Feeds:
Posts

## SAGE tip: More about Differential Forms

In a previous post we discuss the definition of the coordinated patch on a manifold, how to define differential forms, wedge them or calculate their exterior derivative… even simplify’em.

This time a zero form will be defined and a list of forms will be created… So, let’s begin!

# Define a 0-form

Once created the coordinated patch and the differential forms algebra

sage: reset()
sage: var('t,x,y,z')
sage: U = CoordinatePatch((t,x,y,z))
sage: Omega = DifferentialForms(U)

A 0-form is defined as an element of $\Omega^0(U)$, but the value of the 0-form is given inside the declaration command,

sage: A = DifferentialForm(Omega, 0, exp(x*y))

I tried addition, multiplication, wedge product and exterior differentiation on 0-forms and they worked!

Of course you can combine them with forms of different degrees.

# New method of defining a form

I wrote to Joris this morning… but before he was able to answer, from the documentation of the differential form package.

When one calls the generators of the differential form,

sage: Omega.gen(1)
dx

and the result is a differential form… Thus, one can assign a form as follow,

sage: A = sin(x)* Omega.gen(2)
sage: B = cos(y) * Omega.gen(0)
sage: C = sin(z) *Omega.gen(1)
sage: D = cos(y) * Omega.gen(2)

And this forms can be wedged, differentiated, et cetera. ðŸ™‚

# List of forms

Finally, after discovering the above behavior I tried the following, a list of differential forms ;-), for example,

sage: pro = matrix([[A, B], [C, D]])
sage: for i in range(2):
...       for j in range(2):
...           show(pro[i,j].diff())

returns
$cos(x)dx\wedge dz$
$sin(y)dx\wedge dy$
$-cos(z)dy\wedge dz$
$-sin(y)dy\wedge dz$

This implies that somehow one can manage a series of forms by using list properties… I expect to go deeper on this subject in the future! ðŸ˜€

Enjoy people!!!

Dox

## SAGE tip: GRmodule. Day 07.

This code is supposed to be (if some one does the work in the future) located in sage.tensor.differential_form_element.

The code presented below is a slight modification of Joris code for differential forms manipulation on SAGE.

## Needed modules

from sage.symbolic.ring import SymbolicRing, SR
from sage.rings.ring_element import RingElement
from sage.algebras.algebra_element import AlgebraElement
from sage.rings.integer import Integer
from sage.combinat.permutation import Permutation

The advantage of using this is that, tensors defined here are an Algebra element, not just a python object as in the previous code.

The sage.combinat.permutation won’t be used (yet), but could be useful if tensor symmetries are defined.

#### TensorFormatter

class TensorFormatter:
r"""
This class contains all the functionality to print a tensor in a
graphically pleasing way.  This class is called by the _latex_ and
_repr_ methods of the Tensor class.
"""
def __init__(self, space):
r"""
Construct a tensor formatter.  See
TensorFormatter for more information.
"""
self._space = space

def repr(self, comp, fun):
r"""
String representation of a primitive tensor, i.e. a function
times a tensor product of d's of the coordinate functions.

INPUT:

- comp -- a subscript of a differential form.

- fun -- the component function of this form.

EXAMPLES::

sage: from sage.tensor.tensor_element import TensorFormatter
sage: x, y, z = var('x, y, z')
sage: U = CoordinatePatch((x, y, z))
sage: D = TensorFormatter(U)
sage: D.repr((0, 1), z^3)
'z^3*dx@dy'

"""

str = "@".join( \
[('d%s' % self._space.coordinate(c).__repr__()) for c in comp])

if fun == 1 and len(comp) > 0:
# We have a non-trivial form whose component function is 1,
# so we just return the formatted form part and ignore the 1.
return str
else:
funstr = fun._repr_()

if not self._is_atomic(funstr):
funstr = '(' + funstr + ')'

if len(str) > 0:
return funstr + "*" + str
else:
return funstr

def latex(self, comp, fun):
r"""
Latex representation of a primitive differential form, i.e. a function
times a tensor product of d's of the coordinate functions.

INPUT:

- comp -- a subscript of a differential form.

- fun -- the component function of this form.

EXAMPLES::

sage: from sage.tensor.tensor_element import TensorFormatter
sage: x, y, z = var('x, y, z')
sage: U = CoordinatePatch((x, y, z))
sage: D = TensorFormatter(U)
sage: D.latex((0, 1), z^3)
'z^{3} d x \otimes d y'

"""

from sage.misc.latex import latex

str = " \otimes ".join( \
[('d %s' % latex(self._space.coordinate(c))) for c in comp])

if fun == 1 and len(comp) > 0:
return str
else:
funstr = latex(fun)

if not self._is_atomic(funstr):
funstr = '(' + funstr + ')'

return funstr + " " + str

def _is_atomic(self, str):
r"""
Helper function to check whether a given string expression
is atomic.

