JDK-8294539: Augment discussion of equivalence relations on floating-point values #10498

jddarcy · 2022-09-29T22:14:24Z

While the floating-point == operation is not an equivalence relation, there are useful equivalence relations that can be defined over floating-point values. Text is added to java.lang.Double to discuss and name those relations.

Progress

Change must be properly reviewed (1 review required, with at least 1 Reviewer)
Change must not contain extraneous whitespace
Commit message must refer to an issue

Issue

JDK-8294539: Augment discussion of equivalence relations on floating-point values

Reviewers

Paul Sandoz (@PaulSandoz - Reviewer)
Stuart Marks (@stuart-marks - Reviewer) ⚠️ Review applies to 3a7dfbeb

Reviewing

Using git

Checkout this PR locally:
$ git fetch https://git.openjdk.org/jdk pull/10498/head:pull/10498
$ git checkout pull/10498

Update a local copy of the PR:
$ git checkout pull/10498
$ git pull https://git.openjdk.org/jdk pull/10498/head

Using Skara CLI tools

Checkout this PR locally:
$ git pr checkout 10498

View PR using the GUI difftool:
$ git pr show -t 10498

Using diff file

Download this PR as a diff file:
https://git.openjdk.org/jdk/pull/10498.diff

…oint values

bridgekeeper · 2022-09-29T22:16:24Z

👋 Welcome back darcy! A progress list of the required criteria for merging this PR into master will be added to the body of your pull request. There are additional pull request commands available for use with this pull request.

openjdk · 2022-09-29T22:18:58Z

@jddarcy The following label will be automatically applied to this pull request:

core-libs

When this pull request is ready to be reviewed, an "RFR" email will be sent to the corresponding mailing list. If you would like to change these labels, use the /label pull request command.

rose00 · 2022-09-29T22:32:52Z

src/java.base/share/classes/java/lang/Double.java

+ * <br>{@code Double.doubleToRawLongBits(a) == Double.doubleToRawLongBits(b)}<br>
+ * Under this relation, {@code +0.0} and {@code -0.0} are distinguished
+ * from each other and every bit pattern encoding a NaN is distinguished
+ * form other bit patterns encoding a NaN.


s/form/from

rose00 · 2022-09-29T22:39:21Z

src/java.base/share/classes/java/lang/Double.java

+ * Note that representation equivalence is often an appropriate notion
+ * of equivalence to test the behavior of {@linkplain StrictMath math
+ * libraries}.
+ * </dd>


This list deserves to include the not-quite-equivalence relation ==. Suggest:

numerical equivalence

{@code +0.0} and {@code -0.0} are equivalent to each other.

infinite numbers of the same sign are equivalent to each other

The two floating point numbers represent the same mathematical finite value or signed infinity.

Expressions implementing this ER include x == y, x > y | x < y, etc.

Numerical equivalence does not apply to NaN, as NaN violates the reflexive property required of all equivalences, but on the domain of all non-NaNs, numerical equivalence is in fact an equivalence relation.

mlbridge · 2022-09-30T01:08:11Z

Webrevs

PaulSandoz · 2022-09-30T20:08:36Z

src/java.base/share/classes/java/lang/Double.java

+ * <dd>The bits of the two floating-point values are the same. This
+ * equivalence relation for {@code double} values {@code a} and {@code b} is
+ * implemented by the expression
+ * <br>{@code Double.doubleTo}<code><b>Raw</b></code>{@code LongBits(a) == Double.doubleTo}<code><b>Raw</b></code>{@code LongBits(b)}<br>


Place in a code snippet since it's hard to read in source otherwise? Snippets allow for highlighting e.g.

/** * {@snippet : * Double.doubleToRawLongBits(a) == Double.doubleToRawLongBits(b) // @highlight substring="Raw" * } */

Before sending out the PR, I used a snippet, but it formatted the code using too much vertical space for the definition list. I didn't look to see if there was a "compact" styling option that could be used.

Ok, i am not aware of a compact style.

Filed JDK-8294814

openjdk · 2022-09-30T20:24:11Z

@jddarcy This change now passes all automated pre-integration checks.

ℹ️ This project also has non-automated pre-integration requirements. Please see the file CONTRIBUTING.md for details.

After integration, the commit message for the final commit will be:

8294539: Augment discussion of equivalence relations on floating-point values

Reviewed-by: psandoz, smarks

You can use pull request commands such as /summary, /contributor and /issue to adjust it as needed.

