def bdf2_imr_ptinfos(pt):
# assert pt == sRat(1,2)
return [
PTInfo(pt.time, None, "imr"),
PTInfo(pt.time + sRat(1, 2) + 1, "corr_val", None),
PTInfo(pt.time + sRat(1, 2) + 2, "corr_val", None)
]
def test_sum_dts():
# Check a simple fractional case
utils.assert_sym_eq(sum_dts(sRat(1, 2), 1), Sdts[0]/2)
# Check two numbers the same gives zero always
utils.assert_sym_eq(sum_dts(1, 1), 0)
utils.assert_sym_eq(sum_dts(0, 0), 0)
utils.assert_sym_eq(sum_dts(sRat(1, 2), sRat(1, 2)), 0)
def check_sum_dts(a, b):
# Check we can swap the sign
utils.assert_sym_eq(sum_dts(a, b), -sum_dts(b, a))
# Starting half a step earlier
utils.assert_sym_eq(sum_dts(a, b + sRat(1, 2)),
sRat(1, 2)*Sdts[int(b)] + sum_dts(a, b))
# Check that we can split it up
utils.assert_sym_eq(sum_dts(a, b),
sum_dts(a, b-1) + sum_dts(b-1, b))
# Make sure b>=1 or the last check will fail due to negative b.
cases = [(0, 1),
(5, 8),
(sRat(1, 2), 3),
(sRat(2, 2), 3),
(sRat(3, 2), 3),
(0, sRat(9, 7)),
(sRat(3/4), sRat(9, 7)),
]
for a, b in cases:
yield check_sum_dts, a, b
def test_exact_predictors():
def check_exact_predictors(exact, predictor):
p1_func, y_np1_p1_expr = \
generate_predictor_scheme(*predictor, symbolic=exact)
# LTE for IMR: just natural lte:
y_np1_imr_expr = y_np1_exact - imr_lte(Sdts[0], Sdddynph, SFddynph)
# Generate another predictor
p2_func, y_np1_p2_expr = \
generate_predictor_scheme([PTInfo(sRat(1, 2), None, "imr"),
PTInfo(2, "corr_val", None),
PTInfo(3, "corr_val", None)], "ibdf2")
A = system2matrix([y_np1_p1_expr, y_np1_p2_expr, y_np1_imr_expr],
[Sdddynph, SFddynph, y_np1_exact])
# Solve for dddy and Fddy:
x = A.inv()
dddy_symb = sum([
y_est * xi.factor()
for xi, y_est in zip(x.row(0), [y_np1_p1, y_np1_p2, y_np1_imr])
])
Fddy_symb = sum([
y_est * xi.factor()
for xi, y_est in zip(x.row(1), [y_np1_p1, y_np1_p2, y_np1_imr])
])
# Check we got the right matrix and the right formulae
if predictor[1] == "use exact dddy":
assert A.row(0) == sympy.Matrix([[1, 0, 0]]).row(0)
utils.assert_sym_eq(dddy_symb.simplify(), y_np1_p1)
elif predictor[1] == "use exact Fddy":
assert A.row(0) == sympy.Matrix([[0, 1, 0]]).row(0)
utils.assert_sym_eq(Fddy_symb.simplify(), y_np1_p1)
else:
assert False
exacts = [
sympy.exp(St),
sympy.sin(St),
]
exact_predictors = [
([PTInfo(0, None, None),
PTInfo(sRat(1, 2), None, None)], "use exact dddy"),
([PTInfo(0, None, None),
PTInfo(sRat(1, 2), None, None)], "use exact Fddy"),
]
for exact in exacts:
for p in exact_predictors:
yield check_exact_predictors, exact, p
def test_exact_predictors():
def check_exact_predictors(exact, predictor):
p1_func, y_np1_p1_expr = \
generate_predictor_scheme(*predictor, symbolic=exact)
# LTE for IMR: just natural lte:
y_np1_imr_expr = y_np1_exact - imr_lte(Sdts[0], Sdddynph, SFddynph)