EXAMPLES::

sage: x, y, z = var('x, y, z')
sage: U = CoordinatePatch((x, y, z))
sage: from sage.tensor.tensor_element import TensorFormatter
sage: D = TensorFormatter(U)
sage: D._is_atomic('a + b')
False
sage: D._is_atomic('(a + b)')
True
"""
level = 0
for n, c in enumerate(str):
if c == '(':
level += 1
elif c == ')':
level -= 1

if c == '+' or c == '-':
if level == 0 and n > 0:
return False
return True

The only I’ve changed here is “DifferentialForm” by “Tensor” and “\wedge” by “\otimes”

The above code allows to write the tensor product in a basis, $dx^1\otimes\cdots\otimes dx^n$. The chosen symbol for denoting the tensor product was @.

## Tensor Class

This code is incomplete due to:

• I’ve not defined the TensorsAlgebra, which should be done in parallel.
• There are a lot of attributes not presented in this class.
• class Tensor(AlgebraElement):
r"""
Tensor class.
"""

def __init__(self, parent, degree, fun = None):
r"""
Construct a tensor.

INPUT:

- parent -- Parent algebra of tensors.

- degree -- Degree of the tensor.

- fun (default: None) -- Initialize this differential form with the given function.  If the degree is not zero, this argument is silently ignored.

EXAMPLES::

sage: x, y, z = var('x, y, z')
sage: F = Tensors(); F
Algebra of tensors in the variables x, y, z
sage: f = Tensor(F, 0, sin(z)); f
sin(z)

"""

if not isinstance(parent, Tensors):
raise TypeError, "Parent not an algebra of tensors."

RingElement.__init__(self, parent)

self._degree = degree
self._components = {}

if degree == 0 and fun is not None:
self.__setitem__([], fun)

def __getitem__(self, subscript):
r"""
Return a given component of the tensor.

INPUT:

- subscript: subscript of the component.  Must be an integer
or a list of integers.

EXAMPLES::

sage: x, y, z = var('x, y, z')
sage: F = Tensors(); F
Algebra of tensors in the variables x, y, z
sage: f = Tensor(F, 0, sin(x*y)); f
sin(x*y)
sage: f[()]
sin(x*y)
"""

if isinstance(subscript, (Integer, int)):
subscript = (subscript, )
else:
subscript = tuple(subscript)

dim = self.parent().base_space().dim()
if any([s >= dim for s in subscript]):
raise ValueError, "Index out of bounds."

if len(subscript) != self._degree:
raise TypeError, "%s is not a subscript of degree %s" %\
(subscript, self._degree)

"""sign, subscript = sort_subscript(subscript)"""

if subscript in self._components:
return sign*self._components[subscript]
else:
return 0

def __setitem__(self, subscript, fun):
r"""
Modify a given component of the tensor.

INPUT:

- subscript: subscript of the component.  Must be an integer or a list of integers.

EXAMPLES::

sage: F = Tensors(); F
Algebra of tensors in the variables x, y, z
sage: f = Tensor(F, 2)
sage: f[1, 2] = x; f
x*dy@dz
"""

if isinstance(subscript, (Integer, int)):
subscript = (subscript, )
else:
subscript = tuple(subscript)

dim = self.parent().base_space().dim()
if any([s >= dim for s in subscript]):
raise ValueError, "Index out of bounds."

if len(subscript) != self._degree:
raise TypeError, "%s is not a subscript of degree %s" %\
(subscript, self._degree)

"""sign, subscript = sort_subscript(subscript)"""
self._components[subscript] = SR(fun)

Ok, so again I’ve changed “DifferentialForm(s)” by “Tensor(s)”, drop the permutation of indices (’cause tensors do not need to be neither symmetric nor anti-symmetric.

I’ll keep working with this code… It’s all by now. Oh! I’ll post next week the rest of the code based in Sergey’s GRPy.

Enjoy.

Dox

## SAGE tip: GRmodule. Day 06.

I’ve just updated the SAGE worksheet which uses the definitions described in the previous posts.

• There are some explanations in text format
• The code has been hidden… because is long.
• Moreover… I’ve discover something really amazing! Joris Vankerschaver‘s code of the differential form package in SAGE. Thus, I could use some ideas from Joris’ code to improve GRmodule. Nice, Isn’t it?

Let’s hope I could so something nice this weekend!

Don’t forget check the worksheet, and post some comment for feedback! ðŸ˜‰

Enjoy!

