def par2(api: GBD, query, runtimes, timeout, divisor):
for name in runtimes:
times = api.query_search(query, [], [name])
div = len(times) if divisor is None else divisor
par2 = sum(float(time[1]) if util.is_number(time[1]) and float(time[1]) < timeout else 2*timeout for time in times) / div
solved = sum(1 if util.is_number(time[1]) and float(time[1]) < timeout else 0 for time in times)
print(str(round(par2, 2)) + " " + str(solved) + "/" + str(div) + " " + name)
times = api.query_search(query, [], runtimes)
div = len(times) if divisor is None else divisor
vbs_par2 = sum([min(float(val) if util.is_number(val) and float(val) < timeout else 2*timeout for val in row[1:]) for row in times]) / div
solved = sum(1 if t < timeout else 0 for t in [min(float(val) if util.is_number(val) else 2*timeout for val in row[1:]) for row in times])
print(str(round(vbs_par2, 2)) + " " + str(solved) + "/" + str(div) + " VBS")
def calculate_par2_score(self, query, feature):
info = self.database.meta_record(feature)
if not "timeout" in info:
eprint(
"Time-limit 'timeout' missing in meta-record of table '{}'.".
format(feature))
eprint("Unable to calculate score.")
return
if not "memout" in info:
eprint(
"Memory-limit 'memout' missing in meta-record of table '{}'.".
format(feature))
if not "machine" in info:
eprint(
"Machine-id 'machine' missing in meta-record of table '{}'.".
format(feature))
timeout = int(info["timeout"])
times = self.query_search(query, [], [feature])
score = 0
penalized = set()
for time in times:
if is_number(time[1]):
score += int(time[1])
else:
score += 2 * timeout
penalized.add(time[1])
print(score / len(times))
print(penalized)
def greedy_comb(api: GBD, query, runtimes, timeout, size):
result = api.query_search(query, [], runtimes)
result = [[float(val) if util.is_number(val) and float(val) < float(timeout) else 2*timeout for val in row] for row in result]
runtimes.insert(0, "dummy")
for comb in combinations(range(1, len(runtimes)), size):
comb_par2 = sum([min(itemgetter(*comb)(row)) for row in result]) / len(result)
print(str(itemgetter(*comb)(runtimes)) + ": " + str(comb_par2))
def table_values(self, table):
result = {"numeric": [None, None], "discrete": []}
values = self.value_query(
'SELECT DISTINCT value FROM {}'.format(table))
for val in values:
if is_number(val):
value = float(val)
if result["numeric"][0] == None:
result["numeric"] = [value, value]
elif value < result["numeric"][0]:
result["numeric"][0] = value
elif value > result["numeric"][1]:
result["numeric"][1] = value
else:
result["discrete"].append(val)
return result
def optimal_comb(api: GBD, args):
result = api.query_search(args.query, [], args.runtimes)
result = [[
int(float(val)) if is_number(val) and float(val) < float(args.tlim)
else int(2 * args.tlim) for val in row[1:]
] for row in result]
cnf = dimacs.DIMACSPrinter()
_ACT = [cnf.create_literal() for _ in range(0, len(args.runtimes))]
_MAX = get_bitvector(cnf, int(pow(2, _BW - 1) - 1))
for c in encode_at_most_k_constraint_ltseq(cnf, args.size, _ACT):
cnf.consume_clause(c)
# encode row-wise minima
MINS = []
for row in result:
i = 0
B0_ = get_bitvector(cnf, int(row[0]))
B0 = if_then_else(cnf, B0_, _MAX, _ACT[i])
for rt in row[1:]:
i = i + 1
B1_ = get_bitvector(cnf, int(rt))
B1 = if_then_else(cnf, B1_, _MAX, _ACT[i])
Bcarry = get_carry_bits(cnf, B1, [-i for i in B0])
Bmin = if_then_else(cnf, B0, B1, Bcarry[_BW - 1])
B0 = Bmin
MINS.append(B0) # B0 is now minimum of row
# encode sum of minima
A = MINS[0]
for B in MINS[1:]:
SUM = get_sum_bits(cnf, A, B)
A = SUM
solver = Cadical(bootstrap_with=cnf.clauses, with_proof=False)
result = solver.solve()
if result == True:
model = solver.get_model()
print(slice_model(model, _ACT))
print(decode_bitvector(slice_model(model, A)))
