Make positive -> finite with new facts like is_extended_positive #16666

oscarbenjamin · 2019-04-16T23:40:05Z

References to other Issues or PRs

Based on #16603
Alternative to #16655

Brief description of what is fixed or changed

This is an attempt to build on #16603 which makes is_real imply is_finite. The idea is to also make positive, nonpositive etc also imply finite so that positive implies real. In #16655 I tried to do this without introducing new facts and hit a dead end. This attempt introduces new facts such as is_extended_positive which allow for positive testing with oo and -oo included. The assumption rules used here are:

_assume_rules = FactRules([

    'integer        ->  rational',
    'rational       ->  real',
    'rational       ->  algebraic',
    'algebraic      ->  complex & finite',
    'transcendental ==  complex & !algebraic & finite',
    'real           ->  hermitian',
    'imaginary      ->  complex & finite',
    'imaginary      ->  antihermitian',
    'extended_real  ->  commutative',
    'complex        ->  commutative',

    'odd            ==  integer & !even',
    'even           ==  integer & !odd',

    'real           ->  complex',
    'extended_real  ->  real | infinite',
    'real           ==  extended_real & finite',

    'extended_real        ==  extended_negative | zero | extended_positive',
    'extended_negative    ==  extended_nonpositive & extended_nonzero',
    'extended_positive    ==  extended_nonnegative & extended_nonzero',

    'extended_nonpositive ==  extended_real & !extended_positive',
    'extended_nonnegative ==  extended_real & !extended_negative',

    'real           ==  negative | zero | positive',
    'negative       ==  nonpositive & nonzero',
    'positive       ==  nonnegative & nonzero',

    'nonpositive    ==  real & !positive',
    'nonnegative    ==  real & !negative',

    'positive       ==  extended_positive & finite',
    'negative       ==  extended_negative & finite',
    'nonpositive    ==  extended_nonpositive & finite',
    'nonnegative    ==  extended_nonnegative & finite',
    'nonzero        ==  extended_nonzero & finite',

    'zero           ->  even & finite',
    'zero           ==  extended_nonnegative & extended_nonpositive',
    'zero           ==  nonnegative & nonpositive',

    'prime          ->  integer & positive',
    'composite      ->  integer & positive & !prime',
    '!composite     ->  !positive | !even | prime',

    'irrational     ==  real & !rational',

    'imaginary      ->  !extended_real',

    'infinite       ->  !finite',
    'noninteger     ==  extended_real & !integer',
    'extended_nonzero == extended_real & !zero',
])

Note that many of these rules are redundant. I initially had a more economical set of implications but deduce_all_facts was unable to make all possible inferences so I added additional rules to help it.

Other comments

Release Notes

core
- The assumptions system is changed so that only finite numbers can be considered real, positive, negative, nonnegative, nonpositive or nonzero (since nonzero implies real). This means that any symbol declared with e.g. real=True is now automatically considered finite. It also means that infinities can not be considered positive or negative since they are not real (e.g. oo.is_positive is now False).
- New assumptions extended_real, extended_positive etc are added that allow for positive and negative infinity. The equivalent of Symbol('x', real=True) in version 1.4 is now Symbol('x', extended_real=True). The equivalent of Symbol('x', negative=False) is now Symbol('x', extended_negative=False) although it is usually better to use Symbol('x', nonnegative=True) (which implies both real=True and finite=True as well). Code that previously checked if x.is_positive should now be written as if x.is_extended_positive if it is intended that infinities should be allowed.
- A number of corner cases in assumptions (e.g. for Pow.is_zero, Mul.is_positive etc) have been fixed to properly take account of the possibility of infinities.

sympy-bot · 2019-04-16T23:40:16Z

✅

Hi, I am the SymPy bot (v147). I'm here to help you write a release notes entry. Please read the guide on how to write release notes.

Your release notes are in good order.