Dox

## SAGE tip: GRmodule. Day 05

Hi everyone!

This time the Christoffel connection will be defined.

# The code

As usual, here is the code:

class Christoffel(Tensor):
'''The class to represent Christoffel Symbols of the second kind. Please
note that while it inherits from Tensor, Christoffel symbols are
NOT tensors'''

def __init__(self,metr,symbol='C',rank=(1,2),sh=(1,-1,-1)):

# The metric
self.g_down = metr

# Since we have a metric we do indeed have a coordinate system
self.rep  = self.g_down.rep

self.g_up = metr.inverse

# Please note that this call will trigger a call to allocate in
# the Tensor class, but the allocate will actually be the allocate
# defined below
super(Christoffel,self).__init__(symbol,rank,sh,coords=metr.coord)

def allocate(self,rank):
Tensor.allocate(self,rank)
# Now that we have allocated things, time to actually compute things
for i in xrange(self.dim):
for k in xrange(self.dim):
for l in xrange(self.dim):
sum = 0
for m in xrange(self.dim):
term1 = diff(self.g_down[m,k],self.g_down.coords[l])
term2 = diff(self.g_down[m,l],self.g_down.coords[k])
term3 = diff(self.g_down[k,l],self.g_down.coords[m])

tr = self.g_up[i,m] * (term1+term2-term3)

sum += tr
res = sum/2
self.components[i,k,l] = res
self.getNonZero()

This code is almost a copy of Sergey’s one, except for the use of xrange instead of np.arange, and the fact that I’ve dropped the minus signs denoting the shape of the tensors.

## Sage implementation

This time I won’t present a Python file, but a SAGE file, GRmodule.

Enjoy!

Dox

## SAGE tip: Working with Differential Forms.

A few weeks ago I realize there is a package for working with differential forms in SAGE. So, I’ve tried to take advantage of that!

This is what I’ve learnt so far.

If you work with differential forms you should know (probably better than me), that you define forms in a coordinate patch… rarely globally.

sage: reset()
sage: # The variables
sage: var('theta,phi,r')
sage: var("bps", latex_name=r"\bar{\psi}")
sage: var("bph", latex_name=r"\bar{\phi}")
sage: var("bth", latex_name=r"\bar{\theta}")
sage: # The coordinate system
sage: coords = [r,theta, phi, bps, bth, bph]
sage: U = CoordinatePatch((r,theta, phi, bps, bth, bph))
sage: Omega = DifferentialForms(U)

From a previous post a weird declaration of variables has been made.

But don’t lose yourself into the code!!!… the important here is:

• Seven variables have been declared,
• These are converted into a list,
• and also are the coordinated of the Coordinate Patch.
• Finally, $\Omega$ is the algebra of differential forms. Just like $\Omega^\bullet(M)$, where $M$ is the manifold (in this case the patch).
• ## Define the differential forms

Differential form have to be defined… like this,

sage: A = DifferentialForm(Omega, 1)
sage: A[1] = -1/4
sage: A[2] = 1/4*(-cos(theta) - sin(theta))
sage: A[3] = cos(bps)*sin(bth)*sin(bph)
sage: A[4] = sin(bps)*cos(bth)*sin(bph)
sage: A[5] = sin(bps)*sin(bth)*cos(bph)

The first line defines $A$ as an element of $\Omega^1(M)$.

The rest of the lines are for setting the components of $A$. Remember that the coordinates run from 0 to 6 in this example! ðŸ˜› And, of course, they are ordered strictly as we declared them, i.e.,

$A = -\frac{1}{4}d\theta -\frac{1}{4}(\cos(\theta) + \sin(\theta))d\phi + \cos(\bar{\psi})\sin(\bar{\theta})\sin(\bar{\phi})d\bar{\psi} +\cdots$

## Exterior Differentiation

Once the form has been declared, one might differentiate it by using the diff command,

sage: A.diff()

returns

$(\frac{1}{4} \, \sin\left(\theta\right) - \frac{1}{4} \, \cos\left(\theta\right)) d \theta \wedge d \phi$

## Showing the forms

It’s very useful to see the formulas in a nice written way… this is one of the features I love the more from SAGE, to see your form, use

 sage: show(A)

## Wedge Product

Of course one of the most important operations when working with forms is the wedge product. For using this, try

sage: C = A.wedge(A.diff())
sage: show(C)

it vanishes. One might try to define another form,

sage: B = DifferentialForm(Omega,2)
sage: B[0,1] = sin(bth)

Note that $B\in \Omega^2(M)$, and has only one non-vanishing component.

Thus, $A\wedge B$ is calculated,

sage: A.wedge(B)

One can also define multiple wedge product,

sage: A.wedge(B.diff()).wedge(B)

Needless to say one can multiply a form by a function, or number.

