def constraint_consistency_check(make_element_func,
rdm,
ntests=50,
dt=1e-3,
nodal_range=None,
strain_range=None):
if nodal_range is None:
nodal_range = 1.0
if strain_range is None:
strain_range = nodal_range
EXPECTED_DECIMAL = -int(np.log10(min(nodal_range, 2 * strain_range) * dt))
if rdm is None:
rdm = np.random # default random number generator
for i in range(ntests):
# Make an element with random parameters
el = make_element_func(rdm)
# Initialise with random conditions
set_random_state(el, rdm, nodal_range, strain_range)
# el.xstrain[[0, 1, 2, 3, 4, 5]] = 0
# el.vstrain[[0, 1, 2, 3, 4]] = 0
# el.astrain[[0, 1, 2, 3, 4, 5]] = 0
# Save a copy of the current distal node state. NB need
# update_kinematics() to update vd & ad, not calc_kinematics()
el.update_kinematics()
rd1 = el.rd.copy()
Rd1 = el.Rd.copy()
vd1 = el.vd.copy()
ad1 = el.ad.copy()
# Peturb by small timestep & recalculate
el.vp[:] += el.ap[:] * dt
el.rp[:] += el.vp[:3] * dt
el.Rp[:] += dot(skewmat(el.vp[3:]), el.Rp) * dt
el.vstrain[:] += el.astrain[:] * dt
el.xstrain[:] += el.vstrain[:] * dt + el.astrain[:] * dt**2 / 2
el.update_kinematics()
# Estimate velocity and acceleration from change in
# position and velocity respectively
vd_est = (el.rd - rd1) / dt
Rd_dot_est = (el.Rd - Rd1) / dt
wd_est = _deskew(dot(Rd_dot_est, el.Rd.T))
ad_est = (el.vd - vd1) / dt
# Check calculated velocity against approx distal position
print(i)
assert_array_almost_equal(vd1[:3], vd_est, EXPECTED_DECIMAL - 1,
'Wrong distal translational velocity')
assert_array_almost_equal(vd1[3:], wd_est, EXPECTED_DECIMAL - 1,
'Wrong distal angular velocity')
assert_array_almost_equal(ad1, ad_est, EXPECTED_DECIMAL - 1,
'Wrong distal acceleration')
def constraint_consistency_check(make_element_func, rdm, ntests=50,
dt=1e-3, nodal_range=None, strain_range=None):
if nodal_range is None:
nodal_range = 1.0
if strain_range is None:
strain_range = nodal_range
EXPECTED_DECIMAL = -int(np.log10(min(nodal_range, 2*strain_range) * dt))
if rdm is None:
rdm = np.random # default random number generator
for i in range(ntests):
# Make an element with random parameters
el = make_element_func(rdm)
# Initialise with random conditions
set_random_state(el, rdm, nodal_range, strain_range)
# el.xstrain[[0, 1, 2, 3, 4, 5]] = 0
# el.vstrain[[0, 1, 2, 3, 4]] = 0
# el.astrain[[0, 1, 2, 3, 4, 5]] = 0
# Save a copy of the current distal node state. NB need
# update_kinematics() to update vd & ad, not calc_kinematics()
el.update_kinematics()
rd1 = el.rd.copy()
Rd1 = el.Rd.copy()
vd1 = el.vd.copy()
ad1 = el.ad.copy()
# Peturb by small timestep & recalculate
el.vp[:] += el.ap[:] * dt
el.rp[:] += el.vp[:3] * dt
el.Rp[:] += dot(skewmat(el.vp[3:]), el.Rp) * dt
el.vstrain[:] += el.astrain[:] * dt
el.xstrain[:] += el.vstrain[:] * dt + el.astrain[:] * dt**2 / 2
el.update_kinematics()
# Estimate velocity and acceleration from change in
# position and velocity respectively
vd_est = (el.rd - rd1) / dt
Rd_dot_est = (el.Rd - Rd1) / dt
wd_est = _deskew(dot(Rd_dot_est, el.Rd.T))
ad_est = (el.vd - vd1) / dt
# Check calculated velocity against approx distal position
print(i)
assert_array_almost_equal(vd1[:3], vd_est, EXPECTED_DECIMAL - 1,
'Wrong distal translational velocity')
assert_array_almost_equal(vd1[3:], wd_est, EXPECTED_DECIMAL - 1,
'Wrong distal angular velocity')
assert_array_almost_equal(ad1, ad_est, EXPECTED_DECIMAL - 1,
'Wrong distal acceleration')
def test_distal_node_position(self):
NTESTS = 5
for i in range(NTESTS):
el = _make_random_element(self.rdm)
set_random_state(el, self.rdm, 1)
last_node_x0 = [el._params['length'], 0, 0]
Xd = dot(el.shapes[-6:], el.xstrain)
last_node_defl = Xd[0:3]
last_node_rotn = Xd[3:6]
el.calc_distal_pos()
assert_aae(el.rd,
el.rp + dot(el.Rp, last_node_x0 + last_node_defl))
assert_aae(el.Rd, dot(el.Rp, np.eye(3) + skewmat(last_node_rotn)))
def test_distal_node_position(self):
NTESTS = 5
for i in range(NTESTS):
el = _make_random_element(self.rdm)
set_random_state(el, self.rdm, 1)
last_node_x0 = [el._params['length'], 0, 0]
Xd = dot(el.shapes[-6:], el.xstrain)
last_node_defl = Xd[0:3]
last_node_rotn = Xd[3:6]
el.calc_distal_pos()
assert_aae(el.rd,
el.rp + dot(el.Rp, last_node_x0 + last_node_defl))
assert_aae(el.Rd,
dot(el.Rp, np.eye(3) + skewmat(last_node_rotn)))
def test_deskew():
rdm = np.random.RandomState(123456789)
for i in range(10):
vec = rdm.uniform(-10, 10, 3)
assert_array_equal(vec, _deskew(skewmat(vec)))
def test_deskew():
rdm = np.random.RandomState(123456789)
for i in range(10):
vec = rdm.uniform(-10, 10, 3)
assert_array_equal(vec, _deskew(skewmat(vec)))