def t0():
# Incoming quark mom
p1 = 0
p2 = 0
p3 = Symbol('p', real=True)
p0 = p3
p = tensor1d([p0,p1,p2,p3],'mu')
# Transverse (w.r.t p) component k_t of daughter momenta (spacelike)
kt1 = 0
kt2 = Symbol('kt', real=True)
kt3 = 0
kt0 = 0
kt = tensor1d([kt0,kt1,kt2,kt3],'mu')
# Auxiliary light-like vector transverse to k_t
n1 = Symbol('n', real=True)
n2 = 0
n3 = 0
n0 = n1
n = tensor1d([n0,n1,n2,n3],'mu')
# Longitudinal momentum fraction
z = tensor([Symbol('z', real=True)], None)
omz = tensor([1.0 - Symbol('z', real=True)], None)
# Outgoing quark momentum
pa = z*p + kt - tensor([(kt2**2)], None)/ z* n/(2.0*n1*p3)
# Outgoing gluon momentum
pb = omz*p - kt - tensor([(kt2**2)], None)/omz* n/(2.0*n1*p3)
# Outgoing gluon reference mom
g1 = Symbol('g1', real=True)
g2 = Symbol('g2', real=True)
g3 = Symbol('g3', real=True)
g0 = sqrt(g1**2+g2**2+g3**2)
m2 = (uminusbar(pb._array[0]._array[0],
pb._array[1]._array[0],
pb._array[2]._array[0],
pb._array[3]._array[0],'a')*
Gamma('mu','a','b')*
uplus(p0,p1,p2,p3,'b')*mink_metric('mu','nu')*
conjugate_tensor1d(epsilonplus(pa._array[0]._array[0],
pa._array[1]._array[0],
pa._array[2]._array[0],
pa._array[3]._array[0], 'nu', g0,g1,g2,g3)))
M2 = m2._array[0]*cgt(m2._array[0])
m2 = m2._array[0].simplify()
from sympy import Q, refine, Abs
from sympy.assumptions.refine import refine_abs
from IPython import embed
m2 =refine_abs(m2, Q.real(z._array[0]))
embed()
def totensor(self):
sz = self.tsize
order = numpy.concatenate((self.rindices, self.cindices))
order = order.tolist()
data = self.data.reshape(tools.getelts(sz, order))
data = tensor.tensor(data).ipermute(order).data
return tensor.tensor(data, sz)
def totensor(self):
sz = self.tsize;
order = numpy.concatenate((self.rindices, self.cindices));
order = order.tolist();
data = self.data.reshape(tools.getelts(sz,order));
data = tensor.tensor(data).ipermute(order).data;
return tensor.tensor(data, sz);
def boardcast(data1,data2):
data1=ts.tensor(data1)
data2=ts.tensor(data2)
try:
return expand(data1,data2.size()),data2
except:
try:
return data1,expand(data2,data1.size())
except:
raise ValueError('can not boardcast with size %s and %s'%(data1.size(),data2.size()))
def convolve(self, h, idx):
obj = t.tensor(self)
if isinstance(h, t.tensor): h = h.arr
N = obj.len(idx)
obj.transpose([obj.index(idx), 0])
z = zeros(obj.arr.shape)
obj.arr = append(obj.arr, z, axis=0)
obj.transpose([obj.index(idx), -1],opt=True)
X = fft.rfft(obj.arr)
H = fft.rfft(append(h, zeros(2*N - h.size)))
return t.tensor(fft.irfft(X*H)[0:N], obj.idx, obj.ud)
def __init__(self,i,n):
# our inde scheme starts at 0, so subtract 1 from n
# when naming the momenta
k = n-1
il = lorentz_key(i) if (i<0) else i
p_str = "p{0}[{1}]"
array = [tensor([c_variable(p_str.format(k,vect_gauge_dict[0]))], None),
tensor([c_variable(p_str.format(k,vect_gauge_dict[1]))], None),
tensor([c_variable(p_str.format(k,vect_gauge_dict[2]))], None),
tensor([c_variable(p_str.format(k,vect_gauge_dict[3]))], None)]
super(P,self).__init__(array,il)
def __init__(self, i, n):
# our inde scheme starts at 0, so subtract 1 from n
# when naming the momenta
k = n - 1
il = lorentz_key(i) if (i < 0) else i
p_str = "p{0}[{1}]"
