[remislp]

Second half of M3: the binning arithmetic comes from the shared record layer,
ekplot keeps the drawing. Depends on DCPROGS/dcio#3.

This is a bug fix, not only a deduplication

The bin edges were built here, and the upper limit was computed as
exp(ceil(log(max(X)))) — a power of e, where the whole scheme is bins per
decade. That limit can land below the longest interval, and
np.histogram then drops the tail of the distribution without saying so.

Across 400 randomly drawn exponential samples the top edge fell below max(X)
in 119 of them. On glyr_experimental.scn it happens not to bite, which is
presumably how it survived.

What stays

__exponential_scale_factor looked like a third copy of HJCFIT's
ideal_pdf_scale_factor and is a different quantity — this one normalises an
exponential mixture already fitted to these data, HJCFIT's renormalises an
ideal pdf onto the resolved intervals from the Q-matrix survival. Both stay
where they are, and the docstring now says why.

prepare_xlog_hist keeps its signature and its list return type, so callers
and notebooks are unaffected.

Dependency

dcio is declared in install_requires. It is not on PyPI yet, so a source
install needs pip install git+https://github.com/DCPROGS/dcio first — the
comment in setup.py says so. That constraint goes away when dcio is
published.

Tests

6 new, including a regression that checks every interval is counted and
asserts the sample still exercises the old bug. 36 passing.

🤖 Generated with Claude Code