ulaff / notebooks Goto Github PK

8.2.5 Alternative Gauss Jordon Algorithm

Running in LAFF v3 environment, this was working before the latest commit but now throws the following error:

AssertionError Traceback (most recent call last)
in ()
6 A = np.matrix('1 -1 2; 2 0 2;-2 0 -4')
7 B = np.matrix('7;4;-10')
----> 8 GJ_Inverse_alt( A, B )
9 print(B)
10

in GJ_Inverse_alt(A, B)
26
27 laff.ger( -1.0, a01, b10t, B00 )
---> 28 laff.copy( -a01, b01 ) #b01 = -a01
29 laff.ger( -1.0, a21, b10t, B20 )
30 laff.copy( -a21, b21 ) #b21 = -a21

Web/laff/LAFF/notebooks/laff/vecvec/copy.py in copy(x, y)
23
24 if m_x is 1 and m_y is 1: # x is a row, y is a row
---> 25 assert n_x == n_y, "laff.copy: size mismatch between x and y"
26 for i in range(n_x): y[0, i] = x[0, i]
27

AssertionError: laff.copy: size mismatch between x and y

However there is nothing wrong in matrices:

A = np.matrix('1 -1 2; 2 0 2;-2 0 -4')

B = np.matrix('7;4;-10')

A*B

Out[11]:

matrix([[-17],

[ -6],

[ 26]])

np.linalg.inv(A) * B

Out[12]:

matrix([[-1.],

[-2.],

[ 3.]])

Hi, I created a new issue here while I was working on 6.2.5 notebook

The original answers notebook throws back error and I've cited many previous notebook throwing errors on laff methods:

• copy
• invscal
• gemm

Please checkout the whole issue and clarify.

Thanks

In laff.matvec.trsv function, b is a vector with row or column size = 1. At line 75 ,the function trsv_ltu is called as trsv_ltu(A,b) .However,the function laff.matvec.trsv_ltu treats b as a 2D matrix

def trsv_ltu(A,B):
...

So, a call
laff.trsv('Lower triangular','Transpose','Unit diagonal',A,b)
where b is a column vector
will cause an error message
"dots( -l21, b2, beta1 )
NameError: name 'b2' is not defined"