to disk as {}'.format(
'df_preflop_hand_distrib',
os.path.join('Tables', 'df_preflop_hand_distrib.csv'))
print "--------------- end create_df_preflop_hand_distrib"
return df
def test():
"""display all EvalSeven tables, run getSevenRank_ evaluators on sample hands, compare getSevenRank_ with EvalFive.getSevenRank_"""
global seven_cards, array_seven_cards, hr1, hr2, hr3, N, rank7, rank5, sparse_flush_rank, sparse_face_rank
print '\n--------------- start EvalSeven.test'
print '-------- EvalSeven tables'
sparse_flush_rank = sparse.csc_matrix(flush_rank)
print 'flush_rank = \n{}\n{}\nnb of nonzero elements = {}'.format(sparse_flush_rank.nonzero()[1], sparse_flush_rank.data, sparse_flush_rank.nonzero()[1].size)
sparse_face_rank = sparse.csc_matrix(face_rank)
print 'face_rank = \n{}\n{}\nnb of nonzero elements = {}'.format(sparse_face_rank.nonzero()[1], sparse_face_rank.data, sparse_face_rank.nonzero()[1].size)
print 'flush_suit = \n{}'.format(flush_suit)
print 'nb of elements different from -2 = {}'.format(np.array(flush_suit[flush_suit!=-2].size))
print '-------- EvalSeven getSevenRank_ evaluators sample runs'
np.random.seed(99)
array_seven_cards = np.array([np.random.choice(keys.DECK_SIZE, 7, replace=False) for k in xrange(10)])
print 'array_seven_cards=\n{}'.format(array_seven_cards)
seven_cards = array_seven_cards[0]
print 'seven_cards={}'.format(seven_cards)
hr1 = get_seven_rank_simple(seven_cards)
print '\nget_seven_rank_simple(seven_cards)={}'.format(hr1)
hr2 = get_seven_rank_full(array_seven_cards, keys.CARD_FLUSH_KEY, keys.CARD_FACE_KEY, keys.CARD_SUIT, keys.SUIT_MASK, keys.SUIT_BIT_SHIFT, flush_rank, face_rank, flush_suit)
print 'get_seven_rank_full(array_seven_cards)={}'.format(hr2)
hr3 = get_seven_rank_fast(array_seven_cards, keys.CARD_FLUSH_KEY, keys.CARD_FACE_KEY, keys.CARD_SUIT, keys.SUIT_MASK, keys.SUIT_BIT_SHIFT, flush_rank, face_rank, flush_suit)
print 'get_seven_rank_fast(array_seven_cards)={}'.format(hr3)
print '-------- compare getSevenRank_ with EvalFive.get_seven_rank_ on all hands'
t0 = timer()
array_all_seven_cards = create_array_all_seven_fast()
t1 = timer()
print 'array_all_seven_cards time = {:8.4f} s'.format(t1-t0)
t0 = timer()
rank7 = get_seven_rank_fast(array_all_seven_cards[:], keys.CARD_FLUSH_KEY, keys.CARD_FACE_KEY, keys.CARD_SUIT, keys.SUIT_MASK, keys.SUIT_BIT_SHIFT, flush_rank, face_rank, flush_suit)
t1 = timer()
print 'EvalSeven.get_seven_rank_fast time = {:8.4f} s'.format(t1-t0)
N = 1000000*140
t0 = timer()
rank5 = EvalFive.get_seven_rank_fast(array_all_seven_cards[:N], keys.FLUSH_KEY_FIVE, keys.FACE_KEY_FIVE, keys.CARD_SUIT, keys.CARD_FACE, EvalFive.flush_rank, EvalFive.face_rank)
t1 = timer()
print 'EvalFive.get_seven_rank_fast time = {:8.4f} s'.format(t1-t0)
print 'result(EvalSeven.get_seven_rank_fast)==result(EvalFive.get_seven_rank_fast) ? {}'.format((rank5==rank7).all())
print '--------------- end EvalSeven.test'
def create_df_hand_seven():
"""create summary dataframe for all seven card hands"""
hand_type = EvalFive.hand_type
uniq, idx = np.unique(hand_type, return_index=True)
sorted_idx = np.sort(idx)[::-1]
uniq_hand_type = hand_type[sorted_idx]
min_rank_5 = sorted_idx
