def abc_work_function_impulsive_k(c):
u0 = [1, 1, 0, 0, 0, 0, 0, 0]
# axisymmetric expansion
tspan = np.linspace(0, 1.6 / np.abs(g.Strain.axi_exp[0]), 200)
Ynke = odeint(rans.rans_impulsive, u0, tspan, args=(c, g.Strain.axi_exp), atol=1e-8, mxstep=200)
sum_stat_k1 = calc_sum_stat(np.abs(g.Strain.axi_exp[0]) * tspan, Ynke[:, 0], g.Truth.axi_exp_k[:, 0])
# axisymmetric contraction
tspan = np.linspace(0, 1.6 / np.abs(g.Strain.axi_con[0]), 200)
Ynke = odeint(rans.rans_impulsive, u0, tspan, args=(c, g.Strain.axi_con), atol=1e-8, mxstep=200)
sum_stat_k2 = calc_sum_stat(np.abs(g.Strain.axi_con[0]) * tspan, Ynke[:, 0], g.Truth.axi_con_k[:, 0])
# pure shear
tspan = np.linspace(0, 5.2 / (2 * g.Strain.pure_shear[3]), 200)
Ynke = odeint(rans.rans_impulsive, u0, tspan, args=(c, g.Strain.pure_shear), atol=1e-8, mxstep=200)
sum_stat_k3 = calc_sum_stat(2 * g.Strain.pure_shear[3] * tspan, Ynke[:, 0], g.Truth.shear_k[:, 0])
# plane strain
tspan = np.linspace(0, 1.6 / g.Strain.plane_strain[0], 200)
Ynke = odeint(rans.rans_impulsive, u0, tspan, args=(c, g.Strain.plane_strain), atol=1e-8, mxstep=200)
sum_stat_k4 = calc_sum_stat(np.abs(g.Strain.plane_strain[0]) * tspan, Ynke[:, 0], g.Truth.plane_k[:, 0])
sum_stat = np.hstack((sum_stat_k1, sum_stat_k2, sum_stat_k3, sum_stat_k4)) / g.Truth.norm
# noise = np.random.normal(loc=0.0, scale=0.0008, size=len(sum_stat))
err = calc_err(sum_stat, g.Truth.sumstat_true)
result = np.hstack((c, sum_stat, err)).tolist()
return result
def abc_work_function_decay(c):
u0 = [1, 1, 0.36, -0.08, -0.28, 0, 0, 0]
tspan = np.linspace(0, 0.3, 200)
Ynke = odeint(rans.rans_decay, u0, tspan, args=(c, ), atol=1e-8, mxstep=200)
sum_stat1 = calc_sum_stat(tspan, Ynke[:, 2], g.Truth.decay_a11[:, 0])
sum_stat2 = calc_sum_stat(tspan, Ynke[:, 3], g.Truth.decay_a22[:, 0])
sum_stat3 = calc_sum_stat(tspan, Ynke[:, 4], g.Truth.decay_a33[:, 0])
sum_stat = np.hstack((sum_stat1, sum_stat2, sum_stat3))
err = calc_err(sum_stat, g.Truth.sumstat_true)
result = np.hstack((c, sum_stat, err)).tolist()
return result
def abc_work_function_impulsive_periodic(c):
u0 = [1, 1, 0, 0, 0, 0, 0, 0]
# axisymmetric expansion
tspan = np.linspace(0, 1.6 / np.abs(g.Strain.axi_exp[0]), 200)
Ynke = odeint(rans.rans_impulsive, u0, tspan, args=(c, g.Strain.axi_exp), atol=1e-8, mxstep=200)
sum_stat_k1 = calc_sum_stat(np.abs(g.Strain.axi_exp[0]) * tspan, Ynke[:, 0], g.Truth.axi_exp_k[:, 0])
# sum_stat_b1 = calc_sum_stat(np.abs(g.Strain_axi_exp[0]) * tspan, Ynke[:, 2], g.Truth.axi_exp_b[:, 0])
# axisymmetric contraction
tspan = np.linspace(0, 1.6 / np.abs(g.Strain.axi_con[0]), 200)
Ynke = odeint(rans.rans_impulsive, u0, tspan, args=(c, g.Strain.axi_con), atol=1e-8, mxstep=200)
sum_stat_k2 = calc_sum_stat(np.abs(g.Strain.axi_con[0]) * tspan, Ynke[:, 0], g.Truth.axi_con_k[:, 0])