Here is what the release notes will look like:

core
- The assumptions system is changed so that only finite numbers can be considered real, positive, negative, nonnegative, nonpositive or nonzero (since nonzero implies real). This means that any symbol declared with e.g. real=True is now automatically considered finite. It also means that infinities can not be considered positive or negative since they are not real (e.g. oo.is_positive is now False). (#16666 by @oscarbenjamin and @smichr)
- New assumptions extended_real, extended_positive etc are added that allow for positive and negative infinity. The equivalent of Symbol('x', real=True) in version 1.4 is now Symbol('x', extended_real=True). The equivalent of Symbol('x', negative=False) is now Symbol('x', extended_negative=False) although it is usually better to use Symbol('x', nonnegative=True) (which implies both real=True and finite=True as well). Code that previously checked if x.is_positive should now be written as if x.is_extended_positive if it is intended that infinities should be allowed. (#16666 by @oscarbenjamin and @smichr)
- A number of corner cases in assumptions (e.g. for Pow.is_zero, Mul.is_positive etc) have been fixed to properly take account of the possibility of infinities. (#16666 by @oscarbenjamin and @smichr)

This will be added to https://github.com/sympy/sympy/wiki/Release-Notes-for-1.5.

Note: This comment will be updated with the latest check if you edit the pull request. You need to reload the page to see it.

Click here to see the pull request description that was parsed.

<!-- Your title above should be a short description of what
was changed. Do not include the issue number in the title. -->

#### References to other Issues or PRs
<!-- If this pull request fixes an issue, write "Fixes #NNNN" in that exact
format, e.g. "Fixes #1234". See
https://github.com/blog/1506-closing-issues-via-pull-requests . Please also
write a comment on that issue linking back to this pull request once it is
open. -->

Based on #16603 
Alternative to #16655 

#### Brief description of what is fixed or changed

This is an attempt to build on #16603 which makes is_real imply is_finite. The idea is to also make positive, nonpositive etc also imply finite so that positive implies real. In #16655 I tried to do this without introducing new facts and hit a dead end. This attempt introduces new facts such as is_extended_positive which allow for positive testing with oo and -oo included. The assumption rules used here are:
```python
_assume_rules = FactRules([

    'integer        ->  rational',
    'rational       ->  real',
    'rational       ->  algebraic',
    'algebraic      ->  complex & finite',
    'transcendental ==  complex & !algebraic & finite',
    'real           ->  hermitian',
    'imaginary      ->  complex & finite',
    'imaginary      ->  antihermitian',
    'extended_real  ->  commutative',
    'complex        ->  commutative',

    'odd            ==  integer & !even',
    'even           ==  integer & !odd',

    'real           ->  complex',
    'extended_real  ->  real | infinite',
    'real           ==  extended_real & finite',

    'extended_real        ==  extended_negative | zero | extended_positive',
    'extended_negative    ==  extended_nonpositive & extended_nonzero',
    'extended_positive    ==  extended_nonnegative & extended_nonzero',

    'extended_nonpositive ==  extended_real & !extended_positive',
    'extended_nonnegative ==  extended_real & !extended_negative',

    'real           ==  negative | zero | positive',
    'negative       ==  nonpositive & nonzero',
    'positive       ==  nonnegative & nonzero',

    'nonpositive    ==  real & !positive',
    'nonnegative    ==  real & !negative',

    'positive       ==  extended_positive & finite',
    'negative       ==  extended_negative & finite',
    'nonpositive    ==  extended_nonpositive & finite',
    'nonnegative    ==  extended_nonnegative & finite',
    'nonzero        ==  extended_nonzero & finite',

    'zero           ->  even & finite',
    'zero           ==  extended_nonnegative & extended_nonpositive',
    'zero           ==  nonnegative & nonpositive',

    'prime          ->  integer & positive',
    'composite      ->  integer & positive & !prime',
    '!composite     ->  !positive | !even | prime',

    'irrational     ==  real & !rational',

    'imaginary      ->  !extended_real',

    'infinite       ->  !finite',
    'noninteger     ==  extended_real & !integer',
    'extended_nonzero == extended_real & !zero',
])
```
Note that many of these rules are redundant. I initially had a more economical set of implications but deduce_all_facts was unable to make all possible inferences so I added additional rules to help it.