## Simplifying a form

Forms have not implemented the simplify_full attribute, but their components, which are functions, do. So, after a very complicated calculation one might try to implement a long routine of simplification, say,

sage: D = A.wedge(B.diff()).wedge(B)
for i in xrange(Omega.ngens()):
for j in xrange(i+1, Omega.ngens()):
for k in xrange(j+1, Omega.ngens()):
for l in xrange(k+1, Omega.ngens()):
for m in xrange(l+1, Omega.ngens()):
D[i,j,k,l,m] = D[i,j,k,l,m].simplify_full()


In the above, Omega.ngens() returns the dimension of the patch. Additionally, each index runs from the value of the previous plus one, in order to avoid repetition (due to anti-symmetric property) or the zeros values.

Well, I think this is much of it! I’ll keep you posted, in case I learn more about ot.

Enjoy life! and Happy Valentine’s day! ðŸ˜‰

Dox

Hi everyone, today I start to write a new class for dealing with non-Abelian forms, i.e., forms with values in a certain Lie-algebra. This could be really useful when computing Yang-Mills theories in physics.

So, I started by defining a new object which have two entries, a differential form and a matrix, and call it nAform. The code I wrote was,

class nAform(object):
def __init__(self, a, b):
self._form = a
self._matrix = b
if isinstance(other, nAform):
if (self._matrix == other._matrix):
return nAform(self._form + other._form, self._matrix)
else:
return NotImplemented
return NotImplemented
def __mul__(self, other):
if isinstance(other, nAform):
return nAform(self._form.wedge(other._form), self._matrix.commutator(other._matrix))
return NotImplemented
def diff(self):
return nAform(self._form.diff(), self._matrix)
def __repr__(self):
return str((self._form, self._matrix))
def __str__(self):
return self.__repr__()

## Explanation

One should enter a couple of arguments when defining the nAform object. The __init__ attribute recognize them.

Then an addition attribute is defined, this is incomplete!!!

Another attribute is the multiplication, which take the wedge product of the forms and the commutator of the matrices.

I also implement the exterior derivative on nAform objects.

Finally the __repr__ and __str__ are attributes for returning the data.

## TO-DO

• I couldn’t implement the addition of nAform’s if the matrices are different. As
Nicolas M. Thiery note, this objects should define a Monoid (or something quit close to it). But my programming skills are not so developed.
• It would be great if one could define the multiplication by a constant or function.
• The show attribute is not implemented, but one can show either forms of matrices by themselves… I think that is someone knows how they work, would be easy to do that! ðŸ˜‰
• One could try to implement the simplification attributes on differential forms.
• I don’t remember exactly why I was looking for it, but I think could be useful to define and attribute on forms which show the generator, like .gens(), but for a given differential form, say
sage: x, y, z = var('x, y, z')
sage: U = CoordinatePatch((x, y, z)); U
Open subset of R^3 with coordinates x, y, z
sage: F = DifferentialForms(U); F
Algebra of differential forms in the variables x, y, z
sage: F.gens()
(dx, dy, dz)
sage: F.ngens()
3

I’d like one which do like this,

sage: form1 = DifferentialForm(F, 1); form1.ngens()
3
sage: form1 = DifferentialForm(F, 1); form1.gens()
(dx, dy, dz)
form2 = DifferentialForm(F, 2); form2.ngens()
3
sage: form1 = DifferentialForm(F, 1); form1.gens()
(dx/\dy, dy/\dz, dz/\dx)
• ## Simple SAGE sheet

Worksheet

Hope you can help me with these plenty tasks.

Enjoy!

Dox

## SAGE tip: Hiding LaTeX code in the notebook

A quick tip today, thanks to Jason from the google group for the tip. ðŸ™‚

I must admit to be a LaTeX addict, I’ve never used a wysiwyg editor… I do everything in LaTeX code,

• Reports,
• Magazine articles,
• Letters,
• Scientific articles,
• Tried once to do a music sheet,
• Posters,
• I tried sagetex,
• and plenty more!…
• When I discover that SAGE notebook accepted LaTeX code, that help convince me to use it!!!

Example:

%latex
$\vec{F} = m\vec{a}$

generates

$\vec{F} = m\vec{a}$

I use this method to explain in the calculation worksheet what I’m doing. However, when trying to print the sheet the LaTeX code was around and the file was `messy’.

If you use either %hide or %hideall character, the LaTeX output is hidden (at least when you want to print). Sometimes I get the cose hidden as well in the notebook, but thats not that important as hiding it when printing ðŸ™‚

Thank you again Jason.

Cheers!!

Dox