FlexELA
Loading...
Searching...
No Matches
Output Files

Volume Tracking Matrix

Contains the (non-normalized) volume tracking matrix (VTM) \(\mathbf{Q}^{(n-1\rightarrow n)}\) and is output by calling ELA_OutputWriteVTM(). The VTM is stored in CSR matrix format. The first entry is the row count (RC) and the second is the number of non-zero entries (NNZ).

Description Type
RC uint32_t
NNZ uint32_t
ROW_INDEX uint32_t[RC]
COLUMN_INDEX uint32_t[NNZ]
VALUES double[NNZ]

Naming

This file is named afwd_[n].bin, where n is the snapshot index and is zero-padded to 6 digits.

Notes

  • The column count (CC) is not given by this file. If required, obtain from `timelog.bin`.
  • The zero entry that formally starts ROW_INDEX is omitted.
  • To verify the file, check that ROW_INDEX(end)==NNZ.

Volume Vector

Contains the volume vector \(\mathbf{v}^{n}\) (see [Gaylo2022], Eq. 8) and is output by calling ELA_OutputWriteV. The first entry gives \(M^{n}\), the row count (RC). This is followed values corresponding to \(v_l^{n}\) for \(l\in 1\dots M^{n}\), which may contain zeros.

Description Type
RC uint32_t
VALUES double[RC]

Naming

This file is named v_[n].bin, where n is the snapshot index and is zero-padded to 6 digits.

timelog.bin

This file is appended to each time ELA_OutputWriteVTM() is called and records \( n \), the row count (RC) \( M^{n} \), and the time \( t^{n} \).

Description Type
Index, \( n \) uint32_t
RC, \( M^{n} \) uint32_t
Time, \( t^{n} \) double

tracking.log

This file's purpose is to allow one to monitor the volume conservativeness of the tracking data. When ELA_OutputLog() is called, the following data is appended:

Time Max Label Max Source Entry Min Source Entry Volume Error (abs) Volume Error (rel) Max NNZ
$$t^n$$ $$M^{n}$$ $$1-\max\left[\mathbf{s}^{n}_{ijk}\right]$$ $$\min\left[\{\mathbf{s}^{n}_{ijk} : \mathbf{s}^{n}_{ijk}\ne 0\}\right]$$ $$\sum_{ijk}\left[\Omega_{ijk} \sum_l {(s^{n}_l)}_{ijk} - \Omega_{ijk} (1-f_{ijk})\right]$$ $$\frac{\text{Volume Error (abs)}}{\sum_{ijk}\left[\Omega_{ijk} (1-f_{ijk})\right]}$$ $$\max\text{nnz}[\mathbf{s}^{n}]\}$$