#### Other comments


#### Release Notes

<!-- Write the release notes for this release below. See
https://github.com/sympy/sympy/wiki/Writing-Release-Notes for more information
on how to write release notes. The bot will check your release notes
automatically to see if they are formatted correctly. -->

<!-- BEGIN RELEASE NOTES -->
* core
    * The assumptions system is changed so that only finite numbers can be considered real, positive, negative, nonnegative, nonpositive or nonzero (since nonzero implies real). This means that any symbol declared with e.g. real=True is now automatically considered finite. It also means that infinities can not be considered positive or negative since they are not real (e.g. oo.is_positive is now False).
    * New assumptions extended_real, extended_positive etc are added that allow for positive and negative infinity. The equivalent of `Symbol('x', real=True)` in version 1.4 is now `Symbol('x', extended_real=True)`. The equivalent of `Symbol('x', negative=False)` is now `Symbol('x', extended_negative=False)` although it is usually better to use `Symbol('x', nonnegative=True)` (which implies both real=True and finite=True as well). Code that previously checked `if x.is_positive` should now be written as `if x.is_extended_positive` if it is intended that infinities should be allowed.
    * A number of corner cases in assumptions (e.g. for Pow.is_zero, Mul.is_positive etc) have been fixed to properly take account of the possibility of infinities.
<!-- END RELEASE NOTES -->

Update

The release notes on the wiki have been updated.

sympy/calculus/util.py

sympy/core/add.py

oscarbenjamin · 2019-04-17T10:19:27Z

I've just pushed a commit to fix the failing test. The issue was due to Expr not having _eval_is_positive.

I think that maybe the code for Expr._eval_is_extended_positive/negative should be factored out. If I understand correctly the code currently in _eval_is_extended_positive doesn't handle infinities and should really be in a _eval_is_positive method. Expr needs to define _eval_is_extended_positive so that it can be called from Add._eval_is_extended_positive in

     return super(Add, self)._eval_is_extended_positive()

I'm not sure if that can just be changed back to calling _eval_is_positive because I don't know if super would always find Expr as the superclass and other classes might implement _eval_is_extended_positive differently taking account of infinities as well.

The commit that I just pushed includes (in Expr):

     def _eval_is_extended_positive(self):
        if self.is_extended_real is False:
            return False
        if Expr._eval_is_positive(self):
            return True

oscarbenjamin · 2019-04-17T10:26:42Z

This PR currently has 3 XFAILS.

There are two for limits:

$ pytest sympy/utilities/tests/test_wester.py --runxfail -k test_T6
============================================================= test session starts =============================================================
...
================================================================== FAILURES ===================================================================
___________________________________________________________________ test_T6 ___________________________________________________________________

    @XFAIL
    def test_T6():
>       assert limit(1/n * factorial(n)**(1/n), n, oo) == exp(-1)
E       assert 0 == exp(-1)
E        +  where 0 = limit(((1 / n) * (factorial(n) ** (1 / n))), n, oo)
E        +    where factorial(n) = factorial(n)
E        +  and   exp(-1) = exp(-1)

sympy/utilities/tests/test_wester.py:2093: AssertionError
                                                               DO *NOT* COMMIT!                                                               
================================================== 1 failed, 396 deselected in 1.03 seconds ===================================================

$ pytest sympy/series/tests/test_limits.py --runxfail -k test_factorial
============================================================= test session starts =============================================================
================================================================== FAILURES ===================================================================
_______________________________________________________________ test_factorial ________________________________________________________________

    @XFAIL
    def test_factorial():
        from sympy import factorial, E
        f = factorial(x)
        assert limit(f, x, oo) == oo
>       assert limit(x/f, x, oo) == 0
E       AssertionError: assert Limit(x/factorial(x), x, oo, dir='-') == 0
E        +  where Limit(x/factorial(x), x, oo, dir='-') = limit((x / factorial(x)), x, oo)

sympy/series/tests/test_limits.py:412: AssertionError
                                                               DO *NOT* COMMIT!                                                               
=================================================== 1 failed, 61 deselected in 0.70 seconds ===================================================

