Skip to content

Latest commit

 

History

History
119 lines (95 loc) · 3.73 KB

directory.md

File metadata and controls

119 lines (95 loc) · 3.73 KB
Raw

Directories on Blue Waters

Read process overview before this. This file explains necessary conventions to decipher the filenames on Google Drive https://drive.google.com/drive/folders/1zYv2R-oqepX1vJ_Fr5JBmrVNdle0mi9M.

The project ID of our computation is baus. The following is a note on what is inside the directory /projects/sciteam/baus.

  • helloworld, rm16test, rm17pivot, rm32pivot

    Testing by repeating the computation of something easy. These jobs can be done, and had been done, on my laptop.

  • rm32rref, rm33rref

    Trial computation. Debug (overflow), optimize the code (pass by reference), and check against the pivot-signatutre polynomial. The main reason we give up these folders is pure stupidity #define five 4.

  • rm34rref

    The RREF-signature polynomial of RM32 is computed here. The computation is basically enumerating all 2^32 subsets. That resulting polynomial has 17,818,745 monomials. Each monomial is encoded as a line so the resulting file, rm32rref.txt, has 17,818,745 lines. After compressing, rm32rref.zip is 68.7 MB.

    • rm32rref.txt

    • rm32rref.zip

  • rm35rref

    Since rm32rref.txt has 17,818,745 lines, we divide them evenly into 311 files of the form rm35rref123.txt or rm35rref${i}.txt in general . Each file has 57,295 lines, which represents 57,295 monomials out of 17,818,745 monomials.

    • rm35rref000.txt

    • rm35rref001.txt

    • vdots

    • rm35rref310.txt

  • rm64square, rm65square, rm66square, rm67square, rm68square, rm69square, rm70square, rm70tutte

    Wrong computation. (Due to overflow of bitfiled.)

  • rm71square

    From the RREF-signature polynomial of RM32 we want to compute the pivot-signature polynomial of RM64. The computation is like computing the square of the polynomial rm32rref.txt. Since we had divided the polynomial rm32rref.txt into 311 subpolynomials, it suffices to compute the product of any two subpolynomials. The resulting files are called rm64pivot123x234.txt or rm64pivot${i}x$(( ($i+$j)%310 )).txt in general. Here $i runs from 000 to 310; and $j runs from 000 to 155.

    • rm64pivot000x000.txt

    • rm64pivot000x001.txt

    • vdots

    • rm64pivot000x155.txt

    • (Note that 000x156 does not exist because it is named 156x000)
    • (Note that 001x000 does not exist because 000x001 does.)

    • rm64pivot001x001.txt

    • rm64pivot001x002.txt

    • vdots

    • rm64pivot001x156.txt

    • (Note that 001x157 does not exist because it is named 157x001)
    • (Note that 002x000 does not exist because 000x002 does.)
    • (Note that 002x001 does not exist because 001x002 does.)

    • rm64pivot002x002.txt

    • rm64pivot002x003.txt

    • vdots

    • rm64pivot002x157.txt

    • VDOTS

    • rm64pivot310x310.txt

    • rm64pivot310x000.txt

    • vdots

    • rm64pivot310x154.txt

  • rm72square

    Recall that the pivot-signature polynomial of RM64 is computed distributedly and stored across 48,516 files. (48,516 = 1 + 2 + ... + 311.) We gather them and store the final result in this file. It has 7,828,354 lines/monomials; and is 274 MB. After compressing 42.7 MB.

    • rm64pivotall.txt

  • rm72tutte

    From the pivot-signature polynomial of RM64 we can compute the Tutte polynomials. In particular that of the [64,22,16]-Reed-Muller codes.

    • rm72tutte.txt

  • rm73square

    Wrong zip files.

  • rm74square

    The list of zip files of the form 123.zip. A file like 123.zip is the compression of files rm64pivot123*. That is, each zip file contains 156 txt files, starting from $i to $(( ($i+155)%310 )).

    • 000.zip

    • 001.zip

    • vdots

    • 310.zip