def __solver__(self, p):
p.kernelIterFuncs.pop(SMALL_DELTA_X, None)
p.kernelIterFuncs.pop(SMALL_DELTA_F, None)
xf, ff, code, opt_struct = ASA.asa(p.f, p.x0, p.lb, p.ub, full_output=True)
if p.istop == 0: p.istop = 1000
def __solver__(self, p):
p.kernelIterFuncs.pop(SMALL_DELTA_X, None)
p.kernelIterFuncs.pop(SMALL_DELTA_F, None)
xf, ff, code, opt_struct = ASA.asa(p.f,
p.x0,
p.lb,
p.ub,
full_output=True)
if p.istop == 0: p.istop = 1000
s, t, c = .2, .05, .15
d = np.array([1., 1000., 10., 100.])
def cost(x):
#raise ValueError("rrr")
k = np.rint(x/s)
r = np.fabs(k*s-x)
p = np.sign(k)
q = np.where(r<t, c*(p*p*k*s-p*t)**2, x**2)
return (d*q).sum()
x0 = np.array([999., -1007, 1001, -903])
xmax = 1e4*np.ones_like(x0)
print asa.asa(cost, x0, -xmax, xmax, full_output=True)
# simple leak check
#while True:
# try:
# asa.asa(cost, x0, -xmax, xmax, full_output=True)
# except Exception:
# pass
d = np.arange(5.)
def cost(x):
return -(np.sin(1./(x-d))*np.exp(-(x-d)**2)).sum()
x0 = -d
xmax = 1e1*np.ones_like(x0)
d = np.array([1., 1000., 10., 100.])
def cost(x):
#raise ValueError("rrr")
k = np.rint(x / s)
r = np.fabs(k * s - x)
p = np.sign(k)
q = np.where(r < t, c * (k * s - p * t)**2, x**2)
return (d * q).sum()
x0 = np.array([999., -1007, 1001, -903])
xmax = 1e4 * np.ones_like(x0)
print asa.asa(cost, x0, -xmax, xmax, full_output=True)
# simple leak check
#while True:
# try:
# asa.asa(cost, x0, -xmax, xmax, full_output=True)
# except Exception:
# pass
d = np.arange(5.)
def cost(x):
return -(np.sin(1. / (x - d)) * np.exp(-(x - d)**2)).sum()