# Generate another predictor
p2_func, y_np1_p2_expr = \
generate_predictor_scheme([PTInfo(sRat(1, 2), None, "imr"),
PTInfo(2, "corr_val", None),
PTInfo(3, "corr_val", None)], "ibdf2")
A = system2matrix([y_np1_p1_expr, y_np1_p2_expr, y_np1_imr_expr],
[Sdddynph, SFddynph, y_np1_exact])
# Solve for dddy and Fddy:
x = A.inv()
dddy_symb = sum([y_est * xi.factor() for xi, y_est in
zip(x.row(0), [y_np1_p1, y_np1_p2, y_np1_imr])])
Fddy_symb = sum([y_est * xi.factor() for xi, y_est in
zip(x.row(1), [y_np1_p1, y_np1_p2, y_np1_imr])])
# Check we got the right matrix and the right formulae
if predictor[1] == "use exact dddy":
assert A.row(0) == sympy.Matrix([[1, 0, 0]]).row(0)
utils.assert_sym_eq(dddy_symb.simplify(), y_np1_p1)
elif predictor[1] == "use exact Fddy":
assert A.row(0) == sympy.Matrix([[0, 1, 0]]).row(0)
utils.assert_sym_eq(Fddy_symb.simplify(), y_np1_p1)
else:
assert False
exacts = [sympy.exp(St),
sympy.sin(St),
]
exact_predictors = [
([PTInfo(0, None, None),
PTInfo(sRat(1, 2), None, None)],
"use exact dddy"),
([PTInfo(0, None, None),
PTInfo(sRat(1, 2), None, None)],
"use exact Fddy"),
]
for exact in exacts:
for p in exact_predictors:
yield check_exact_predictors, exact, p
def check_sum_dts(a, b):
# Check we can swap the sign
utils.assert_sym_eq(sum_dts(a, b), -sum_dts(b, a))
# Starting half a step earlier
utils.assert_sym_eq(sum_dts(a, b + sRat(1, 2)),
sRat(1, 2) * Sdts[int(b)] + sum_dts(a, b))
# Check that we can split it up
utils.assert_sym_eq(sum_dts(a, b),
sum_dts(a, b - 1) + sum_dts(b - 1, b))
def check_sum_dts(a, b):
# Check we can swap the sign
utils.assert_sym_eq(sum_dts(a, b), -sum_dts(b, a))
# Starting half a step earlier
utils.assert_sym_eq(sum_dts(a, b + sRat(1, 2)),
sRat(1, 2)*Sdts[int(b)] + sum_dts(a, b))
# Check that we can split it up
utils.assert_sym_eq(sum_dts(a, b),
sum_dts(a, b-1) + sum_dts(b-1, b))
def test_t_at_time_point():
tests = [
0,
1,
sRat(1, 2),
sRat(3, 2),
1 + sRat(1, 2),
sRat(4, 5),
]
ts = range(11)
for pt in tests:
utils.assert_almost_equal(t_at_time_point(ts, pt), 10 - pt)
def test_is_half_integer():
tests = [(0, False),
(1239012424481273, False),
(sRat(1, 2), True),
(1.5, True),
(sRat(5, 2), True),
(2.0 + sRat(1, 2), True),
(1.5 + 1e-15, False), # 1e-16 fails (gives true)
(1.51, False),
]
for t, result in tests:
assert is_half_integer(t) == result
def test_is_half_integer():
tests = [
(0, False),
(1239012424481273, False),
(sRat(1, 2), True),
(1.5, True),
(sRat(5, 2), True),
(2.0 + sRat(1, 2), True),
(1.5 + 1e-15, False), # 1e-16 fails (gives true)
(1.51, False),
]
for t, result in tests:
assert is_half_integer(t) == result
def rational_as_mixed(x):
"""Convert a rational number to (integer_part, remainder_as_fraction)
"""