array = [
tensor([c_variable(p_str.format(k, vect_gauge_dict[0]))], None),
tensor([c_variable(p_str.format(k, vect_gauge_dict[1]))], None),
tensor([c_variable(p_str.format(k, vect_gauge_dict[2]))], None),
tensor([c_variable(p_str.format(k, vect_gauge_dict[3]))], None)
]
super(P, self).__init__(array, il)
def __init__(self, i, n):
# our inde scheme starts at 0, so subtract 1 from n
# when naming the momenta
k = n - 1
il = lorentz_key(i) if (i < 0) else i
p_str = "p{0}{1}"
array = [
tensor([parse_expr(p_str.format(k, 0))], None),
tensor([parse_expr(p_str.format(k, 1))], None),
tensor([parse_expr(p_str.format(k, 2))], None),
tensor([parse_expr(p_str.format(k, 3))], None)
]
super(P, self).__init__(array, il)
def compute_materialtangent(self):
# =============================================================================================
#%% Compute error on material tangent operator by finite (central) difference of stress
# =============================================================================================
d = self.FE_model.dim
DC = 1.e-5
TOLMAX = 1.e-3
PRECISION = 1.e-16
printerror = False
e = 1 # element
KerrMax = 1e-30
for e in range(self.FE_model.nElements()):
Ed = self.E[e].reshape(d, d) # Lagrange deformation
Cd = 2 * Ed + tensor.I(d) # right Cauchy-Green tensor
Sp = tensor.tensor(d) # S(C+DC/2) Piola Kirchhoff II tensor "plus"
# S(C-DC/2) Piola Kirchhoff II tensor "minus"
Sm = tensor.tensor(d)
S = self.FE_model.material.stress(Cd)
# Lagrangian tangent operator 2*dS/dC
M = self.FE_model.material.stiffness(Cd)
for k in range(d):
for l in range(d):
Cd[k, l] += DC / 2.
Sp = self.FE_model.material.stress(Cd)
Cd[k, l] -= DC
Sm = self.FE_model.material.stress(Cd)
Cd[k, l] += DC / 2. # come back to before perturbation
for i in range(d):
for j in range(d):
Knum = (Sp[i, j] - Sm[i, j] + Sp[j, i] -
Sm[j, i]) / DC # 2*(Sp-Sm + Sp-Sm)/2/DC
# K_ijkl = ( S_ij(C_kl+DC/2) - S_ji(C_kl-DC/2) ) SYM
if (np.fabs(M[i, j, k, l]) > 1.0):
Ktest = np.fabs(M[i, j, k, l])
else:
Ktest = 1.0
Kerr = np.fabs(M[i, j, k, l] - Knum) / Ktest
if (Kerr > KerrMax):
KerrMax = Kerr
if (printerror) and (Kerr > TOLMAX) and (np.fabs(Knum) > PRECISION):
print('%d %d %d %d %13.6e %13.6e %13.6e' %
(i, j, k, l, M[i, j, k, l], Knum, Kerr))
# print('Maximal error = %13.6e' % KerrMax)
if KerrMax > TOLMAX:
msg = "Material tangent not valided: KerrMax = %.2e" % KerrMax
raise Exception(msg)
def stress(self, C):
"""
Compute 2nd Piola-Kirchhoff stress
PK2 = 2*dphi/dC
= lambda * tr(E) * I + 2 * mu * E
"""
dim = len(C)
PK2 = tensor.tensor(dim)
lamda = self.get1LAME()
mu = self.get2LAME()
E = tensor.tensor(dim)
E = 0.5 * (C - tensor.I(dim))
II = tensor.I(dim)
PK2 = lamda * tensor.trace(E) * II + 2 * mu * E
return PK2
def m_1Inner(b):
b = b[1].clean()
answer = tensor()
if b.degree() != 0:
for i in range(b.degree()):
answer += b.coefficient * pureTensor([b[0:i],b[i],b[i+1:]])
return answer
def ctor(verbose):
dat = numpy.arange(24).reshape([2, 3, 4])
t = tensor.tensor(dat)
print t
if (verbose):
obj = tenmat.tenmat(t, [1, 0])
print obj
print obj.copy()
dat = dat.reshape([4, 6])
t = tensor.tensor(dat)
if (verbose):
obj = tenmat.tenmat(t, [0], [1], [4, 6])
print obj
