Linear algebra/power iteration #2190
Conversation
|
Hey @zakademic, TravisCI finished with status TravisBuddy Request Identifier: a88eb620-c19e-11ea-8e31-7fd6e6dac888 |
There was a problem hiding this comment.
Go through the CONTRIBUTING.md and make sure it passes the flake8 style standard
cclauss
added
tests are failing
|
@QuantumNovice Your review please. |
Oops I missed that when writing power iteration. The rayleigh quotient I wrote within it is a duplicate; I will remove. But power iteration is not a duplicate. Thanks! |
OK will do. |
| def rayleigh_quotient(input_matrix: np.array, vector: np.array) -> float: | ||
|
|
||
| """ | ||
| Rayleigh Quotient. | ||
| Fine the Rayliegh quotient of matrix and vector. Given an eigenvector, | ||
| the Rayleigh quotient is the corresponding eigenvalue. | ||
|
|
||
| Input | ||
| input_matrix: input matrix. | ||
| Numpy array. np.shape(input_matrix) == (N,N). | ||
| vector: input vector. | ||
| Numpy array. np.shape(vector) == (N,) or (N,1) | ||
| """ | ||
|
|
||
| num = np.dot(vector.T, np.dot(input_matrix, vector)) | ||
| den = np.dot(vector.T,vector) | ||
|
|
||
| return num/den |
There was a problem hiding this comment.
Can you remove these lines and use this instead?
from rayleigh_quotient import rayleigh_quotientThere was a problem hiding this comment.
Working on this now, thank you!
|
@QuantumNovice I actually removed Rayleigh quotient since in the power iteration algorithm the vectors are already normalized. This removes a dot product and division operation. I had not written doctests before so I referenced what was happening in Rayleigh Quotient and tried to emulate that. I am using VS Code now and set Thanks all for your help! |
Changed convergence check line. Co-authored-by: Christian Clauss <[email protected]>
| Numpy array. np.shape(largest_eigenvector) == (N,) or (N,1). | ||
|
|
||
| >>> import numpy as np | ||
| >>> A = np.array([ |
There was a problem hiding this comment.
Please use the same variable names as the function parameters so the reader does not get lost.
There was a problem hiding this comment.
Thanks @cclauss this is a great point; I made the change :)
| eigs, eigvs = np.linalg.eigh(A) | ||
| eig_max = eigs[-1] | ||
| eigv_max = eigvs[:, -1] |
There was a problem hiding this comment.
Longer variable names would improve readability for learners.
There was a problem hiding this comment.
I made changes to be more clear and added some comments as well. Appreciate it!
Named tests more clearly. Co-authored-by: Christian Clauss <[email protected]>
|
Hey @zakademic, TravisCI finished with status TravisBuddy Request Identifier: 1f69e7e0-cf02-11ea-9502-ed86ad10d0d8 |
…octests to match function call.
…akademic/Python into linear_algebra/power_iteration
|
Nice!! |
* Initial commit of power iteration. * Added more documentation for power iteration and rayleigh quotient * Type hinting for rayleigh quotient * Changes after running black and flake8. * Added doctests, added unit tests. Removed Rayleigh quotient as it is not needed. * Update linear_algebra/src/power_iteration.py Changed convergence check line. Co-authored-by: Christian Clauss <[email protected]> * Update linear_algebra/src/power_iteration.py Named tests more clearly. Co-authored-by: Christian Clauss <[email protected]> * Changed naming in test function to be more clear. Changed naming in doctests to match function call. * Self running tests Co-authored-by: Zeyad Zaky <[email protected]> Co-authored-by: Christian Clauss <[email protected]>
Describe your change:
Added power iteration algorithm. Given a symmetric matrix and a random vector of same space, will find largest eigenvalue and corresponding eigenvector. Rayleigh quotient needed to compute eigenvalue given an eigenvector.
Checklist:
Fixes: #{$ISSUE_NO}.