assert is_rational(x) # Don't want to accidentally end up with floats
# in here as very long rationals!
x_int = int(x)
return x_int, sRat(x - x_int)
def rational_as_mixed(x):
"""Convert a rational number to (integer_part, remainder_as_fraction)
"""
assert is_rational(x) # Don't want to accidentally end up with floats
# in here as very long rationals!
x_int = int(x)
return x_int, sRat(x - x_int)
def test_const_step_explicit():
# Get explicit BDF2 (implicit midpoint)'s dydt approximation G&S pg 715
a = sympy.solve(-y0 + y1 + (1 + Delta0/Delta1)*Delta0*Dy1
- (Delta0/Delta1)**2*(y1 - y2), Dy1)
assert(len(a) == 1)
IMR_bdf_form = a[0].subs({k: Delta0 for k in dts})
orders = [1, 2, 3]
exacts = [(y0 - y1)/Delta0,
IMR_bdf_form,
#Hairer pg 364
(sRat(1, 3)*y0 + sRat(1, 2)*y1 - y2 + sRat(1, 6)*y3)/Delta0
]
for order, exact in zip(orders, exacts):
yield check_const_step, order, exact, 1
def test_const_step_explicit():
# Get explicit BDF2 (implicit midpoint)'s dydt approximation G&S pg 715
a = sympy.solve(-y0 + y1 + (1 + Delta0/Delta1)*Delta0*Dy1
- (Delta0/Delta1)**2*(y1 - y2), Dy1)
assert(len(a) == 1)
IMR_bdf_form = a[0].subs({k: Delta0 for k in dts})
orders = [1, 2, 3]
exacts = [(y0 - y1)/Delta0,
IMR_bdf_form,
#Hairer pg 364
(sRat(1, 3)*y0 + sRat(1, 2)*y1 - y2 + sRat(1, 6)*y3)/Delta0
]
for order, exact in zip(orders, exacts):
yield check_const_step, order, exact, 1
def test_imr_predictor_equivalence():
# Check the we generate imr's lte if we plug in the right times and
# approximations to ebdf2 (~explicit midpoint rule).
ynp1_imr = y_np1_exact - imr_lte(Sdts[0], Sdddynph, SFddynph)
_, ynp1_p1 = generate_predictor_scheme([
PTInfo(0, None, None),
PTInfo(sRat(1, 2), "bdf2 imr", "imr"),
PTInfo(1, "corr_val", None)
], "ebdf2")
utils.assert_sym_eq(ynp1_imr, ynp1_p1)
# Should be exactly the same with ebdf to calculate the y-est at the
# midpoint (because the midpoint value is ignored).
_, ynp1_p2 = generate_predictor_scheme([
PTInfo(0, None, None),
PTInfo(sRat(1, 2), "bdf3 imr", "imr"),
PTInfo(1, "corr_val", None)
], "ebdf2")
utils.assert_sym_eq(ynp1_imr, ynp1_p2)
def test_imr_predictor_equivalence():
# Check the we generate imr's lte if we plug in the right times and
# approximations to ebdf2 (~explicit midpoint rule).
ynp1_imr = y_np1_exact - imr_lte(Sdts[0], Sdddynph, SFddynph)
_, ynp1_p1 = generate_predictor_scheme([PTInfo(0, None, None),
PTInfo(
sRat(1, 2), "bdf2 imr", "imr"),
PTInfo(1, "corr_val", None)],
"ebdf2")
utils.assert_sym_eq(ynp1_imr, ynp1_p1)
# Should be exactly the same with ebdf to calculate the y-est at the
# midpoint (because the midpoint value is ignored).
_, ynp1_p2 = generate_predictor_scheme([PTInfo(0, None, None),
PTInfo(
sRat(1, 2), "bdf3 imr", "imr"),
PTInfo(1, "corr_val", None)],
"ebdf2")
utils.assert_sym_eq(ynp1_imr, ynp1_p2)
def test_const_step_implicit():
"""Check that the implicit methods are correct for fixed step size by
comparison with Hairer et. al. 1991 pg 366.