# sum_stat_b2 = calc_sum_stat(np.abs(g.Strain_axi_con[0]) * tspan, Ynke[:, 2], g.Truth.axi_con_b[:, 0])
# pure shear
tspan = np.linspace(0, 5.2 / (2*g.Strain.pure_shear[3]), 200)
Ynke = odeint(rans.rans_impulsive, u0, tspan, args=(c, g.Strain.pure_shear), atol=1e-8, mxstep=200)
sum_stat_k3 = calc_sum_stat(2 * g.Strain.pure_shear[3] * tspan, Ynke[:, 0], g.Truth.shear_k[:, 0])
# plane strain
tspan = np.linspace(0, 1.6 / g.Strain.plane_strain[0], 200)
Ynke = odeint(rans.rans_impulsive, u0, tspan, args=(c, g.Strain.plane_strain), atol=1e-8, mxstep=200)
sum_stat_k4 = calc_sum_stat(np.abs(g.Strain.plane_strain[0]) * tspan, Ynke[:, 0], g.Truth.plane_k[:, 0])
time1 = time()
s0 = 3.3
beta = [0.125, 0.25, 0.5, 0.75, 1]
u0 = [1, 1, 0, 0, 0, 0, 0, 0]
# Periodic shear(five different frequencies)
tspan = np.linspace(0, 50/s0, 500)
sum_stat = []
for i in range(5):
Ynke = odeint(rans.rans_periodic, u0, tspan, args=(c, s0, beta[i], StrainTensor.periodic_strain),
atol=1e-8, mxstep=200)
ss_new = calc_sum_stat(s0 * tspan, take_safe_log10(Ynke[:, 0]), g.Truth.periodic_k[i][:, 0])
sum_stat = np.hstack((sum_stat, ss_new))
sum_stat = np.hstack((sum_stat_k1, sum_stat_k2, sum_stat_k3, sum_stat_k4, sum_stat)) / g.Truth.norm
err = calc_err(sum_stat, g.Truth.sumstat_true)
result = np.hstack((c, err)).tolist()
time2 = time()
if np.isnan(err) or np.isinf(err):
print(time2-time1, 'nan/inf = {}'.format(np.isnan(err) or np.isinf(err)), err)
return result
def abc_work_function_strain_relax(c):
u0 = [1, 1, 0.36, -0.08, -0.28, 0, 0, 0]
# strain-relaxation
tspan = np.linspace(0.0775, 0.953, 500)
Ynke = odeint(rans.rans_strain_relax, u0, tspan, args=(c, g.Strain.strain_relax), atol=1e-8, mxstep=200)
sum_stat = calc_sum_stat(tspan, Ynke[:, 2], g.Truth.strain_relax_a11[:, 0])
err = calc_err(sum_stat, g.Truth.sumstat_true)
result = np.hstack((c, sum_stat, err)).tolist()
return result
def abc_work_function_validation_nominal(c):
time1 = time()
s0 = 3.3
beta = 1
u0 = [1, 1, 0, 0, 0, 0, 0, 0]
# Periodic shear(five different frequencies)
tspan = np.linspace(0, 50/s0, 500)
Ynke = odeint(rans.rans_periodic, u0, tspan, args=(c, s0, beta, StrainTensor.periodic_strain),
atol=1e-8, mxstep=200)
sum_stat = calc_sum_stat(s0 * tspan, Ynke[:, 0], g.Truth.validation_nominal[:, 0])
err = calc_err(np.log10(sum_stat), g.Truth.sumstat_true)
result = np.hstack((c, sum_stat, err)).tolist()
time2 = time()
if np.isnan(err) or np.isinf(err):
print(time2-time1, 'nan/inf = {}'.format(np.isnan(err) or np.isinf(err)), err)
return result
def abc_work_function_periodic(c):
time1 = time()
s0 = 3.3
beta = [0.125, 0.25, 0.5, 0.75, 1]
u0 = [1, 1, 0, 0, 0, 0, 0, 0]
# Periodic shear(five different frequencies)
tspan = np.linspace(0, 50/s0, 500)
sum_stat = []