At the time when this comment was updated there had been 15 new commits pushed to the master branch:

b2e86a6: 8294255: Add link to DEFAULT_WAIT_TIME in javadoc for SunToolKit.realsSync
b22a38d: 8292309: Fix java/awt/PrintJob/ConstrainedPrintingTest/ConstrainedPrintingTest.java test
121d4a5: 8293579: tools/jpackage/share/jdk/jpackage/tests/UnicodeArgsTest.java fails on Japanese Windows platform
1166a8a: 8292214: Memory leak in getAllConfigs of awt_GraphicsEnv.c:386
3644e26: 8294673: JFR: Add SecurityProviderService#threshold to TestActiveSettingEvent.java
085949a: 8294712: G1: Use index-base iteration for G1FlushHumongousCandidateRemSets
b850f05: 8294758: JFR: Docs build fails after changes to RecordedObject and Timespan
2dbedf0: 8294406: Test runtime/handshake/HandshakeDirectTest.java failed: JVMTI_ERROR_WRONG_PHASE
4d6668e: 8294242: JFR: jfr print doesn't handle infinite duration well
5a9cd33: 8294509: The sign extension bug applies to 'public static int[] convertSeedBytesToInts(byte[] seed, int n, int z)' in RandomSupport
... and 5 more: https://git.openjdk.org/jdk/compare/ae79af2ad67b51a7608f4c9060421dd175cabf3f...master

As there are no conflicts, your changes will automatically be rebased on top of these commits when integrating. If you prefer to avoid this automatic rebasing, please check the documentation for the /integrate command for further details.

➡️ To integrate this PR with the above commit message to the master branch, type /integrate in a new comment.

stuart-marks · 2022-10-03T19:57:35Z

src/java.base/share/classes/java/lang/Double.java

+ * <p>While {@code ==} is not an equivalence relation, several useful
+ * equivalence relations can be defined over floating-point
+ * values. Comparing numerical equality to various kinds of
+ * equivalence:


The overall discussion is good, and the inclusion of numerical equality is good, but now this introductory paragraph is invalid. It starts off:

While == is not an equivalence relation,

OK, this continues the discussion, picking up from the assertion several paragraphs above that == is not an equivalence relation. Then,

several useful equivalence relations can be defined over floating-point values. Comparing numerical equality to various kinds of equivalence:

It sounds like what follows will be a list of equivalence relations that will be contrasted with numerical equality, yet the very first example is numerical equality and is not an equivalence relation.

I think this just needs to be reworded to be more inclusive. Maybe something like, "the following is a comparison of various relations among floating-point values."

stuart-marks · 2022-10-03T20:04:09Z

src/java.base/share/classes/java/lang/Double.java

+ * <ul>
+ * <li> {@code +0.0} and {@code -0.0} are distinguished from each other.
+ * <li> every bit pattern encoding a NaN is considered equivalent to each other
+ * <li> an infinite value is equivalent to an infinite value of the same sign


Seems like this line on infinities could be reworded. I wouldn't quibble over this except that I had to read it several times to figure out what it meant. The statement on NaN is universally quantified, whereas the statement on infinite values starts off sounding like an existential quantifier. Possibly: "all infinite values of the same sign are considered equivalent to each other."

Update as suggested and added some cross-links in from BigDecimal; thanks.

PS To further spell things out, I added an additional trailing paragraph:

"For two binary floating-point values a and b, if neither of a and b is zero or NaN, then the three relations numerical equality, bit-wise equivalence, and representation equivalence of a and b have the same true/false value. In other words, for binary floating-point values, the three relations only differ if at least one argument is zero or NaN."

The additional paragraph sounds fine.

jddarcy · 2022-10-05T00:17:27Z

/integrate

openjdk · 2022-10-05T00:19:08Z

Going to push as commit 1dafbe3.
Since your change was applied there have been 15 commits pushed to the master branch:

b2e86a6: 8294255: Add link to DEFAULT_WAIT_TIME in javadoc for SunToolKit.realsSync
b22a38d: 8292309: Fix java/awt/PrintJob/ConstrainedPrintingTest/ConstrainedPrintingTest.java test
121d4a5: 8293579: tools/jpackage/share/jdk/jpackage/tests/UnicodeArgsTest.java fails on Japanese Windows platform
1166a8a: 8292214: Memory leak in getAllConfigs of awt_GraphicsEnv.c:386
3644e26: 8294673: JFR: Add SecurityProviderService#threshold to TestActiveSettingEvent.java
085949a: 8294712: G1: Use index-base iteration for G1FlushHumongousCandidateRemSets
b850f05: 8294758: JFR: Docs build fails after changes to RecordedObject and Timespan
2dbedf0: 8294406: Test runtime/handshake/HandshakeDirectTest.java failed: JVMTI_ERROR_WRONG_PHASE
4d6668e: 8294242: JFR: jfr print doesn't handle infinite duration well
5a9cd33: 8294509: The sign extension bug applies to 'public static int[] convertSeedBytesToInts(byte[] seed, int n, int z)' in RandomSupport
... and 5 more: https://git.openjdk.org/jdk/compare/ae79af2ad67b51a7608f4c9060421dd175cabf3f...master

Your commit was automatically rebased without conflicts.

openjdk · 2022-10-05T00:19:21Z

@jddarcy Pushed as commit 1dafbe3.