"""
exacts = [(y0 - y1)/Delta0,
(sRat(3, 2)*y0 - 2*y1 + sRat(1, 2)*y2)/Delta0,
(sRat(11, 6)*y0 - 3*y1 + sRat(3, 2)*y2 - sRat(1, 3)*y3)/Delta0,
(sRat(25, 12)*y0 - 4*y1 + 3*y2 - sRat(4, 3)*y3 + sRat(1, 4)*y4)/Delta0]
orders = [1, 2, 3, 4]
for order, exact in zip(orders, exacts):
yield check_const_step, order, exact, 0
def test_const_step_implicit():
"""Check that the implicit methods are correct for fixed step size by
comparison with Hairer et. al. 1991 pg 366.
"""
exacts = [(y0 - y1)/Delta0,
(sRat(3, 2)*y0 - 2*y1 + sRat(1, 2)*y2)/Delta0,
(sRat(11, 6)*y0 - 3*y1 + sRat(3, 2)*y2 - sRat(1, 3)*y3)/Delta0,
(sRat(25, 12)*y0 - 4*y1 + 3*y2 - sRat(4, 3)*y3 + sRat(1, 4)*y4)/Delta0]
orders = [1, 2, 3, 4]
for order, exact in zip(orders, exacts):
yield check_const_step, order, exact, 0
def test_sum_dts():
# Check a simple fractional case
utils.assert_sym_eq(sum_dts(sRat(1, 2), 1), Sdts[0] / 2)
# Check two numbers the same gives zero always
utils.assert_sym_eq(sum_dts(1, 1), 0)
utils.assert_sym_eq(sum_dts(0, 0), 0)
utils.assert_sym_eq(sum_dts(sRat(1, 2), sRat(1, 2)), 0)
def check_sum_dts(a, b):
# Check we can swap the sign
utils.assert_sym_eq(sum_dts(a, b), -sum_dts(b, a))
# Starting half a step earlier
utils.assert_sym_eq(sum_dts(a, b + sRat(1, 2)),
sRat(1, 2) * Sdts[int(b)] + sum_dts(a, b))
# Check that we can split it up
utils.assert_sym_eq(sum_dts(a, b),
sum_dts(a, b - 1) + sum_dts(b - 1, b))
# Make sure b>=1 or the last check will fail due to negative b.
cases = [
(0, 1),
(5, 8),
(sRat(1, 2), 3),
(sRat(2, 2), 3),
(sRat(3, 2), 3),
(0, sRat(9, 7)),
(sRat(3 / 4), sRat(9, 7)),
]
for a, b in cases:
yield check_sum_dts, a, b
def ptinfo2dyfunc(pt, dydt_func):
"""Construct a python function to calculate the approximation to dy
requested. If not using "exact" for dy_est the dydt_func can/should
be None.
"""
# Use the dy estimate from implicit midpoint rule (obviously only works
# at the midpoint).
if pt.dy_est == "imr":
if not is_half_integer(pt.time):
raise ValueError(
"imr dy approximation Only works for half integer " +
"points but given the point " + str(pt.time))
# Need to drop some of the later time values if time point is
# not the midpoint of 0 and 1 points (i.e. 1/2). Specifically, we
# need:
# 1/2 -> ts[:None]
# 3/2 -> ts[:-1]
# 5/2 -> ts[:-2]
if pt.time == sRat(1, 2):
# dy_func = ode.imr_dydt
def dy_func(ts, ys):
val = ode.imr_dydt(ts, ys)
print "imr f(t) =", val
return val
else:
x = -math.floor(pt.time)
def dy_func(ts, ys):
val = ode.imr_dydt(ts[:-int(math.floor(pt.time))],
ys[:-int(math.floor(pt.time))])
print "imr f(t) =", val
return val
elif pt.dy_est == "fd4":
def dy_func(ts, ys):
coeffs = [-1 / 12, -2 / 3, 0, 2 / 3, -1 / 12]
ys_used = ys[-6:-1]