The third XFAIL is to do with the new assumptions calling the old assumptions. I think that's because I haven't updated the new assumptions yet but I'm not sure:

$ pytest sympy/assumptions/tests/test_sathandlers.py --runxfail -k test_CheckOldAssump
============================================================= test session starts =============================================================
================================================================== FAILURES ===================================================================
_____________________________________________________________ test_CheckOldAssump _____________________________________________________________

    @XFAIL
    def test_CheckOldAssump():
        # TODO: Make these tests more complete
    
        class Test1(Expr):
            def _eval_is_positive(self):
                return True
            def _eval_is_negative(self):
                return False
    
        class Test2(Expr):
            def _eval_is_finite(self):
                return True
            def _eval_is_positive(self):
                return True
            def _eval_is_negative(self):
                return False
    
        t1 = Test1()
        t2 = Test2()
    
        # We can't say if it's positive or negative in the old assumptions without
        # bounded. Remember, True means "no new knowledge", and
        # Q.positive(t2) means "t2 is positive."
>       assert CheckOldAssump(Q.positive(t1)) == True
E       assert Q.positive(Test1()) == True
E        +  where Q.positive(Test1()) = CheckOldAssump(Q.positive(Test1()))
E        +    where Q.positive(Test1()) = Q.positive(Test1())
E        +      where Q.positive = <sympy.assumptions.ask.AssumptionKeys object at 0x108696320>.positive

sympy/assumptions/tests/test_sathandlers.py:94: AssertionError
                                                               DO *NOT* COMMIT!                                                               
=================================================== 1 failed, 5 deselected in 0.18 seconds ====================================================

oscarbenjamin · 2019-04-17T10:29:20Z

sympy/concrete/products.py

-    pi/2
+
+    #>>> limit(W2e, n, oo)
+    #pi/2


I have commented out this doctest because it failed:

092 >>> from sympy import limit 093 >>> W2 = Product(2*i/(2*i-1)*2*i/(2*i+1), (i, 1, n)) 094 >>> W2 095 Product(4*i**2/((2*i - 1)*(2*i + 1)), (i, 1, n)) 096 >>> W2e = W2.doit() 097 >>> W2e 098 2**(-2*n)*4**n*factorial(n)**2/(RisingFactorial(1/2, n)*RisingFactorial(3/2, n)) 099 100 >>> limit(W2e, n, oo) Expected: pi/2 Got: Limit(2**(-2*n)*4**n*factorial(n)**2/(RisingFactorial(1/2, n)*RisingFactorial(3/2, n)), n, oo, dir='-') sympy/concrete/products.py:100: DocTestFailure

codecov · 2019-04-17T12:17:03Z

Codecov Report

Merging #16666 into master will decrease coverage by 0.008%.
The diff coverage is 91.925%.

@@              Coverage Diff              @@
##            master    #16666       +/-   ##
=============================================
- Coverage   73.906%   73.897%   -0.009%     
=============================================
  Files          619       619               
  Lines       160001    160101      +100     
  Branches     37554     37578       +24     
=============================================
+ Hits        118251    118311       +60     
- Misses       36269     36308       +39     
- Partials      5481      5482        +1

oscarbenjamin · 2019-04-17T17:34:12Z

The tests have passed (with 3 XFAILED and one doctest commented out) but have taken longer than usual to run. I can reproduce a slowdown with:

$ bin/test sympy/simplify/tests/test_hyperexpand.py -k test_formulae --slow

That takes 24 seconds with this PR and 14 seconds on master.

oscarbenjamin · 2019-04-18T11:43:23Z

There is an intermittent test failure that I can reproduce on Python 3.7 with:

