ASKSAGE: Sage Q&A Forum - RSS feedhttps://ask.sagemath.org/questions/Q&A Forum for SageenCopyright Sage, 2010. Some rights reserved under creative commons license.Sat, 30 May 2020 07:51:15 +0200Are there fast(er) ways to compute inverses of Hermitian matrices?https://ask.sagemath.org/question/51638/are-there-faster-ways-to-compute-inverses-of-hermitian-matrices/I'm dealing with some Hermitian matrices valued in symbolic expressions. I need to be able to compute inverses of these matrices, but they seem to get big pretty fast. That is it would nice to be able to do this in a reasonable amount of time with 10x10 matrices at least, and hopefully larger than that.
I'm currently running a cell where the inverses of 2 10x10 matrices are being computed, and it's taken well over 10 minutes.
Are there ways to exploit the fact that these matrices are Hermitian to compute the inverses faster, and/or are there better ways to compute inverses of symbolic matrices than the standard .inverse()??
EDIT: I originally avoided an example, because their construction is somewhat convoluted as you'll see below.
First we have two variables which for which I want to run the larger program for various choices thereof.
p = 1, q= 5
X = {(i): var("X_{}".format(i), latex_name="X_{}") for i in range(2*p*q)}
e = {(i,j,k): var("e_{}{}{}".format(i,j,k), latex_name="e_{{{}{}{}}}".format(i,j,k)) for i in range(2) for j in range(q) for k in range((p+q))}
The following creates a list L such that L[i] is the collection of lists of length i whose elements are integers between 0 and q-1 in strictly increasing order. This is used throughout the larger program.
L = [[[]]]
LL = []
for i in range(q):
LL.append([i])
L.append(LL)
if q>=2:
for k in range(2,q+1):
LLL = []
for i in range(binomial(q, k-1)):
for j in range(q):
if L[k-1][i][len(L[k-1][i])-1]<j:
mm = []
for t in L[k-1][i]:
mm.append(t)
mm.append(j)
LLL.append(mm)
L.append(LLL)
Here, a matrix is defined which we be used to construct the matrices I'm interested in.
HE = matrix(SR, q, q)
for i in range(q):
for j in range(q):
prod = 0
for k in range(p):
prod -= (e[(0, i, k)]*e[(1, j, k)])
for l in range(p+q):
if l>=p:
prod += (e[(0, i, l)]*e[(0, j, l)])
HE[i,j] = prod
h = {(i): var("h_{}".format(i)) for i in range(q+1)}
for i in range(q+1):
h[i] = matrix(SR, binomial(q, i), binomial(q, i))
h[0][0,0] = 1
for i in range(1, q+1):
for z in L[i]:
for w in L[i]:
h[i][L[i].index(z), L[i].index(w)] = HE[z, w].det()
Finally, we come to the inverse matrices I would like to compute. It is absolutely necessary that I have these inverse matrices.
hinv = {(i): var("hinv_{}".format(i)) for i in range(q+1)}
for i in range(q+1):
hinv[i] = h[i].inverse()sum8tionSat, 30 May 2020 07:51:15 +0200https://ask.sagemath.org/question/51638/<symbolic matrix>.simplify_rational() accepts no keywords?https://ask.sagemath.org/question/49934/symbolic-matrixsimplify_rational-accepts-no-keywords/I have a symbolic matrix whose entries I would like to simplify for display. Minimal working example (in the `sage` REPL):
```
a,b=var('a,b'); m=matrix(SR,2,[[(a*b-a)/a,0],[0,1]]).simplify_rational()
```
works fine and upon calling `m` produces
```[b - 1 0]
[ 0 1]
```
as expected. On the other hand, running
```
a,b=var('a,b'); m=matrix(SR,2,[[(a*b-a)/a,0],[0,1]]).simplify_rational(algorithm='simple')
```
as per the documentation produces
TypeError Traceback (most recent call last)
<ipython-input-30-3edf7642200d> in <module>()
----> 1 a,b=var('a,b'); m=matrix(SR,Integer(2),[[(a*b-a)/a,Integer(0)],[Integer(0),Integer(1)]]).simplify_rational(algorithm='simple')
TypeError: simplify_rational() takes no keyword argument
If I apply the method to a *single* expression it works fine:
```
((a*b-a)/a).simplify_rational(algorithm='simple')
```
gives back `b-1`.
grobberSat, 15 Feb 2020 18:03:18 +0100https://ask.sagemath.org/question/49934/Optimal method to invert symbolic matrixhttps://ask.sagemath.org/question/47401/optimal-method-to-invert-symbolic-matrix/ If I read the source code correctly, a matrix defined over SR is inverted by the Moore-Penrose pseudo-inversion. For a real, symmetric, symbolic, full-rank matrix, is this algorithm optimal? Or is there another algorithm that would be a) less memory intensive, and/or b) less time-consuming. [In my particular case, the attempted inversion of a 4x4 matrix runs out of memory (8 GB) after approximately 15 hours.]Richard_LTue, 06 Aug 2019 01:04:16 +0200https://ask.sagemath.org/question/47401/Formal determinant of symbolic matrixhttps://ask.sagemath.org/question/46624/formal-determinant-of-symbolic-matrix/ I have some sparse symbolic matrices, and want to compute their formal determinant (without cancellation of terms). In other words, if I have the matrix
x,y = var('x,y')
M = Matrix(SR, [[x,y],[x,y]])
I would like the result of
M.determinant()
to be x*y - x*y, rather than just 0. The variables in each monomial are allowed to commute with each other, but on the other hand I would like all monomials containing a 0 to vanish (i.e if in the example above M = Matrix(SR, [[x,0],[x,y]])), then the determinant should be just x*y, rather than x*y - 0*x).
Is there a way to achieve this (without using the expansion of the determinant as permutations, since the dimension of the matrices gets quite big!)? Thanks in advance!
danieleCWed, 22 May 2019 16:32:20 +0200https://ask.sagemath.org/question/46624/