assert len(ys) >= 6
# ??ds only for constant step! with variable steps we can put
# the 0 at a midpoint to reduce n-prev-steps needed.
return sp.dot(coeffs, ys_used)
# ??ds Use real dydt for exp
elif pt.dy_est == "exp test":
def dy_func(ts, ys):
return ys[-(pt.time + 1)]
# Use the provided f with known values to calculate dydt. Only for
# integer time points (i.e. where we already know y etc.) for now.
elif pt.dy_est == "exact":
assert dydt_func is not None
if is_integer(pt.time):
dy_func = lambda ts, ys: dydt_func(ts[pt.time])
else:
# Can't get y without additional approximations...
def dy_func(ts, ys):
val = dydt_func(t_at_time_point(ts, pt.time))
print "f(t) =", val
return val
# None = "this value at this point should not be used"
elif pt.dy_est is None:
dy_func = None
else:
raise ValueError("Unrecognised dy_est name in function construction " +
str(pt.dy_est))
return dy_func
def ptinfo2dyfunc(pt, dydt_func):
"""Construct a python function to calculate the approximation to dy
requested. If not using "exact" for dy_est the dydt_func can/should
be None.
"""
# Use the dy estimate from implicit midpoint rule (obviously only works
# at the midpoint).
if pt.dy_est == "imr":
if not is_half_integer(pt.time):
raise ValueError("imr dy approximation Only works for half integer "
+ "points but given the point " + str(pt.time))
# Need to drop some of the later time values if time point is
# not the midpoint of 0 and 1 points (i.e. 1/2). Specifically, we
# need:
# 1/2 -> ts[:None]
# 3/2 -> ts[:-1]
# 5/2 -> ts[:-2]
if pt.time == sRat(1, 2):
# dy_func = ode.imr_dydt
def dy_func(ts, ys):
val = ode.imr_dydt(ts, ys)
print "imr f(t) =", val
return val
else:
x = -math.floor(pt.time)
def dy_func(ts, ys):
val = ode.imr_dydt(ts[:-int(math.floor(pt.time))],
ys[:-int(math.floor(pt.time))])
print "imr f(t) =", val
return val
elif pt.dy_est == "fd4":
def dy_func(ts, ys):
coeffs = [-1/12, -2/3, 0, 2/3, -1/12]
ys_used = ys[-6:-1]
assert len(ys) >= 6
# ??ds only for constant step! with variable steps we can put
# the 0 at a midpoint to reduce n-prev-steps needed.
return sp.dot(coeffs, ys_used)
# ??ds Use real dydt for exp
elif pt.dy_est == "exp test":
def dy_func(ts, ys):
return ys[-(pt.time+1)]
# Use the provided f with known values to calculate dydt. Only for
# integer time points (i.e. where we already know y etc.) for now.
elif pt.dy_est == "exact":
assert dydt_func is not None
if is_integer(pt.time):
dy_func = lambda ts, ys: dydt_func(ts[pt.time])
else:
# Can't get y without additional approximations...
def dy_func(ts, ys):
val = dydt_func(t_at_time_point(ts, pt.time))
print "f(t) =", val
return val
# None = "this value at this point should not be used"
elif pt.dy_est is None:
dy_func = None
else:
raise ValueError(
"Unrecognised dy_est name in function construction " + str(pt.dy_est))
return dy_func
def bdf3_imr_ptinfos(pt):
# assert pt == sRat(1,2)
return [PTInfo(pt.time, None, "imr"),
PTInfo(pt.time + sRat(1, 2), "corr_val", None),
PTInfo(pt.time + sRat(1, 2) + 1, "corr_val", None),
PTInfo(pt.time + sRat(1, 2) + 2, "corr_val", None)]
def test_t_at_time_point():
tests = [0, 1, sRat(1, 2), sRat(3, 2), 1 + sRat(1, 2), sRat(4, 5), ]
ts = range(11)
for pt in tests:
utils.assert_almost_equal(t_at_time_point(ts, pt), 10 - pt)
def is_half_integer(x):
return x == (math.floor(x) + float(sRat(1, 2)))
def is_half_integer(x):
return x == (math.floor(x) + float(sRat(1, 2)))