$ PYTHONHASHSEED=2937395614 bin/test sympy/core/tests/test_assumptions.py -k test_Mul_is_prime_composite --seed=73309503
============================================================= test process starts =============================================================
executable:         /Users/enojb/current/sympy/venv/bin/python  (3.7.1-final-0) [CPython]
architecture:       64-bit
cache:              yes
ground types:       gmpy 1.17
numpy:              1.16.1
random seed:        73309503
hash randomization: on (PYTHONHASHSEED=2937395614)

sympy/core/tests/test_assumptions.py[1] F                                                                                                [FAIL]

_______________________________________________________________________________________________________________________________________________
______________________________________ sympy/core/tests/test_assumptions.py:test_Mul_is_prime_composite _______________________________________
Traceback (most recent call last):
  File "/Users/enojb/current/sympy/sympy/sympy/core/tests/test_assumptions.py", line 834, in test_Mul_is_prime_composite
    assert ( (x+1)*(y+1) ).is_prime is False
AssertionError

============================================= tests finished: 0 passed, 1 failed, in 0.28 seconds =============================================
DO *NOT* COMMIT!

oscarbenjamin · 2019-04-18T22:48:42Z

I think I've fixed the performance problem. I've also fixed the 2 XFAILed tests and one commented test that were all to do with limits.

That leaves one XFAILed test to do with the new assumptions:

$ pytest sympy/assumptions/tests/test_sathandlers.py --runxfail -k test_CheckOldAssump

I think that's just because I should update the new assumptions now.

oscarbenjamin · 2019-04-18T23:06:10Z

It looks like the new assumptions already define positive as finite:

sympy/sympy/assumptions/sathandlers.py

Line 186 in 3335542

ret = fuzzy_and([e.is_finite, e.is_positive])

So I think the only problem there is that the failing test is no longer valid under the changes in this PR:

$ pytest sympy/assumptions/tests/test_sathandlers.py --runxfail -k test_CheckOldAssump
============================================================= test session starts =============================================================
================================================================== FAILURES ===================================================================
_____________________________________________________________ test_CheckOldAssump _____________________________________________________________

    @XFAIL
    def test_CheckOldAssump():
        # TODO: Make these tests more complete
    
        class Test1(Expr):
            def _eval_is_positive(self):
                return True
            def _eval_is_negative(self):
                return False
    
        class Test2(Expr):
            def _eval_is_finite(self):
                return True
            def _eval_is_positive(self):
                return True
            def _eval_is_negative(self):
                return False
    
        t1 = Test1()
        t2 = Test2()
    
        # We can't say if it's positive or negative in the old assumptions without
        # bounded. Remember, True means "no new knowledge", and
        # Q.positive(t2) means "t2 is positive."
>       assert CheckOldAssump(Q.positive(t1)) == True
E       assert Q.positive(Test1()) == True
E        +  where Q.positive(Test1()) = CheckOldAssump(Q.positive(Test1()))
E        +    where Q.positive(Test1()) = Q.positive(Test1())
E        +      where Q.positive = <sympy.assumptions.ask.AssumptionKeys object at 0x108696320>.positive

sympy/assumptions/tests/test_sathandlers.py:94: AssertionError

oscarbenjamin · 2019-04-18T23:13:44Z

I've fixed that test with positive -> extended_positive. That means there are no failing tests with this PR now.

Todo:

Go back through the diff
Audit uses of is_positive, is_real throughout codebase
Investigate performance (tests are running maybe 5% slower right now)

Reviews/suggestions welcome!

oscarbenjamin · 2019-04-22T17:02:01Z

I've rebased on #16603 which was rebased onto master because of #16592

sympy/core/numbers.py

oscarbenjamin · 2019-05-27T19:23:53Z

Thanks you very much for taking the time to review this @ShubhamKJha. I think I've responded to all of your comments now but please let me know if I've missed something.

I think I've done everything needed in this PR now apart from squashing the commits. Is there any useful way to format them?