max_rank_5 = np.r_[hand_type.size, sorted_idx[:-1]]-1
nb_rank_5 = max_rank_5-min_rank_5+1
array_all_seven_cards = create_array_all_seven_fast()
all_hand_rank = EvalSeven.get_seven_rank_fast(array_all_seven_cards, keys.CARD_FLUSH_KEY, keys.CARD_FACE_KEY, keys.CARD_SUIT, keys.SUIT_MASK, keys.SUIT_BIT_SHIFT, EvalSeven.flush_rank, EvalSeven.face_rank, EvalSeven.flush_suit)
bins = np.r_[sorted_idx[::-1], max_rank_5[0]]
nb_occurence = np.histogram(all_hand_rank, bins)[0][::-1]
proba = 1.0*nb_occurence/np.sum(nb_occurence)
odd = 1.0/proba
cum_proba = np.cumsum(proba)
bins = np.r_[sorted_idx[::-1], max_rank_5[0]+1][::-1]
digitized = np.digitize(all_hand_rank, bins)
min_rank_7 = np.zeros(uniq.size, dtype=np.int32)
max_rank_7 = np.zeros(uniq.size, dtype=np.int32)
nb_rank_7 = np.zeros(uniq.size, dtype=np.int32)
for k in range(1, uniq.size+1):
subset = all_hand_rank[digitized==k]
min_rank_7[k-1] = subset.min()
max_rank_7[k-1] = subset.max()
nb_rank_7[k-1] = np.unique(subset).size
proba_exact = map(lambda x : Rational(x, array_all_seven_cards.shape[0]), nb_occurence)
df = pd.DataFrame(data=np.array([uniq_hand_type, nb_rank_5, nb_rank_7, min_rank_5, min_rank_7, max_rank_5, max_rank_7, nb_occurence, proba_exact, odd, proba, cum_proba]).T,
columns=['HandType', 'NbHands_5', 'NbHands_7', 'MinRank_5', 'MinRank_7', 'MaxRank_5', 'MaxRank_7', 'NbOccurence', 'ProbaExact', 'Odd', 'ProbaApprox', 'CumulativeProba'])
# df.loc[uniq.size] = ['All', nb_rank_5.sum(), nb_rank_7.sum(), min_rank_5.min(), min_rank_7.min(), max_rank_5.max(), max_rank_7.max(), array_all_seven_cards.shape[0], 1.0, 1.0, 1.0, np.nan]
s = pd.Series(data=['All', nb_rank_5.sum(), nb_rank_7.sum(), min_rank_5.min(), min_rank_7.min(), max_rank_5.max(), max_rank_7.max(), array_all_seven_cards.shape[0], 1.0, 1.0, 1.0, np.nan],
index=['HandType', 'NbHands_5', 'NbHands_7', 'MinRank_5', 'MinRank_7', 'MaxRank_5', 'MaxRank_7', 'NbOccurence', 'ProbaExact', 'Odd', 'ProbaApprox', 'CumulativeProba'], name=uniq.size)
df = df.append(s)
df.to_pickle(os.path.join('Tables', 'df_hand_seven.pd'))
return df
def create_df_preflop_hand_distrib(save=False):
"""returns df containing the hand rank distribution for all combinations of preflop hands"""
global df_preflop_hand_distrib
print "\n--------------- start create_df_preflop_hand_distrib"
print 'all preflop hands = \nstart = {}\nend = {}\nnb of elements = {}'.format(all_preflop_hands[:10], all_preflop_hands[-10:], len(all_preflop_hands))
print "\nhand rank distribution for all preflop hands"
t0 = timer()
preflop_hand_rank_distrib = np.zeros([1+len(all_preflop_hands), 7462], dtype=np.int32)
sys.stdout.write('\rk=%3d / %3d' % (1, 1+len(all_preflop_hands)))
sys.stdout.flush()
N = C(52, 7)
h1_array_all_seven_cards = create_array_all_seven_fast()
h1_rank7 = EvalSeven.get_seven_rank_fast(h1_array_all_seven_cards, keys.CARD_FLUSH_KEY, keys.CARD_FACE_KEY, keys.CARD_SUIT, keys.SUIT_MASK, keys.SUIT_BIT_SHIFT, EvalSeven.flush_rank, EvalSeven.face_rank, EvalSeven.flush_suit)
preflop_hand_rank_distrib[0, :] = np.bincount(h1_rank7)
for k, hand in enumerate(all_preflop_hands):
sys.stdout.write('\rk = %3d / %3d' % (2+k, 1+len(all_preflop_hands)))
sys.stdout.flush()
hand_no = hand_str_to_no(hand)
