UMFInfo

class sksparse.umfpack.UMFInfo[source]

A data class to store UMFPACK info.

status[source]

Return status of the last UMFPACK call.

Type:

int

n_row[source]

Number of rows in the input matrix.

Type:

int

n_col[source]

Number of columns in the input matrix.

Type:

int

nz[source]

Number of nonzeros in the input matrix.

Type:

int

size_of_unit[source]

Size of a unit in bytes.

Type:

int

size_of_int[source]

Size of an int32_t in bytes.

Type:

int

size_of_long[source]

Size of an int64_t in bytes.

Type:

int

size_of_pointer[source]

Size of a void * pointer in bytes.

Type:

int

size_of_entry[source]

Size of an entry in bytes, real or complex.

Type:

int

ndense_row[source]

Number of dense rows in the input matrix.

Type:

int

nempty_row[source]

Number of empty rows in the input matrix.

Type:

int

ndense_col[source]

Number of dense columns in the input matrix.

Type:

int

nempty_col[source]

Number of empty columns in the input matrix.

Type:

int

symbolic_defrag[source]

Number of memory compactions performed.

Type:

int

symbolic_peak_memory[source]

Peak memory usage during symbolic factorization.

Type:

int

symbolic_size[source]

Size of symbolic factorization, in Units.

Type:

int

symbolic_time[source]

Time spent in symbolic factorization, in seconds.

Type:

float

symbolic_walltime[source]

Wall-clock time spent in symbolic factorization, in seconds.

Type:

float

strategy_used[source]

Strategy used in the factorization. One of: {"auto", "unsymmetric", "symmetric"}.

Type:

str

ordering_used[source]

Ordering method used in the factorization. One of: {"cholmod", "amd", "given", "none", "metis", "best", "user", "metis_guard"}

Type:

str

qfixed[source]

Whether the column permutation Q was fixed.

Type:

bool

diag_preferred[source]

Whether diagonal pivoting was preferred.

Type:

bool

pattern_symmetry[source]

Symmetry of the nonzero pattern of the input matrix, excluding dense rows and columns (aka \(S\)).

Type:

float

nz_a_plus_at[source]

Number of nonzeros in \(S + S^{\top}\), excluding the diagonal.

Type:

int

nzdiag[source]

Number of nonzeros on the diagonal of \(S\).

Type:

int

symmetric_lunz[source]

Number of non-zeros in \(L + U\), if AMD ordering was used.

Type:

int

symmetric_flops[source]

Number of floating-point operations for the factorization, if AMD ordering was used.

Type:

int

symmetric_ndense[source]

Number of dense rows and columns in \(S + S^{\top}\).

Type:

int

symmetric_dmax[source]

Maximum number of entries in any column of \(L\), for AMD.

Type:

int

col_singletons[source]

Number of column singletons.

Type:

int

row_singletons[source]

Number of row singletons.

Type:

int

n2[source]

Size of \(S\).

Type:

int

s_symmetric[source]

1 if \(S\) is square and symmetrically permuted.

Type:

int

numeric_size_estimate[source]

Estimated size of numeric factorization, in Units.

Type:

int

peak_memory_estimate[source]

Estimated peak memory usage during numeric factorization.

Type:

int

flops_estimate[source]

Estimated number of floating-point operations for the factorization.

Type:

int

lnz_estimate[source]

Estimated number of nonzeros in \(L\).

Type:

int

unz_estimate[source]

Estimated number of nonzeros in \(U\).

Type:

int

variable_init_estimate[source]

Initial size of memory usage in numeric factorization.

Type:

int

variable_peak_estimate[source]

Peak size of memory usage in numeric factorization.

Type:

int

variable_final_estimate[source]

Final size of memory usage in numeric factorization.

Type:

int

max_front_size_estimate[source]

Maximum frontal matrix size, estimated.

Type:

int

max_front_nrows_estimate[source]

Maximum number of rows in any frontal matrix, estimated.

Type:

int

max_front_ncols_estimate[source]

Maximum number of columns in any frontal matrix, estimated.

Type:

int

numeric_size[source]

Size of numeric factorization, in Units.

Type:

int

peak_memory[source]

Peak memory usage during symbolic and numeric factorization.

Type:

int

flops[source]

Number of floating-point operations for the factorization.

Type:

int

lnz[source]

Number of nonzeros in \(L\).

Type:

int

unz[source]

Number of nonzeros in \(U\).

Type:

int

variable_init[source]

Initial size of memory usage in numeric factorization.

Type:

int

variable_peak[source]

Peak size of memory usage in numeric factorization.

Type:

int

variable_final[source]

Final size of memory usage in numeric factorization.

Type:

int

max_front_size[source]

Maximum frontal matrix size.

Type:

int

max_front_nrows[source]

Maximum number of rows in any frontal matrix.

Type:

int

max_front_ncols[source]

Maximum number of columns in any frontal matrix.

Type:

int

numeric_defrag[source]

Number of memory compactions performed.

Type:

int

numeric_realloc[source]

Number of memory reallocations performed.

Type:

int

numeric_costly_realloc[source]

Number of costly memory reallocations performed.

Type:

int

compressed_pattern[source]

Number of integers in LU pattern.

Type:

int

lu_entries[source]

Number of real entries in \(L\) and \(U\).

Type:

int

numeric_time[source]

Time spent in numeric factorization, in seconds.

Type:

float

nz_udiag[source]

Number of nonzeros on the diagonal of \(U\).

Type:

int

rcond[source]

Estimate of the reciprocal of the condition number of \(A\).

Type:

float

was_scaled[source]

Scaling method used. One of: {"none", "sum", "max"}.

Type:

str

rsmin[source]

min(max(row)) or min(sum(row)), depending on the scaling method.

Type:

float

rsmax[source]

max(max(row)) or max(sum(row)), depending on the scaling method.

Type:

float

umin[source]

Minimum absolute value of a diagonal entry of \(U\).

Type:

float

umax[source]

Maximum absolute value of a diagonal entry of \(U\).

Type:

float

alloc_init_used[source]

Initial memory allocation used, as a fraction of total numeric memory.

Type:

float

forced_updates[source]

Number of forced updates during numeric factorization.

Type:

int

numeric_walltime[source]

Wall-clock time spent in numeric factorization, in seconds.

Type:

float

noff_diag[source]

Number of off-diagonal pivots.

Type:

int

all_lnz[source]

Total number of entries in \(L\), if no dropped entries.

Type:

int

all_unz[source]

Total number of entries in \(U\), if no dropped entries.

Type:

int

nzdropped[source]

Number of dropped entries in \(L\) and \(U\).

Type:

int

ir_taken[source]

Number of iterative refinement steps taken.

Type:

int

ir_attempted[source]

Number of iterative refinement steps attempted.

Type:

int

omega1[source]

Factor for sparse backdward error estimate.

Type:

int

omega2[source]

Factor for sparse backdward error estimate.

Type:

int

solve_flops[source]

Number of floating-point operations for solve.

Type:

int

solve_time[source]

Time spent in solve, in seconds.

Type:

float

solve_walltime[source]

Wall-clock time spent in solve, in seconds.

Type:

float

Added in version 0.5.0.

Methods

__init__