def getStrainEulerAlmansi(self, n):
""" Return average of EA strain at all integration points of element n"""
avg = tensor.tensor(self.dim)
for tens in self.elems[n].E_EA:
avg += tens
avg /= self.elems[n].getNIntPts()
return avg
def mov_ave(self, wid, dx, idx, fs):
f = fft.rfftfreq(self.len(idx), float(fs))
L = int(float(wid) / float(dx))
H = ((sin(pi * f * L / fs) / sin(pi * f / fs)) * exp(-1j * pi * f * (L-1) / fs)) / L
H[0] = 1.
X = self.rfft(idx).arr * H
return t.tensor(fft.irfft(X), self.idx, self.ud)
def ctor(verbose):
dat = numpy.arange(24).reshape([2,3,4]);
t = tensor.tensor(dat);
print t;
if (verbose):
obj = tenmat.tenmat(t, [1,0]);
print obj;
print obj.copy();
dat = dat.reshape([4,6]);
t = tensor.tensor(dat);
if (verbose):
obj = tenmat.tenmat(t, [0], [1], [4,6]);
print obj;
def tensorProduct(self,other):
"""Add other as a component(s) on the right of self. e.g. (a|b).tensorProduct(c) = a|b|c"""
if type(other) in [relation.relation,doublyDefined.doublyDefined,coefficient.coefficient, str, float, int, monomial.monomial]:
return self.tensorProduct(pureTensor(other))
if not isinstance(other,pureTensor):
return reduce(lambda x,y: x+y, [self.tensorProduct(z) for z in other], tensor.tensor())
return pureTensor(self.monomials + other.monomials, self.coefficient * other.coefficient)
def i_3Inner(pT):
answer = tensor()
doublyDefined = pT[1]
for generator, rel in doublyDefined.leftHandRepresentation:
rightHandSide = generator * i_2(pureTensor([1,rel,1]),alg)
answer = answer + pureTensor(1).tensorProduct(rightHandSide)
return answer
def generateNeighborInteractions(interaction_vector):
"""
Generate the interaction between two sites
interaction_vector (list): A list of length (num_qubits) with 1 on index i if the
site is a constituent of the interaction. For example,
for interaction between site 1 and 2 in a 4 qubit system,
the vector would be,[1, 1, 0, 0] and for stie 1 and 3, [1, 0, 1, 0]
"""
pauli_z = tensor(2, {(0, 0): 1, (1, 1): -1})
id = tensor(2, {(0, 0): 1, (1, 1): 1})
_tensor = pauli_z if interaction_vector[0] == 1 else id
for site_i in interaction_vector[1:]:
field = pauli_z if site_i == 1 else id
_tensor = _tensor.tensor_prod(field)
return _tensor
def i_2Inner(pT):
answer = tensor()
rel = pT[1]
for term in rel.leadingMonomial:
answer = answer + term.coefficient * pureTensor((1,term[0],term[1],1))
for term in rel.lowerOrderTerms:
answer = answer - term.coefficient * pureTensor((1,term[0],term[1],1))
return answer
def k_4Inner(pT):
answer= tensor()
doublyDefined = pT[1]
for generator, rel in doublyDefined.leftHandRepresentation:
answer = answer + pureTensor((generator,rel,1)).clean()
for rel, generator in doublyDefined.rightHandRepresentation:
answer = answer - pureTensor((1,rel,generator)).clean()
return answer
def totensorTests(verbose):
dat = numpy.arange(24).reshape([2,3,4]);
t = tensor.tensor(dat);
obj = tenmat.tenmat(t,[2,1]);
if(verbose):
print obj;
print obj.totensor();
def totensorTests(verbose):
dat = numpy.arange(24).reshape([2, 3, 4])
t = tensor.tensor(dat)
obj = tenmat.tenmat(t, [2, 1])
if (verbose):
print obj
print obj.totensor()
def getStrainGreenLagrange(self, n):
"""
