I'm computing exp using BigFloat library in python. But then I have to compute the inverse of a matrix of BigFloats. I use function numpy.linalg.inv but I'm getting the following error:
No loop matching the specified signature and casting was found for ufunc inv
So I'd say I need to convert BigFloat to some other type. How can I do it?
Considering a 2D list of Bigfloat values (with exp):
from bigfloat import *
import numpy as np
row = 4
col = 4
mat = []
with precision(1000):
for i in range(0,row):
row = []
for j in range(0,col):
row.append(exp(1./(i+j+1)))
mat.append(row)
print "A single element in bigfloat: \n",mat[0][0]
mat_numpivized = np.array([[np.float(col) for col in row] for row in mat])
print "\nNumpived matrix of bigfloat: \n",mat_numpivized
inversed_mat = np.linalg.inv(mat_numpivized)
print "\nInverse of numpivized bigfloat matrix: \n", inversed_mat
result:
A single element in bigfloat:
2.71828182845904523536028747135266249775724709369995957496696762772407663035354759457138217852516642742746639193200305992181741359662904357290033429526059563073813232862794349076323382988075319525101901157383418793070215408914993488416750924476146066808226480016847741185374234544243710753907774499206950
Numpived matrix of bigfloat:
[[ 2.71828183 1.64872127 1.39561243 1.28402542]
[ 1.64872127 1.39561243 1.28402542 1.22140276]
[ 1.39561243 1.28402542 1.22140276 1.18136041]
[ 1.28402542 1.22140276 1.18136041 1.15356499]]
Inverse of numpivized bigfloat matrix:
[[ 3.82106765 -30.68951855 61.37113204 -34.60880961]
[ -30.68951855 385.05574805 -890.19682475 538.10687764]
[ 61.37113204 -890.19682475 2211.11911647 -1390.16166038]
[ -34.60880961 538.10687764 -1390.16166038 893.29628089]]