Integrate BigDecimal_div and BigDecimal_div2 #329
Merged
Conversation
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
…ecified Fixes exponential precision grouth. `max(divisor_prec, dividend_prec) + extra_prec` is enough for division precision.
tompng
commented
May 27, 2025
| GUARD_OBJ(cv, NewZeroWrapLimited(1, ix + 3 * VpBaseFig())); | ||
|
|
||
| mx = bv->Prec + cv->MaxPrec - 1; | ||
| if (mx <= av->Prec) mx = av->Prec + 1; |
There was a problem hiding this comment.
Correct allocation size calculation of res.
VpDivd requires word_a < word_r && word_b + word_c - 2 < word_r
tompng
commented
May 27, 2025
| ssize_t a_prec, b_prec; | ||
| BigDecimal_count_precision_and_scale(av->obj, &a_prec, NULL); | ||
| BigDecimal_count_precision_and_scale(bv->obj, &b_prec, NULL); | ||
| ix = ((a_prec > b_prec) ? a_prec : b_prec) + BIGDECIMAL_DOUBLE_FIGURES; |
There was a problem hiding this comment.
max(a_prec, b_prec) + extra_prec is enough. Fixes exponential precision growth.
tompng
commented
May 27, 2025
| @@ -6164,7 +6119,7 @@ VpDivd(Real *c, Real *r, Real *a, Real *b) | |||
| word_c = c->MaxPrec; | |||
| word_r = r->MaxPrec; | |||
|
|
|||
| if (word_a >= word_r) goto space_error; | |||
| if (word_a >= word_r || word_b + word_c - 2 >= word_r) goto space_error; | |||
There was a problem hiding this comment.
There are two more goto space_error in this function.
One is on L6218
ind_r = ind_c + ind_b;
if (ind_r >= word_r) goto space_error;And another in L6253
ind_r = b->Prec + ind_c - 1; // L6242
ind_r = b->Prec + ind_c; // L6250 (ind_c + 1 < word_c is ensured in L6247)
if (ind_r >= word_r) goto space_error;Both has the same requirement: (word_b-1) + (word_c-1) < word_r
tompng
commented
May 27, 2025
| @@ -510,10 +510,10 @@ def test_round_up | |||
|
|
|||
| BigDecimal.mode(BigDecimal::ROUND_MODE, BigDecimal::ROUND_HALF_DOWN) | |||
| assert_operator(n4, :>, n4 / 9 * 9) | |||
| assert_operator(n5, :>, n5 / 9 * 9) | |||
| assert_operator(n5, :<, n5 / 9 * 9) | |||
There was a problem hiding this comment.
Half down rules are:
123.5 → 123 (half down)
123.50001 → 124 (round up because it's larger than half)
0.55550 → 0.555
0.55551 → 0.556
0.55555555.. → 0.556 (Because it's larger than half)
In master branch, rounding is inconsistent between a / b and a.div(b, prec).
BigDecimal.mode(BigDecimal::ROUND_MODE, BigDecimal::ROUND_HALF_DOWN)
BigDecimal(5)/9
#=> 0.555555555555555555555555555555555555e0 (ends with 5)
BigDecimal(5).div(9, 36)
#=> 0.555555555555555555555555555555555556e0 (ends with 6)
Merged
Closed
This was referenced May 29, 2025
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
Fixes #220, #222 and #272
BigDecimal_div and BigDecimal_div2 both have rounding problem (#272 and #328)
Integrating two division logic will make it easy to fix them.
Precision calculation for
a / banda.div(b, 0)is improved. There is no exponential precision growth anymore.