log.txt

1/20/16


TODO:
Change DataFrame to a list,
Only add monomials if not relatively prime.
Add them to the list, add new an necessary pieces to the heap
Make sure you keep track of the number of functions, s-polynomials, and monomials in the matrix for future use

Finished:
Add the S-polynomials to said matrix
look in to how to sort a matrix based on the sorting of a list.
    HERE WE ARE GOING TO USE np.argsort
Built a max heap


12/30/16 
Todo today:
Change DataFrame index to lisat
Check for relative primality

1/19/17:
todo:
Abananoed using dicts for now, want to make other progress.
Pushed everything into a module, importable structure.
Built a Power Basis module to test the code


1/20/17:
TODO: 
Need to make sys append your current path directory instead of just mine, that way it works for any repo
Figure out reduction for F4 algorithm, and final steps
Set an attribute to take in [tuple(c)] and extend the size of the matrix
Handle case when len of poly and monomials in add_r_to_matrix are different lengths
Make a function for adding a polynomial to the matrix, and use that all over


1/27/17:
TODO:
Make a function for adding a polynomial to the matrix, and use that all over


2/3/17:
TODO:
Add a monomial multiplication that is more numerically stable that just uses shifting of the matrix
Make conditions to only check new f's and s's, since the f's and old s's don't need to be checked against eachother. 
Remove entirely 0 rows and columns when doing multiplication
Verify that the loop is converging to the correct grobner basis, and then verify it is working in cheb basis




Q's for the big man:
1. Do the f's and old s's need to be checked in each iteration of the algorithm?
2. When I do the LU Decomposition, do I just take the PLU, and grab the columns I care about?