for i in range(5):
Ynke = odeint(rans.rans_periodic, u0, tspan, args=(c, s0, beta[i], StrainTensor.periodic_strain),
atol=1e-8, mxstep=200)
ss_new = calc_sum_stat(s0 * tspan, take_safe_log10(Ynke[:, 0]), g.Truth.periodic_k[i][:, 0])
sum_stat = np.hstack((sum_stat, ss_new))
err = calc_err(sum_stat, g.Truth.sumstat_true)
result = np.hstack((c, err)).tolist()
time2 = time()
if np.isnan(err) or np.isinf(err):
print(time2-time1, 'nan/inf = {}'.format(np.isnan(err) or np.isinf(err)), err)
return result
def abc_work_function_impulsive_decay(c):
u0 = [1, 1, 0, 0, 0, 0, 0, 0]
# axisymmetric expansion
tspan = np.linspace(0, 1.6 / np.abs(g.Strain.axi_exp[0]), 200)
Ynke = odeint(rans.rans_impulsive, u0, tspan, args=(c, g.Strain.axi_exp), atol=1e-8, mxstep=200)
sum_stat_k1 = calc_sum_stat(np.abs(g.Strain.axi_exp[0]) * tspan, Ynke[:, 0], g.Truth.axi_exp_k[:, 0])
sum_stat_a1 = calc_sum_stat(np.abs(g.Strain.axi_exp[0]) * tspan, Ynke[:, 2], g.Truth.axi_exp_a[:, 0])
# axisymmetric contraction
tspan = np.linspace(0, 1.6 / np.abs(g.Strain.axi_con[0]), 200)
Ynke = odeint(rans.rans_impulsive, u0, tspan, args=(c, g.Strain.axi_con), atol=1e-8, mxstep=200)
sum_stat_k2 = calc_sum_stat(np.abs(g.Strain.axi_con[0]) * tspan, Ynke[:, 0], g.Truth.axi_con_k[:, 0])
sum_stat_a2 = calc_sum_stat(np.abs(g.Strain.axi_con[0]) * tspan, Ynke[:, 2], g.Truth.axi_con_a[:, 0])
# pure shear
tspan = np.linspace(0, 5.2 / (2*g.Strain.pure_shear[3]), 200)
Ynke = odeint(rans.rans_impulsive, u0, tspan, args=(c, g.Strain.pure_shear), atol=1e-8, mxstep=200)
sum_stat_k3 = calc_sum_stat(2 * g.Strain.pure_shear[3] * tspan, Ynke[:, 0], g.Truth.shear_k[:, 0])
# plane strain
tspan = np.linspace(0, 1.6 / g.Strain.plane_strain[0], 200)
Ynke = odeint(rans.rans_impulsive, u0, tspan, args=(c, g.Strain.plane_strain), atol=1e-8, mxstep=200)
sum_stat_k4 = calc_sum_stat(np.abs(g.Strain.plane_strain[0]) * tspan, Ynke[:, 0], g.Truth.plane_k[:, 0])
sum_stat_a3 = calc_sum_stat(np.abs(g.Strain.plane_strain[0]) * tspan, Ynke[:, 2], g.Truth.plane_a11[:, 0])
sum_stat_a4 = calc_sum_stat(np.abs(g.Strain.plane_strain[0]) * tspan, Ynke[:, 2], g.Truth.plane_a22[:, 0])
u0 = [1, 1, 0.36, -0.08, -0.28, 0, 0, 0]
tspan = np.linspace(0, 0.3, 200)
Ynke = odeint(rans.rans_decay, u0, tspan, args=(c, ), atol=1e-8, mxstep=200)
sum_stat1 = calc_sum_stat(tspan, Ynke[:, 2], g.Truth.decay_a11[:, 0])
sum_stat2 = calc_sum_stat(tspan, Ynke[:, 3], g.Truth.decay_a22[:, 0])
sum_stat3 = calc_sum_stat(tspan, Ynke[:, 4], g.Truth.decay_a33[:, 0])
sum_stat = np.hstack((sum_stat_k1, sum_stat_k2, sum_stat_k3, sum_stat_k4,
sum_stat_a1, sum_stat_a2, sum_stat_a3, sum_stat_a4,
sum_stat1, sum_stat2, sum_stat3)) / g.Truth.norm
err = calc_err(sum_stat, g.Truth.sumstat_true)
result = np.hstack((c, sum_stat, err)).tolist()
return result