def optimal_comb(api: GBD, query, runtimes, timeout, k):
result = api.query_search(query, [], runtimes)
result = [[
int(float(val))
if is_number(val) and float(val) < float(timeout) else int(2 * timeout)
for val in row[1:]
] for row in result]
dataset = pd.DataFrame(result, columns=runtimes)
dataset = dataset[(dataset != 2 * timeout).any(axis='columns')]
model = mip.Model()
instance_solver_vars = [[
model.add_var(f'x_{i}_{j}', var_type=mip.BINARY)
for j in range(dataset.shape[1])
] for i in range(dataset.shape[0])]
solver_vars = [
model.add_var(f's_{j}', var_type=mip.BINARY)
for j in range(dataset.shape[1])
]
for var_list in instance_solver_vars: # per-instance constraints
model.add_constr(mip.xsum(var_list) == 1)
for j in range(dataset.shape[1]): # per-solver-constraints
model.add_constr(
mip.xsum(instance_solver_vars[i][j]
for i in range(dataset.shape[0])) <=
dataset.shape[0] * solver_vars[j]) # "Implies" in Z3
model.add_constr(mip.xsum(solver_vars) <= k)
model.objective = mip.minimize(
mip.xsum(instance_solver_vars[i][j] * int(dataset.iloc[i, j])
for i in range(dataset.shape[0])
for j in range(dataset.shape[1])))
print(dataset.sum().min())
print(model.optimize())
print(model.objective_value)
for index, item in enumerate([var.x for var in solver_vars]):
if item > 0:
print(runtimes[index])
def vbs(api: GBD, query, runtimes, timeout, separator):
result = api.query_search(query, [], runtimes)
resultset = [(row[0], min(float(val) if util.is_number(val) else 2*timeout for val in row[1:])) for row in result]
for result in resultset:
print(separator.join([(str(item or '')) for item in result]))
def cdf(api: GBD, query, runtimes, timeout, title):
plt.rcParams.update({'font.size': 8})
result = api.query_search(query, [], runtimes)
result = [[
float(val)
if util.is_number(val) and float(val) < float(timeout) else timeout
for val in row[1:]
] for row in result]
df = pd.DataFrame(result)
df.columns = runtimes
df['vbs'] = df[runtimes].min(axis=1)
print(df)
plt.rcParams['axes.prop_cycle'] = plt.cycler(color=coolors2 + coolors)
params = {
'legend.fontsize': 'small',
'axes.labelsize': 6,
'axes.titlesize': 6,
'xtick.labelsize': 6,
'ytick.labelsize': 6,
'axes.titlepad': 10
}
plt.rcParams.update(params)
fig = plt.figure()
ax = fig.add_subplot(111)
plt.xlim(0, timeout + 100)
plt.grid(linestyle='-', linewidth=.5)
plt.axvline(x=timeout, linestyle='dashed', color='black', linewidth=.5)
#plt.ylim(0, len(result))
# Build Title
if (title is None):
title = []
for elem in query.split('=')[1:]:
parts = elem.strip().split(' ')
title = title + [parts[0].replace('_', ' ').title()]
ax.set_title(", ".join(title), fontsize=6, variant='small-caps')
else:
ax.set_title(title, fontsize=6, variant='small-caps')
df2 = pd.DataFrame(index=range(timeout + 100), columns=runtimes)
df2.fillna(0)
for col in ['vbs'] + runtimes:
df2[col] = [0] * (timeout + 100)
for val in df[col]:
if val < timeout:
df2.loc[round(val), col] = df2[col][round(val)] + 1
sum = 0
for val in range(1, timeout + 100):
df2.loc[val, col + "_"] = NaN
if df2[col][val] != 0:
df2.loc[val, col + "_"] = sum
sum = sum + df2.loc[val, col]
df2.loc[val, col] = sum
markers = itertools.cycle([
'o', 'v', '^', '<', '>', 'p', 'P', '*', 'h', 'H', '8', 'X', 'd', 'D',
's'
])
#next(markers)
#next(markers)
order = len(runtimes) + 1
for col in ['vbs'] + runtimes:
color = next(ax._get_lines.prop_cycler)['color']
ax.plot(df2[col],
zorder=order,
linestyle='-',
linewidth=.5,
color=color)
ax.plot(df2[str(col) + "_"],
label=col,
zorder=order,
fillstyle='none',
marker=next(markers)[0],
alpha=.9,
markeredgewidth=.5,
markersize=3,
drawstyle='steps-post',
color=color)
order = order - 1
plt.legend(ncol=2, loc='lower right')
plt.savefig('out.svg', transparent=True, bbox_inches='tight', pad_inches=0)
plt.show()