Repository navigation
paper: one meaning per symbol, theorems as bounds, references and layout - #115
Conversation
|
Important Review skippedAuto reviews are disabled on base/target branches other than the default branch. Please check the settings in the CodeRabbit UI or the
|
[Low impact] This PR appears safe to merge; no actionable issues were found. SummaryThis PR cleans up the paper’s notation, theorem statements, references and layout without changing the arithmetic implementations or computed bounds.
No actionable issues found. Reviews (1) · Last reviewed commit: "paper: rebuild the PDF" · Reviewed by Greptile |
Lambda, beta, alpha, gamma, zeta, pi and f each stood for several things, and R and Delta were absolute in Lemmas 11 and 13 and normalized in Theorem 14. The appendices rename all but one use of each, write the number pi upright, and the scripts' docstrings name the paper's symbols.
Each of these theorems stated a recursion for computing a bound. The recursion now stands in a paragraph before the theorem, which states the bound and its computed value. The numbering is unchanged.
Algorithm 2 took c_0 to c_N with N at most n, and Appendix B and Theorem 9 take N terms c_0 to c_(N-1) with N at most n + 1.
The text had its last parts near 2^(-2j-320). The last of eight parts of 1 - r^2/2 is about u^6 r^2 / 2, near 2^(-2j-319).
The unit, the crate's OD::EPS, was used in a caption and a remark and never defined.
The trigonometric functions run DWTimesDW3 and exp and ln DWTimesDW2. The text now says how they differ and that each appendix's constants are those of its product.
Table I named the operand x for the sine, cosine and tangent and a elsewhere, and Table II listed the operations in another order than Table I.
Shewchuk for expansions, Moller and Knuth for TwoSum, Sterbenz for the exact subtractions and Payne and Hanek for the reduction of large arguments were used without a reference. The list is now in order of first citation, with page ranges for the three ACM articles, the Universal pull request's full title, the crate's version and commit, and the QD authors' release.
The number 3.5 is that of the HAL preprint, as the crate's comment says. The published article could not be opened to confirm its numbering.
Assume with a bare reference printed as Assume 6, and IEEEtran's paragraph headings printed labels a) to d) that nothing refers to. The appendices now write Under Assumption 6 and use an unlabeled run-in heading.
Figure 1's labels were about 5 pt after scaling, the appendix started mid-column, Table V floated two pages past its text, and two long inline chains were stretched across the column. The figure is redrawn larger, the appendix starts on a new page, the table moves before its section, and the chains are displays.
Lemma 33 wrote f_k for the real that the expansion F_k approximates, another use of f. It is now F_k with the superscript e that the lemma's other exact values carry.
The PDF of the notation, references and layout changes.
b31a0a4 to
989b669
Compare
Codecov Report✅ All modified and coverable lines are covered by tests. Additional details and impacted files@@ Coverage Diff @@
## paper/referee-campaign #115 +/- ##
=========================================================
Coverage ? 97.15%
=========================================================
Files ? 21
Lines ? 13794
Branches ? 0
=========================================================
Hits ? 13401
Misses ? 393
Partials ? 0 ☔ View full report in Codecov by Harness. 🚀 New features to boost your workflow:
|
Stacked on #114. Fixes the referee's findings on notation, references and presentation: 21, 22, 26, 29, 30, 31, 37, 39, 40, 42, 43, 44 and 45, and what was left of 24.
Statements
Finding 40: Theorems 19, 21, 22, 40 and 42 each stated a recursion. The recursion now stands in an unnumbered paragraph before the theorem ("The constant kappa_DIV", "The constant delta_J", "The constant delta^DD", "The bound of a case", "The bound E(l)"), and the theorem states the bound and its computed value. No hypothesis, bound or constant changes, and the numbering of every theorem, lemma and table is unchanged, so the numbers cited in
src/,tests/, the READMEs and the scripts stay correct.Finding 21: Algorithm 2 now takes N terms
c_0, ..., c_(N-1)withN <= n + 1, as Appendix B and Theorem 9 do. It tookc_0, ..., c_NwithN <= n.Notation (finding 22)
c_r(1+u)^(n-1)/(1-u)is written out; the accumulator's1/(1-(n+1)Ku)isGamma; Lemma 38's isepsilon_kA-hat_m; the double-word bounds aree_A, e_F, e_M, e_P, as the trig appendix'se_oppi^g; Lemma 38's tail bound isx-bar_t; LOG's value of step (5) isy_c|Z|/MZ-bar_c; the part bounds arepi-hat^z,pi^sandchi''w_im_1 m_2; MUL's chain writess_n; LOG'sx' - 1isvarphi; Horner's exact coefficients areF^e_kR_n,Delta_nThe scripts' docstrings name the paper's symbols; no code name changes.
explog_bits.py scannow printsx_0 c_r.References (findings 37 and 39)
New entries, each checked against Crossref or Open Library: Shewchuk (Discrete Comput. Geom. 18(3), 305-363, 1997), Møller (BIT 5(1), 37-50, 1965), Knuth (TAOCP 2, 3rd ed., ISBN 0-201-89684-2), Sterbenz (Prentice-Hall, ISBN 0-13-322495-3) and Payne and Hanek (SIGNUM Newsl. 18(1), 19-24, 1983). The list is in order of first citation. The three ACM TOMS entries have Crossref's page ranges (1-27, 1-24, 1-34), the Universal entry has the pull request's full title, the crate's entry has its version, commit and date, and the QD entry cites the authors' repository (BL-highprecision/QD, v2.3.24), where I read
expandlogto confirm the squaringss = 2s + s^2and the one Newton step.Presentation
Findings 26, 29, 30, 31: the cosine's last part is given as
u^6 r^2/2, near2^(-2j-319); the undefined "units of 2^-418" are gone; the exp and ln appendix says how DWTimesDW2 and DWTimesDW3 differ; Table I usesafor every operand and Table II has its row order.Findings 42 to 45: Figure 1 is redrawn with labels about twice the size, the appendix starts on a new page, Table V lands on the page of its text, two long inline chains are displays, "Assume 6" reads "Under Assumption 6", and the paragraph headings carry no a) to d) labels. I rendered and read pages 1 to 3, 6, 7, 9 to 11, 13, 14, 17, 21 and 22.
Not done
Finding 32, the accumulator's doc in
src/od.rs, is outsidepaper/and is left alone. The comment atsrc/trig.rs:630-633has2^(-2j-318)and2^-82where the paper now has2^(-2j-319)and the script's2^-81.Still unverified
HAL and the ACM Digital Library refuse automated access, so these rest on secondary sources or are unconfirmed:
src/andREADME.md. Their bounds 3u^2/(1-4u), 2u^2, 5u^2/(1+u)^2, (4u^2 + u^3/2)/(1+u)^2 and 15u^2 + 56u^3 (p >= 7), and the algorithm numbers 4, 6, 7, 9, 11, 12 and 17, agree with the Coq files of Muller and Rideau's formalization as distributed in VeriNum/double-double and with slides of J.-M. Muller;Verified here and not before: the QD paper's statements, read in the text of the authors' technical report; the page ranges of the ITP 2017, ARITH-23 and triple-word entries; the ISBN of Cody and Waite.