gulps.analysis.coverage.CoverageReport

class gulps.analysis.coverage.CoverageReport(rows, decomposer)[source]

Bases: object

Haar coverage and cost of a fixed ISA, integrated in floating point.

Parameters:
  • rows (tuple[SentenceCoverage, ...]) – Every sentence the search emitted through the last entry, in search order, including those whose region earlier sentences already cover.

  • decomposer (GulpsDecomposer) – The instruction set the search ran on.

property entries: tuple[SentenceCoverage, ...]

The cost-ordered rows that reach classes no earlier row reaches.

property expected_cost: float

The Haar-weighted cost summed over the reported regions.

property cost_std: float

The Haar-weighted standard deviation of cost over the reported regions.

property total_coverage: float

The covered Haar fraction, within numerical tolerance of 1.

plot()

Draw each entry’s reachable region in its own Weyl-chamber subplot.

Return type:

matplotlib.figure.Figure | None

plot_tree(names=None)

Draw the search rows and the candidates pruned among them as a tree.

Parameters:

names (Sequence[str] | None) – A display name for each gate of the decomposer, in order. Defaults to each gate’s label, or its name.

Return type:

matplotlib.figure.Figure

property cost_cdf: list[tuple[float, float]]

(cost, cumulative Haar fraction) at each cost, ascending.

percentile(q)

The cost at which the cumulative Haar fraction first reaches q.

percentile(1.0) returns the endpoint of the reported distribution, not a guaranteed worst-case cost: small uncovered regions can remain when integration stops. Returns infinity if q exceeds the reported coverage by more than its numerical tolerance.

Parameters:

q (float) – A cumulative probability from 0 through 1.

Return type:

float

plot_reach(ax=None)

Plot cumulative Haar coverage against cost on ax, or on new axes.

Parameters:

ax (matplotlib.axes.Axes | None)

Return type:

matplotlib.axes.Axes