Return average of GL strain at all integration points of element n
"""
avg = tensor.tensor(self.dim)
for tens in self.elems[n].E_GL:
avg += tens
avg /= self.elems[n].getNIntPts()
return avg
def getStrainHencky(self, n):
"""
Return average of hencky strain at all integration points of element n
"""
avg = tensor.tensor(self.dim)
for tens in self.elems[n].hencky:
avg += tens
avg /= self.elems[n].getNIntPts()
return avg
def getStressPK1(self, n):
"""
Return average of PK1 stress at all integration points of element n
"""
avg = tensor.tensor(self.dim)
for tens in self.elems[n].PK1:
avg += tens
avg /= self.elems[n].getNIntPts()
return avg
def getDeformationGradient(self, n):
"""
Return average of deformation gradient at all integration points of element n
"""
avg = tensor.tensor(self.dim)
for tens in self.elems[n].F:
avg += tens
avg /= self.elems[n].getNIntPts()
return avg
def listOfPolysToTensors(ps):
ps = iter(ps)
pureTensors = [pureTensor.pureTensor(mono) for mono in next(ps)]
for poly in ps:
newPureTensors = []
for mono in poly:
newPureTensors.extend([x.tensorProduct(mono) for x in pureTensors])
pureTensors = newPureTensors
return tensor.tensor(pureTensors)
def getStressCauchy(self, n):
"""
Return average of cauchy stress at all integration points of element n
"""
avg = tensor.tensor(self.dim)
for tens in self.elems[n].sigma:
avg += tens
avg /= self.elems[n].getNIntPts()
return avg
def ctorTests(verbose):
arr = numpy.arange(24).reshape([2,3,4]);
A = numpy.array([[1,2],[3,4],[5,6]]);
B = numpy.array([[1,2,3],[4,5,6]]);
C = numpy.array([[1,2,3,4]]);
obj = ttensor.ttensor(tensor.tensor(arr), [A, B, C]);
print obj;
print obj.shape;
def mtimes(self, B):
tsize = numpy.hstack((self.tsize[self.rindices], B.tsize[B.cindices]))
data = tensor.tensor([])
C = tenmat2(data)
C.rindices = numpy.array(range(0, len(self.rindices)))
C.cindices = numpy.array(range(0, len(B.cindices))) + len(self.rindices)
C.data = numpy.dot(self.data, B.data)
C.tsize = tsize
return C
def ctorTests(verbose):
arr = numpy.arange(24).reshape([2, 3, 4])
A = numpy.array([[1, 2], [3, 4], [5, 6]])
B = numpy.array([[1, 2, 3], [4, 5, 6]])
C = numpy.array([[1, 2, 3, 4]])
obj = ttensor.ttensor(tensor.tensor(arr), [A, B, C])
print obj
print obj.shape
def totensor(self):
"""returns a new tensor object that contains the same values"""
temp = numpy.ndarray(self.shape);
temp.fill(0);
for i in range(0, len(self.vals)):
temp.put(tools.sub2ind(self.shape, self.subs[i])[0], self.vals[i][0]);
return tensor.tensor(temp, self.shape);
def tosptensor(verbose):
dat = numpy.array([1,2,3,4,5,6,7,8,9,10,0,0,13,14,15,16,17,18]);
siz = numpy.array([3,3,2]);
obj = tensor.tensor(dat, siz);
if(verbose == 1):
print obj;
print (obj.tosptensor());
def totensor(self):
"""returns a new tensor object that contains the same values"""
temp = numpy.ndarray(self.shape);
temp.fill(0);
for i in range(0, len(self.vals)):
temp.put(tools.sub2ind(self.shape, self.subs[i])[0], self.vals[i][0]);
return tensor.tensor(temp, self.shape);
def append(self, obj, idx):
## \brief Append a tensor to other tensor.
self.same_trans(obj)
self.transpose([idx,0])
obj.transpose([idx,0])
if self.idx != obj.idx:
raise ValueError, "Index is not much."