N = C(52-2, 5)
array_five_cards = all_n_cards_given_m_cards_fast(5, hand_no, N)
h1_cards = np.ones([N, 2], dtype=np.int32)*hand_no
h1_array_all_seven_cards = np.concatenate((h1_cards, array_five_cards), axis=1)
h1_array_all_seven_cards = np.array(h1_array_all_seven_cards, dtype=np.int32)
h1_rank7 = EvalSeven.get_seven_rank_fast(h1_array_all_seven_cards, keys.CARD_FLUSH_KEY, keys.CARD_FACE_KEY, keys.CARD_SUIT, keys.SUIT_MASK, keys.SUIT_BIT_SHIFT, EvalSeven.flush_rank, EvalSeven.face_rank, EvalSeven.flush_suit)
preflop_hand_rank_distrib[1+k, :] = np.bincount(h1_rank7)
df = pd.DataFrame(data=preflop_hand_rank_distrib.T, columns=['NoHand']+all_preflop_hands, index=np.arange(7462))
t1 = timer()
print '\ndf_preflop_hand_distrib time = {:8.4f} s'.format(t1-t0)
if (save):
df.to_pickle(os.path.join('Tables', 'df_preflop_hand_distrib.pd'))
print '{} saved to disk as {}'.format('df_preflop_hand_distrib', os.path.join('Tables', 'df_preflop_hand_distrib.pd'))
df.to_csv(os.path.join('Tables', 'df_preflop_hand_distrib.csv'), index=False)
print '{} saved to disk as {}'.format('df_preflop_hand_distrib', os.path.join('Tables', 'df_preflop_hand_distrib.csv'))
print "--------------- end create_df_preflop_hand_distrib"
return df
def test():
"""display all EvalSeven tables, run getSevenRank_ evaluators on sample hands, compare getSevenRank_ with EvalFive.getSevenRank_"""
global seven_cards, array_seven_cards, hr1, hr2, hr3, N, rank7, rank5, sparse_flush_rank, sparse_face_rank
print '\n--------------- start EvalSeven.test'
print '-------- EvalSeven tables'
sparse_flush_rank = sparse.csc_matrix(flush_rank)
print 'flush_rank = \n{}\n{}\nnb of nonzero elements = {}'.format(
sparse_flush_rank.nonzero()[1], sparse_flush_rank.data,
sparse_flush_rank.nonzero()[1].size)
sparse_face_rank = sparse.csc_matrix(face_rank)
print 'face_rank = \n{}\n{}\nnb of nonzero elements = {}'.format(
sparse_face_rank.nonzero()[1], sparse_face_rank.data,
sparse_face_rank.nonzero()[1].size)
print 'flush_suit = \n{}'.format(flush_suit)
print 'nb of elements different from -2 = {}'.format(
np.array(flush_suit[flush_suit != -2].size))
print '-------- EvalSeven getSevenRank_ evaluators sample runs'
np.random.seed(99)
array_seven_cards = np.array([
np.random.choice(keys.DECK_SIZE, 7, replace=False) for k in xrange(10)
],
dtype=np.int32)
print 'array_seven_cards=\n{}'.format(array_seven_cards)
seven_cards = array_seven_cards[0]
print 'seven_cards={}'.format(seven_cards)
hr1 = get_seven_rank_simple(seven_cards)
print '\nget_seven_rank_simple(seven_cards)={}'.format(hr1)
hr2 = get_seven_rank_full(array_seven_cards, keys.CARD_FLUSH_KEY,
keys.CARD_FACE_KEY, keys.CARD_SUIT,
keys.SUIT_MASK, keys.SUIT_BIT_SHIFT, flush_rank,
face_rank, flush_suit)
print 'get_seven_rank_full(array_seven_cards)={}'.format(hr2)
hr3 = get_seven_rank_fast(array_seven_cards, keys.CARD_FLUSH_KEY,
keys.CARD_FACE_KEY, keys.CARD_SUIT,
keys.SUIT_MASK, keys.SUIT_BIT_SHIFT, flush_rank,
face_rank, flush_suit)
print 'get_seven_rank_fast(array_seven_cards)={}'.format(hr3)
print '-------- compare getSevenRank_ with EvalFive.get_seven_rank_ on all hands'
t0 = timer()
array_all_seven_cards = create_array_all_seven_fast()