💡 You may see a message that your pull request was closed with unmerged commits. This can be safely ignored.

src/java.base/share/classes/java/lang/Double.java

JDK-8294539: Augment discussion of equivlance relations on floating-p…

7d18585

…oint values

jddarcy marked this pull request as draft September 29, 2022 22:14

openjdk bot added the core-libs core-libs-dev@openjdk.org label Sep 29, 2022

rose00 reviewed Sep 29, 2022

View reviewed changes

Fix typo.

95d11d8

jddarcy marked this pull request as ready for review September 30, 2022 01:01

openjdk bot added the rfr Pull request is ready for review label Sep 30, 2022

jddarcy changed the title ~~JDK-8294539: Augment discussion of equivlance relations on floating-point values~~ JDK-8294539: Augment discussion of equivalence relations on floating-point values Sep 30, 2022

Add discussion of numerical equality.

5854364

PaulSandoz approved these changes Sep 30, 2022

View reviewed changes

openjdk bot added the ready Pull request is ready to be integrated label Sep 30, 2022

stuart-marks reviewed Oct 3, 2022

View reviewed changes

jddarcy added 2 commits October 3, 2022 21:58

Merge branch 'master' into JDK-8294539

5f41936

Respond to review feedback.

3a7dfbe

openjdk bot removed ready Pull request is ready to be integrated rfr Pull request is ready for review labels Oct 4, 2022

stuart-marks approved these changes Oct 4, 2022

View reviewed changes

jddarcy added 2 commits October 4, 2022 12:16

Add additional explanatory paragraph.

aaac731

Appease jcheck; reflow paragraphs.

8aca6b3

openjdk bot added ready Pull request is ready to be integrated rfr Pull request is ready for review labels Oct 4, 2022

PaulSandoz approved these changes Oct 4, 2022

View reviewed changes

openjdk bot added the integrated Pull request has been integrated label Oct 5, 2022

openjdk bot closed this Oct 5, 2022

openjdk bot removed ready Pull request is ready to be integrated rfr Pull request is ready for review labels Oct 5, 2022

biboudis reviewed Oct 18, 2022

View reviewed changes

src/java.base/share/classes/java/lang/Double.java Show resolved Hide resolved

jddarcy deleted the JDK-8294539 branch June 10, 2023 20:13

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

JDK-8294539: Augment discussion of equivalence relations on floating-point values #10498

JDK-8294539: Augment discussion of equivalence relations on floating-point values #10498

jddarcy commented Sep 29, 2022 •

edited by openjdk bot

Loading

bridgekeeper bot commented Sep 29, 2022

openjdk bot commented Sep 29, 2022

rose00 Sep 29, 2022

rose00 Sep 29, 2022

mlbridge bot commented Sep 30, 2022 •

edited

Loading

PaulSandoz Sep 30, 2022 •

edited

Loading

jddarcy Oct 4, 2022

PaulSandoz Oct 4, 2022

jonathan-gibbons Oct 4, 2022

openjdk bot commented Sep 30, 2022 •

edited

Loading

stuart-marks Oct 3, 2022

stuart-marks Oct 3, 2022

jddarcy Oct 4, 2022

jddarcy Oct 4, 2022

stuart-marks Oct 4, 2022

jddarcy commented Oct 5, 2022

openjdk bot commented Oct 5, 2022

openjdk bot commented Oct 5, 2022

JDK-8294539: Augment discussion of equivalence relations on floating-point values #10498

JDK-8294539: Augment discussion of equivalence relations on floating-point values #10498

Conversation

jddarcy commented Sep 29, 2022 • edited by openjdk bot Loading

Progress

Issue

Reviewers

Reviewing

bridgekeeper bot commented Sep 29, 2022

openjdk bot commented Sep 29, 2022

Choose a reason for hiding this comment

Choose a reason for hiding this comment

mlbridge bot commented Sep 30, 2022 • edited Loading

Webrevs

PaulSandoz Sep 30, 2022 • edited Loading

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

openjdk bot commented Sep 30, 2022 • edited Loading

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

jddarcy commented Oct 5, 2022

openjdk bot commented Oct 5, 2022

openjdk bot commented Oct 5, 2022

jddarcy commented Sep 29, 2022 •

edited by openjdk bot

Loading

mlbridge bot commented Sep 30, 2022 •

edited

Loading

PaulSandoz Sep 30, 2022 •

edited

Loading

openjdk bot commented Sep 30, 2022 •

edited

Loading