ret = t.tensor(append(self.arr, obj.arr, axis=0), self.idx, self.ud)
return ret
def form_factors_implementations(self, parameter_map):
evaluated_form_factors = {}
wrapped_parameter_map = {
k: tensor([sym_var(Symbol(v, complex=True))], None)
for k, v in parameter_map.items()
}
for k, v in self._form_factors.items():
evaluated_form_factors[k] = eval(v, globals(),
wrapped_parameter_map)
return evaluated_form_factors
def tosptensor(verbose):
dat = numpy.array(
[1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 0, 0, 13, 14, 15, 16, 17, 18])
siz = numpy.array([3, 3, 2])
obj = tensor.tensor(dat, siz)
if (verbose == 1):
print obj
print(obj.tosptensor())
def mov_grad(self, obj, wid, dx, idx, fs):
x = obj
y = self
xy = x * y
L = abs(int(float(wid)/float(dx)))
h = t.tensor(ones(L), idx, self.gud(idx))
conv_x = x.convolve(h, idx)
a = (L * xy.convolve(h, idx) - conv_x * y.convolve(h, idx)) / (L * (x**2).convolve(h, idx) - conv_x ** 2)
return a
def test2():
# A = numpy.array([[1, 4, 7, 10], [2, 5, 8, 11], [3, 6, 9, 12]]);
# A = numpy.arange(12).reshape([4,3]) + 1;
A = numpy.arange(1000).reshape([10,10,10])+1;
(ans1, ans2) = DTA.DTA(tensor.tensor(A), [2,2,2]);
print ans1;
for i in range(0, len(ans2)):
print "{0} th array\n {1}".format(i, ans2[i]);
print ans1.totensor();
def test2():
# A = numpy.array([[1, 4, 7, 10], [2, 5, 8, 11], [3, 6, 9, 12]]);
# A = numpy.arange(12).reshape([4,3]) + 1;
A = numpy.arange(1000).reshape([10, 10, 10]) + 1
(ans1, ans2) = DTA.DTA(tensor.tensor(A), [2, 2, 2])
print ans1
for i in range(0, len(ans2)):
print "{0} th array\n {1}".format(i, ans2[i])
print ans1.totensor()
def ttv(self, v, dims):
"""
Computes the product of this tensor with the column vector along
specified dimensions.