t1 = timer()
print 'array_all_seven_cards time = {:8.4f} s'.format(t1 - t0)
t0 = timer()
rank7 = get_seven_rank_fast(array_all_seven_cards[:], keys.CARD_FLUSH_KEY,
keys.CARD_FACE_KEY, keys.CARD_SUIT,
keys.SUIT_MASK, keys.SUIT_BIT_SHIFT,
flush_rank, face_rank, flush_suit)
t1 = timer()
print 'EvalSeven.get_seven_rank_fast time = {:8.4f} s'.format(t1 - t0)
N = 1000000 * 140
t0 = timer()
rank5 = EvalFive.get_seven_rank_fast(array_all_seven_cards[:N],
keys.FLUSH_KEY_FIVE,
keys.FACE_KEY_FIVE, keys.CARD_SUIT,
keys.CARD_FACE, EvalFive.flush_rank,
EvalFive.face_rank)
t1 = timer()
print 'EvalFive.get_seven_rank_fast time = {:8.4f} s'.format(t1 - t0)
print 'result(EvalSeven.get_seven_rank_fast)==result(EvalFive.get_seven_rank_fast) ? {}'.format(
(rank5 == rank7).all())
print '--------------- end EvalSeven.test'
def create_df_hand_seven():
"""create summary dataframe for all seven card hands"""
hand_type = EvalFive.hand_type
uniq, idx = np.unique(hand_type, return_index=True)
sorted_idx = np.sort(idx)[::-1]
uniq_hand_type = hand_type[sorted_idx]
min_rank_5 = sorted_idx
max_rank_5 = np.r_[hand_type.size, sorted_idx[:-1]] - 1
nb_rank_5 = max_rank_5 - min_rank_5 + 1
array_all_seven_cards = create_array_all_seven_fast()
all_hand_rank = EvalSeven.get_seven_rank_fast(
array_all_seven_cards, keys.CARD_FLUSH_KEY, keys.CARD_FACE_KEY,
keys.CARD_SUIT, keys.SUIT_MASK, keys.SUIT_BIT_SHIFT,
EvalSeven.flush_rank, EvalSeven.face_rank, EvalSeven.flush_suit)
bins = np.r_[sorted_idx[::-1], max_rank_5[0]]
nb_occurence = np.histogram(all_hand_rank, bins)[0][::-1]
proba = 1.0 * nb_occurence / np.sum(nb_occurence)
odd = 1.0 / proba
cum_proba = np.cumsum(proba)
bins = np.r_[sorted_idx[::-1], max_rank_5[0] + 1][::-1]
digitized = np.digitize(all_hand_rank, bins)
min_rank_7 = np.zeros(uniq.size, dtype=np.int32)
max_rank_7 = np.zeros(uniq.size, dtype=np.int32)
nb_rank_7 = np.zeros(uniq.size, dtype=np.int32)
for k in range(1, uniq.size + 1):
subset = all_hand_rank[digitized == k]
min_rank_7[k - 1] = subset.min()
max_rank_7[k - 1] = subset.max()
nb_rank_7[k - 1] = np.unique(subset).size
proba_exact = map(lambda x: Rational(x, array_all_seven_cards.shape[0]),
nb_occurence)
df = pd.DataFrame(data=np.array([
uniq_hand_type, nb_rank_5, nb_rank_7, min_rank_5, min_rank_7,
max_rank_5, max_rank_7, nb_occurence, proba_exact, odd, proba,
cum_proba
]).T,
columns=[
'HandType', 'NbHands_5', 'NbHands_7', 'MinRank_5',
'MinRank_7', 'MaxRank_5', 'MaxRank_7', 'NbOccurence',
'ProbaExact', 'Odd', 'ProbaApprox', 'CumulativeProba'
])
# df.loc[uniq.size] = ['All', nb_rank_5.sum(), nb_rank_7.sum(), min_rank_5.min(), min_rank_7.min(), max_rank_5.max(), max_rank_7.max(), array_all_seven_cards.shape[0], 1.0, 1.0, 1.0, np.nan]
s = pd.Series(data=[
'All',
nb_rank_5.sum(),
nb_rank_7.sum(),
min_rank_5.min(),
min_rank_7.min(),
max_rank_5.max(),
max_rank_7.max(), array_all_seven_cards.shape[0], 1.0, 1.0, 1.0, np.nan
],
index=[
'HandType', 'NbHands_5', 'NbHands_7', 'MinRank_5',
'MinRank_7', 'MaxRank_5', 'MaxRank_7', 'NbOccurence',
'ProbaExact', 'Odd', 'ProbaApprox', 'CumulativeProba'
],
name=uniq.size)
df = df.append(s)
df.to_pickle(os.path.join('Tables', 'df_hand_seven.pd'))
return df