I don't think that the changes can be separated into independently revertible parts but I can split it into maybe one commit for the changes to the core, another for functions, and several for different parts of the tests.

smichr · 2019-05-27T20:37:21Z

I'm not sure that keeping all the commit separate will help in tracking down any future bugs, but one will be able to use bisection to still tell if it is before or after this batch that something broke so I don't think there is a need to squash them.

This needs a release note telling what has changed, i.e. what worked before vs what works now. Is this backwards incompatible in assumptions in any way?

oscarbenjamin · 2019-05-27T21:21:19Z

I've added a release note.

This is backwards incompatible. I find it hard to gauge how many problems users will have with it. I had to change a lot of things from e.g. is_real to is_extended_real but I suspect that this is more a problem in the core than it would be for users.

Perhaps a more likely problem for users and one that came up a few times in this PR is when assumptions are declared negatively e.g. Symbol('x', negative=False). Under this PR that includes -oo as a possibility so on master we have:

In [1]: x = Symbol('x', negative=False)                                                                                           

In [2]: x < 0                                                                                                                     
Out[2]: False

Where as with this PR:

In [2]: x < 0                                                                                                                     
Out[2]: x < 0

I suspect that most users (and contributors?) who write negative=False don't realise what it actually means and should really use nonnegative=True or include real=True. I've mentioned this in the release notes.

smichr · 2019-05-27T23:17:23Z

@jksuom , @asmeurer - any objections to this being committed?

asmeurer · 2019-05-28T00:12:41Z

I have no objections but let's wait for a sign-off from @ShubhamKJha.

asmeurer · 2019-05-28T00:14:08Z

What is the status of the new assumptions with this?

oscarbenjamin · 2019-05-28T00:42:53Z

I don't understand what the situation with the new assumptions is. There is Q.real and also Q.extended_real but Q.real does not imply finite and Q.real implies Q.extended_real:

In [70]: print(ask(Q.finite(x), Q.real(x)))                                                                                       
None

In [2]: ask(Q.extended_real(x), Q.real(x))                                                                                        
Out[2]: True

Also this code suggests that the new assumptions are different somehow in this respect but I don't see how in terms of output:

sympy/sympy/assumptions/sathandlers.py

Line 186 in 3335542

ret = fuzzy_and([e.is_finite, e.is_positive])

oscarbenjamin · 2019-05-28T00:56:12Z

Whoops! Q.real implies Q.extended_real is not surprising. That's what this PR does with the old assumptions.

In the new assumptions Q.extended_real does not imply Q.real:

In [6]: print(ask(Q.real(x), Q.extended_real(x)))                                                                                 
None

However Q.real does not imply finite:

In [7]: print(ask(Q.finite(x), Q.real(x)))                                                                                        
None

so I don't know what that means.

asmeurer · 2019-05-28T04:25:39Z

We can clean up the new assumptions in a new PR. The two should ultimately be consistent with one another. I just wanted to see if that was already the case or not.

ShubhamKJha · 2019-05-28T05:17:07Z

In the new assumptions Q.extended_real does not imply Q.real:

That is correct. extended_real can also be infinite.

ShubhamKJha · 2019-05-28T05:19:24Z

Thanks @oscarbenjamin this PR looks good.

There are certainly two issues that popped up:

Inconsistent results with nonpositive = False etc.
Inconsistency in new assumptions

These can be taken care of in another PR.

jksuom · 2019-05-28T08:19:43Z

I think it is time to merge this.

smichr · 2019-05-28T15:42:18Z

more likely problem for users and one that came up a few times in this PR is when assumptions are declared negatively

I think you have the right behavior here negative=False doesn't imply anything about being real since I.is_negative is also False.

oscarbenjamin · 2019-05-28T15:59:22Z

Thanks everyone for reviewing and for merging.

There are a number of followup tasks that I'll list here:

Update the docs
Make the new assumptions consistent with the old
Make complex imply finite

For 3. I don't have time for a while now but I can get to it at some point. If someone else wants to do it sooner then go ahead! I didn't update the docs yet because I viewed this as a step towards making complex imply finite. The docs will need updating before the next release in any case though.