Parameters
----------
v - column vector
d - dimensions to multiply the product
Returns
-------
out : a sparse tensor if 50% or fewer nonzeros
"""
(dims, vidx) = tools.tt_dimscheck(dims, self.ndims(), len(v))
remdims = numpy.setdiff1d(range(self.ndims()), dims)
newvals = self.vals
subs = self.subs
# Multiple each value by the appropriate elements of the appropriate vector
for n in range(len(dims)):
idx = subs[:, dims[n]]
# extract indices for dimension n
w = v[vidx[n]]
# extract nth vector
bigw = w[idx]
# stretch out the vector
newvals = numpy.multiply(newvals.flatten(), bigw)
# Case 0: all dimensions specified - return the sum
if len(remdims) == 0:
c = numpy.sum(newvals)
return c
# Otherwise figure out the subscripts and accumulate the results
newsubs = self.subs[:, remdims]
newsiz = numpy.array(self.shape)[remdims]
# Case 1: return a vector
if len(remdims) == 1:
c = tools.accum_np(newsubs, newvals, newsiz[0])
#if numpy.count_nonzero(c) < 0.5*newsiz[0]:
# c = sptensor.sptensor(numpy.arange(newsiz[0]).reshape(newsiz[0],1), c.reshape(newsiz[0],1));
#else:
c = tensor.tensor(c, newsiz)
return c
# Case 2: result is a multi-way array
c = sptensor.sptensor(newsubs, newvals.reshape(len(newvals), 1),
newsiz)
# check to see if it's dense
if c.nnz() > 0.5 * numpy.prod(c.shape):
return c.totensor()
return c
def k_2Inner(tens):
assert isinstance(tens,pureTensor)
answer= tensor()
rel =tens.monomials[1]
for i in rel.leadingMonomial:
answer = answer + i.coefficient * pureTensor((i.submonomial(0,1),i.submonomial(1,2), 1))
answer = answer + i.coefficient * pureTensor((1,i.submonomial(0,1),i.submonomial(1,2)))
for i in rel.lowerOrderTerms:
answer = answer - i.coefficient * pureTensor((i.submonomial(0,1),i.submonomial(1,2), 1))
answer = answer - i.coefficient * pureTensor((1,i.submonomial(0,1),i.submonomial(1,2)))
return answer
def replacer_T(a, i, j):
return sqrt(2.0) * tensor([
T1(i, j),
T2(i, j),
T3(i, j),
T4(i, j),
T5(i, j),
T6(i, j),
T7(i, j),
T8(i, j)
], a)
def ctor(verbose):
dat = numpy.arange(24);
dat[10] = 100;
dat[16] = -1;
siz = numpy.array([4,3,2]);
obj = tensor.tensor(dat, siz);
if(verbose == 1):
print obj;
print obj.ndims();
obj2 = tensor.tensor(dat.reshape([2,3,4]),siz);
if(verbose == 1):
print obj;
print obj.shape;
dat = numpy.array([1,2,3,4,5,6,7,8,9,10,0,0,13,14,15,16,17,18]);
obj2 = tensor.tensor(dat);
if(verbose == 1):
print obj2;
print obj2.shape;
def T(a, i, j):
"""Fundamental representation matrices"""
return tensor([
T1(color_key(i, 'fu'), color_key(j, 'af')),
T2(color_key(i, 'fu'), color_key(j, 'af')),
T3(color_key(i, 'fu'), color_key(j, 'af')),
T4(color_key(i, 'fu'), color_key(j, 'af')),
T5(color_key(i, 'fu'), color_key(j, 'af')),
T6(color_key(i, 'fu'), color_key(j, 'af')),
T7(color_key(i, 'fu'), color_key(j, 'af')),
T8(color_key(i, 'fu'), color_key(j, 'af'))
], color_key(a, 'ad'))
def ttmTest(verbose):
dat = numpy.arange(24).reshape([2,3,4]);
siz = numpy.array([2,3,4]);
obj = tensor.tensor(dat, siz);
A = numpy.arange(18).reshape([6,3]);
print obj.ttm(A,1);
B = numpy.arange(12).reshape([3,4]);
print obj.ttm([A,B],[1,2]);
print "a";
print obj.ttm([A.transpose(),B.transpose()],[1,2],'t');
def ttmTest(verbose):
dat = numpy.arange(24).reshape([2, 3, 4])
siz = numpy.array([2, 3, 4])
obj = tensor.tensor(dat, siz)
A = numpy.arange(18).reshape([6, 3])
print obj.ttm(A, 1)
B = numpy.arange(12).reshape([3, 4])
print obj.ttm([A, B], [1, 2])
print "a"
print obj.ttm([A.transpose(), B.transpose()], [1, 2], 't')
def m_2Inner(PT):
assert isinstance(PT, pureTensor)
assert len(PT) == 4
PT = PT.clean()
w = PT[1] * PT[2]
answer = tensor()
sequence = alg.makeReductionSequence(w)
for reductionFunction, weight in sequence:
answer += PT.coefficient * weight * PT[0] \
* pureTensor([reductionFunction.leftMonomial,
reductionFunction.relation,
reductionFunction.rightMonomial]) * PT[3]
return answer
def mathlogicops(verbose):
dat = numpy.arange(24).reshape([2, 3, 4])
siz = numpy.array([2, 3, 4])
obj = tensor.tensor(dat, siz)
print obj + 100
print obj - 100
print obj + (obj + 10)
print obj * 1.5
print obj < 10
print obj > 10l
print obj == 10
print obj == obj
def ctor(verbose):
dat = numpy.arange(24)
dat[10] = 100
dat[16] = -1
siz = numpy.array([4, 3, 2])
obj = tensor.tensor(dat, siz)
if (verbose == 1):
print obj
print obj.ndims()
obj2 = tensor.tensor(dat.reshape([2, 3, 4]), siz)
if (verbose == 1):
print obj
print obj.shape
dat = numpy.array(
[1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 0, 0, 13, 14, 15, 16, 17, 18])
obj2 = tensor.tensor(dat)
if (verbose == 1):
print obj2
print obj2.shape
def ttv(self, v, dims):
"""
Computes the product of this tensor with the column vector along
specified dimensions.
