Crate m4ri_sys 

Implements the FFI to M4RI

See https://bitbucket.org/malb/m4ri

Structs

Code

Djb

Mzd

Represents the Mzd data type used by M4RI

Mzp

Enums

Srctyp

Statics

MZD_FLAG_MULTIPLE_BLOCKS

Flag for multiple blocks

MZD_FLAG_NONZERO_EXCESS

Flag when ncols%64 == 0

MZD_FLAG_WINDOWED_OWNSBLOCKS

Flag for windowed matrix which owns its memory

MZD_FLAG_WINDOWED_ZEROEXCESS

Flag for windowed matrix where ncols % 64 == 0

MZD_FLAG_WINDOWED_ZEROOFFSET

Flag for windowed matrix

Functions

Mzp_free_window^⚠

Free a permutation window created with Mzp_init_window

_mzd_mul_naive^⚠

brief Naive cubic matrix multiplication with the pre-transposed B.

_mzd_mul_va^⚠

Matrix multiplication optimized for v*A where v is a vector

djb_apply_mzd^⚠

apply the linear map m to V and write the result in W

djb_compile^⚠

Compile an new DJB linear map from A

djb_free^⚠

free a DJB linear map

djb_info^⚠

Print information on linear map mA

djb_init^⚠

Allocate a new DJB linear map

djb_push_back^⚠

Add a new operation out[target] ^= srctype[source] to queue

m4ri_build_all_codes^⚠

Generates global code book

m4ri_build_code^⚠

Fils var ord and var inc with Gray code dat afor a Gray code of length 2^l

m4ri_destroy_all_codes^⚠

Destroy global code book

m4ri_gray_code^⚠

Returns the ith Gray code entry for a gray code of length 2^l

m4ri_opt_k^⚠

Return the optimal var k for the given parameters

mzd_add^⚠

Set C = A + B If C is passed in, the result is written there otherwise a new matrix is created

mzd_addmul^⚠

\brief Matrix multiplication and in-place addition via the Strassen-Winograd matrix multiplication algorithm, i.e. compute C = C+ AB.

mzd_addmul_m4rm^⚠

Set C to C + AB using Konrods Method

mzd_addmul_mp^⚠

Matrix multiplication and in-place additoin via th ecubic matrix multiplication algorithm on multiple cores. C = C + AB

mzd_addmul_naive^⚠

naive cubic matrix multiplication and addition

mzd_apply_p_left^⚠

Apply the permutation P to A from the left

mzd_apply_p_left_trans^⚠

Apply the permutation P to A from the left, but transpose P

mzd_apply_p_right^⚠

Apply the permutation P to A from the right

mzd_apply_p_right_trans^⚠

Apply the permutation P to A from the right, but transpose P

mzd_col_swap^⚠

Swap the two columns cola and colb

mzd_concat^⚠

Concatenate B to A and write the result to C

mzd_copy^⚠

Copy a matrix to dest

mzd_copy_row^⚠

\brief copy row j from A to row i from B.

mzd_echelonize^⚠

(Reduced) row echelon form

mzd_echelonize_m4ri^⚠

Matrix elimination using the method of the four russians

mzd_echelonize_pluq^⚠

(Reduced) row echelon form using PLUQ factorisation

mzd_equal^⚠

Return true if A == B

mzd_first_row^⚠

Get a pointer to the first word

mzd_free^⚠

Free a matrix created with mzd_init. Automatically done by the Deref trait on Mzd

mzd_free_window^⚠

Free a matrix window created with mzd_init_window

mzd_free_window_const^⚠

Free a “const” window created with mzd_init_window_const.

mzd_init^⚠

Create a new rows x columns matrix

mzd_init_window^⚠

\brief Create a window/view into the matrix M.

mzd_init_window_const^⚠

Create a const window/view into a const matrix

mzd_inv_m4ri^⚠

Invert the matrix using Konrod’s method

mzd_invert_naive^⚠

Invert the target matrix using gaussian elimination To avoid recomputing the identity matrix over and over again, I may be passed in as identity parameter The first parameter may be null to have the space automatically allocated

mzd_is_windowed^⚠

Test if a matrix is windowed

mzd_is_zero^⚠

Zero test for matrix

mzd_make_table^⚠

Constructs all possible 2^k row combinations using the gray code table

mzd_mul^⚠

\brief Matrix multiplication via the Strassen-Winograd matrix multiplication algorithm, i.e. compute C = AB.

mzd_mul_m4rm^⚠

Matrix multiplication using Konrods Method

mzd_mul_mp^⚠

Matrix multiplication via the cubic multiplication algorithm on multiple cores

mzd_mul_naive^⚠

naive cubic matrix multiplication the first argument may be null for automatic creation

mzd_owns_blocks^⚠

Test if this mzd_t should free blocks

mzd_ple^⚠

PLE matrix decomposition.

mzd_pluq^⚠

PLUQ matrix decomposition.

mzd_pluq_solve_left^⚠

Solves (P L U Q) X = B

mzd_process_rows^⚠

The function looks up k bits from position i, startcol in each row and adds the appropriate row from T to the row i

mzd_process_rows2^⚠

Same as mzd_process_rows but works with two Gray code tables in parallel

mzd_process_rows3^⚠

Same as mzd_process_rows but works with three Gray code tables in parallel

mzd_process_rows4^⚠

Same as mzd_process_rows but works with four Gray code tables in parallel

mzd_process_rows5^⚠

Same as mzd_process_rows but works with five Gray code tables in parallel

mzd_process_rows6^⚠

Same as mzd_process_rows but works with six Gray code tables in parallel

mzd_randomize^⚠

Fill the matrix m with uniformly distributed bits.

mzd_read_bit^⚠

Read the bit at position M[row, col]

mzd_row^⚠

Get pointer to first word of row

mzd_row_clear_offset^⚠

Clear the given row, but only begins at the column coloffset.

mzd_row_swap^⚠

Swap the two rows rowa and rowb

mzd_set_ui^⚠

Set to identity matrix if the second argument is 1

mzd_solve_left^⚠

Solves A X = B with A and B matrices.

mzd_stack^⚠

Stack A on top of B into C

mzd_sub^⚠

Set C = A - B If C is passed in, the result is written there otherwise a new matrix is created

mzd_submatrix^⚠

Copy a submatrix first argument may be preallocated space or null

mzd_top_echelonize_m4ri^⚠

Matrix elimination using the Method of the four russians (m4ri)

mzd_transpose^⚠

Transpose a matrix Dest may be null for automatic creation

mzd_write_bit^⚠

Write the bit to position M[row, col]

mzp_copy^⚠

Copy permutation Q to P Target may be null

mzp_free^⚠

Free an Mzp

mzp_init^⚠

Construct an identity permutation

mzp_init_window^⚠

Create a window into the permutation

mzp_print^⚠

Print the mzp

mzp_set_ui^⚠

Set the permutation to the identity permutation

Type Aliases

BIT

M4RI Internal representation

Rci

M4RI Internal representation

Word

M4RI Internal representation