petbox-dev / dca

100%

Build Type

push

github

Committed by dsfulf

Commit Message

code-fix: close three shut-in validation holes and correct the measured claims

The 2.3.0 shut-in work established one invariant -- a rate of zero has no
decline -- and left three ways through it. All three are fixed here, each found
by a fresh reader of the diff and each confirmed by running it before believing
it.

- The "restart after a shut-in must state its own D" gate read the immediately
  preceding segment's STATED `q`. A segment that merely INHERITS a shut-in has
  no stated rate, so shut-in -> inheriting continuation -> restart walked past
  the gate and produced a flat 500 BPD forever. It now tracks the resolved
  incoming rate through the chain.

- The same rule's mirror: a segment could state `D` without restating `q` while
  the resolved rate was zero, so `GeneralizedHyperbolic(0.0, 0.8, 1.5,
  (HyperbolicSegment(100.0, D=0.5),))` reported D = 0.00222 for a well whose
  rate is zero at every time. Also keyed on the resolved rate now, and its
  message says "is shut in" rather than "has q == 0", since the segment need
  not have said so itself.

- `_append_terminal_segment` returned silently whenever row 0 did not produce,
  which is right for a well that never flows and wrong for one that comes back:
  a `qi = 0` model restarting at 100 days lost its Dterm without a word, giving
  rate(1e5) = 1.82 uncapped against 1.3e-08 capped. Keyed on whether ANY row
  produces.

Two existing tests move to the stricter rule -- both had been reaching the
generic `D == 0 requires b == 0` contradiction by a weaker route, and now trip
the specific one first.

The measured claims are corrected to a stated basis. The step-accuracy table
quoted 2.4e-07 where the max across the evaluation array is 3.1e-06 -- the
smaller figure was the straddle alone at the two earliest points, before local
refinement, which is a different and flattering measurement. 4.3e-03 was
likewise the error at one point rather than the max, 5.5e-03. Both files now
name the arrays the f... (continued)

Coverage Stats

34 of 34 new or added lines in 3 files covered. (100.0%)

3 existing lines in 1 file now uncovered.

1162 of 1166 relevant lines covered (99.66%)

1.0 hits per line