Parameters
----------
v - column vector
d - dimensions to multiply the product
Returns
-------
out : a sparse tensor if 50% or fewer nonzeros
"""
(dims, vidx) = tools.tt_dimscheck(dims, self.ndims(), len(v));
remdims = numpy.setdiff1d(range(self.ndims()), dims);
newvals = self.vals;
subs = self.subs;
# Multiple each value by the appropriate elements of the appropriate vector
for n in range(len(dims)):
idx = subs[:, dims[n]]; # extract indices for dimension n
w = v[vidx[n]]; # extract nth vector
bigw = w[idx]; # stretch out the vector
newvals = numpy.multiply(newvals.flatten(), bigw);
# Case 0: all dimensions specified - return the sum
if len(remdims) == 0:
c = numpy.sum(newvals);
return c;
# Otherwise figure out the subscripts and accumulate the results
newsubs = self.subs[:, remdims];
newsiz = numpy.array(self.shape)[remdims];
# Case 1: return a vector
if len(remdims) == 1:
c = tools.accum_np(newsubs, newvals, newsiz[0]);
#if numpy.count_nonzero(c) < 0.5*newsiz[0]:
# c = sptensor.sptensor(numpy.arange(newsiz[0]).reshape(newsiz[0],1), c.reshape(newsiz[0],1));
#else:
c = tensor.tensor(c, newsiz);
return c;
# Case 2: result is a multi-way array
c = sptensor.sptensor(newsubs, newvals.reshape(len(newvals), 1), newsiz);
# check to see if it's dense
if c.nnz() > 0.5*numpy.prod(c.shape):
return c.totensor();
return c;
def mathlogicops(verbose):
dat = numpy.arange(24).reshape([2,3,4]);
siz = numpy.array([2,3,4]);
obj = tensor.tensor(dat, siz);
print obj + 100;
print obj - 100;
print obj + (obj + 10);
print obj * 1.5;
print obj < 10;
print obj > 10l
print obj == 10;
print obj == obj;
def permutetest(verbose):
dat = numpy.arange(24).reshape([2, 3, 4])
siz = numpy.array([2, 3, 4])
obj = tensor.tensor(dat, siz)
if (verbose == 1):
print obj
print "permute by {0}".format("[2,0,1]")
print obj.permute([2, 0, 1])
print "permute by {0}".format("[1,2,0]")
print obj.permute([1, 2, 0])
print "ipermuted by {0}".format("[1,2,0]")
print obj.ipermute([1, 2, 0])
print obj.permute([1, 2, 0]).ipermute([1, 2, 0])
def permutetest(verbose):
dat = numpy.arange(24).reshape([2,3,4]);
siz = numpy.array([2,3,4]);
obj = tensor.tensor(dat, siz);
if(verbose == 1):
print obj;
print "permute by {0}".format("[2,0,1]");
print obj.permute([2,0,1]);
print "permute by {0}".format("[1,2,0]");
print obj.permute([1,2,0]);
print "ipermuted by {0}".format("[1,2,0]");
print obj.ipermute([1,2,0]);
print obj.permute([1,2,0]).ipermute([1,2,0]);
def __init__(self,i,j,k,l):
perms = list(permutations([0,1,2,3]))
il = lorentz_key(i) if (i<0) else i
jl = lorentz_key(j) if (j<0) else j
