# Revision history [back]

### A question on symbolic Matrices - unexpected Decimals in algebraic entry

Firstly, I'd just like to say that I am starting out with Sage adn Python at the moment, really impressed by the power of Sage and its potential! I am coming at this from an engineering and C++ background, and am currently trying to look at an engineering problem using some matrix linear algebra. I am setting up a from of stiffness matrix and am trying to extract the symbolic eigenvalues and eigenvectors from this. The stiffness matrix below is generated through a few different matrix multiplication operations, based on node and connectivity matrices. Despite trying to define these as being "SR" rings, I seem to end up with entries like (b + 1.0g + 1.0s) rather than the (b + g + s) I was expecting. I suspect this is due to me not understanding rings properly and how to apply them. I have tried the explanations in the Sage documentation, can anybody recommend some further reading or tutorials on this, because it does seem to be a vital topic in getting to grips with Sage?

I enclose my code below, in case anybody can spot a glaringly stupid thing that I've done...

many thanks,

Brian

N = matrix(QQ,[[1,0,-1,0],[0,1,0,-1]])

C = matrix(QQ,[[-1,0,1,0],[0,-1,0,1],[-1,1,0,0],[0,-1,1,0],[0,0,-1,1],[1,0,0,-1]])

D = C.transpose()

space = N.nrows()

M = N*D
M
II = identity_matrix(QQ,space)
b,s,g = var('b s g')
numMembers = M.ncols()

LVector  = II

#loop through members to create LVector
for i in range(numMembers):

m1 = M.matrix_from_columns([i])
mt = m1.transpose()
mag = m1.norm()
Coeff = m1*mt/(mag^2)
size = m1.norm()

if i<2:
#for bars
L = -g*(II - Coeff) + b*Coeff
else:
L = g*(II - Coeff) + s*Coeff

L.factor()

if i == 0:
LVector = L
else:
LVector = LVector.block_sum(L)
print L.str()

print LVector.str()

Pre = D.tensor_product(II)
Post = C.tensor_product(II)
K = Pre*LVector
K = K*Post

Ks = K.change_ring(SR)
print Ks.str()

Eigen = K.eigenvectors_left()
print Eigen.str()