kl = lorentz_key(k) if (k<0) else k
ll = lorentz_key(l) if (l<0) else l
tmp = new(il, 4,{jl:4,kl:4,ll:4})
for perm in perms:
tmp.__setitem__({il:perm[0], jl:perm[1], kl:perm[2], ll:perm[3]},
tensor([parity(list(perm))], None))
super(Epsilon,self).__init__(tmp._array, tmp._toplevel_key)
def __add__(self,other ):
if other == 0:
return self
new1 = self.clean()
other = other.clean()
if isinstance(other,pureTensor):
if new1.isAddable(other):
if self.isZero():
return other
else:
newCoefficient = new1.coefficient + other.coefficient
return pureTensor(self.monomials,newCoefficient)
else:
return tensor.tensor((new1,other))
else:
return NotImplemented
def stress(self, C):
"""
Compute 2nd Piola-Kirchhoff stress
"""
dim = len(C)
PK2 = tensor.tensor(dim)
detC = tensor.det(C)
detF = np.sqrt(detC)
lnJ = np.log(detF)
lamda = self.get1LAME()
mu = self.get2LAME()
invC = tensor.inv(C)
I = tensor.I(dim)
PK2 = mu * (I - invC) + lamda * lnJ * invC
return PK2
def sum(self, idx):
obj = t.tensor(self)
for c in idx:
obj.transpose([c,0])
obj.arr = obj.arr.sum(axis=0)
a = array([])
if type(obj.arr) != type(a):
return obj.arr
obj.idx = obj.idx[1:]
obj.ud = obj.ud[1:]
#for c in idx:
# if c in obj.idx:
# N = obj.arr.shape[obj.idx.find(c)]
# ud = obj.ud[obj.idx.find(c)]
# o = t.tensor(ones(N), idx=c, ud={"u":"d", "d": "u"}[ud])
#
# obj = obj * o
# check = True
return obj
def reduce(self,tens):
if len(self) == 1: #If this is just an algebra
return self[0].reduce(tens)
newTens = tensor.tensor()
for pure in tens:
polys = []
for index in range(len(pure)):
currentAlgebra = self[index]
polys.append(currentAlgebra.reduce(pure[index]))
def listOfPolysToTensors(ps):
ps = iter(ps)
pureTensors = [pureTensor.pureTensor(mono) for mono in next(ps)]
for poly in ps:
newPureTensors = []
for mono in poly:
newPureTensors.extend([x.tensorProduct(mono) for x in pureTensors])
pureTensors = newPureTensors
return tensor.tensor(pureTensors)
newTens = newTens + pure.coefficient * listOfPolysToTensors(polys)
return newTens
def test1():
A = ttensor.ttensor(tensor.tenrands([2,3,4]),
[numpy.random.random([10,2]),
numpy.random.random([30,3]),
numpy.random.random([40,4])]).totensor();
[a,b] = DTA.DTA(A, [1,2,3]);
#print a;
#print b;
Core = numpy.arange(24).reshape([2,3,4]);
#Core = numpy.array([[1,3,5],[2,4,6]] , [[7,9,11],[8,10,12]])
u1 = numpy.array([[1,2],[3,4]]);
u2 = numpy.array([[0,1,0],[1,0,1],[1,1,1]]);
u3 = numpy.array([[1,1,1,1],[1,2,3,4],[1,1,1,1]]);
tt = ttensor.ttensor(tensor.tensor(Core), [u1,u2,u3]);
print tt;
[a,b] = DTA.DTA(tt.totensor(), [1,2,3]);
print a;
print a.totensor();
print b;