Also it looks like the core _eval_is methods for Pow, Add and Mul have room for improvement and there are a number of missing methods. #16433 will be somewhat invalidated by this merge but you can see there that just focussing closely on Pow.is_zero turned into a substantial patch. So that makes another task:

Closely review _eval_is methods for Pow, Add and Mul and add missing methods.

oscarbenjamin · 2019-05-28T16:08:41Z

more likely problem for users and one that came up a few times in this PR is when assumptions are declared negatively

I think you have the right behavior here negative=False doesn't imply anything about being real since I.is_negative is also False.

It's the right behaviour for the wrong reason because after this PR we still have:

In [1]: x = Symbol('x', extended_negative=False)                                                                                  

In [2]: x < 0                                                                                                                     
Out[2]: False

This is due to this code in Expr.__lt__ (that I didn't substantively change):

sympy/sympy/core/expr.py

Lines 365 to 369 in b9604c5

    
           if self.is_extended_real or other.is_extended_real: 
        
               dif = self - other 
        
               if dif.is_extended_negative is not None and \ 
        
                       dif.is_extended_negative is not dif.is_extended_nonnegative: 
        
                   return sympify(dif.is_extended_negative)

I have to say I don't understand exactly why that code is written the way it is but it doesn't handle the negative assumption case correctly.

sylee957 added assumptions core labels Apr 17, 2019

jksuom reviewed Apr 17, 2019

View reviewed changes

sympy/calculus/util.py Outdated Show resolved Hide resolved

jksuom reviewed Apr 17, 2019

View reviewed changes

sympy/core/add.py Outdated Show resolved Hide resolved

oscarbenjamin commented Apr 17, 2019

View reviewed changes

oscarbenjamin force-pushed the ext_pos_finite branch from cd32f56 to 44a93f6 Compare April 22, 2019 17:00

oscarbenjamin added 14 commits April 27, 2019 13:51

Add extended_real and Make real -> finite

4c5e54e

Replace is_real -> is_extended_real in numbers.py

f267682

Replace is_real -> is_extended_real in add.py

4ff3610

Replace is_real -> is_extended_real in expr.py

11592d1

Replace is_real -> is_extended_real in basic.py

fe1c7c9

Replace is_real -> is_extended_real in mul.py

d806cc3

Replace is_real -> is_extended_real in power.py

c980ad0

Replace is_real -> is_extended_real in complexes.py

8736fe4

Replace is_real -> is_extended_real in trigonometric.py

f407592

Fix tests after making real imply finite

742e3ea

Add extended_real functions/elementary/miscellaneous.py

270d6ed

Use extended real for AccumBounds in calculus/util.py

b2494aa

WIP - ...

431c8c0

WIP - ...

43a8857

oscarbenjamin added 2 commits May 27, 2019 14:21

Add Piecewise._eval_is_extended_positive etc

5e6edf8

Use is_extended_negative in erf

d66e52a

ShubhamKJha reviewed May 27, 2019

View reviewed changes

sympy/core/numbers.py Show resolved Hide resolved

oscarbenjamin added 3 commits May 27, 2019 17:30

Fix Mul._eval_is_positive for positive=False

791ddf8

Use extended_real=True in bessel test

9d0ef12

Update reduce_rational_inequalities docstring

dffb392

oscarbenjamin mentioned this pull request May 27, 2019

nonlinsolve misses corner case when unknowns are declared as finite #16909

Open

jksuom merged commit 50d1b2f into sympy:master May 28, 2019

oscarbenjamin mentioned this pull request May 28, 2019

Negated assumptions not handled correctly by __lt__ #16915

Closed

ShubhamKJha mentioned this pull request Jun 3, 2019

Improved code in Expr.__lt__ and similar functions in Expr #16956

Merged

oscarbenjamin mentioned this pull request Jun 7, 2019

[WIP] Make complex->finite in the codebase. #16978

Closed

oscarbenjamin mentioned this pull request Jun 27, 2019

[WIP] Sets: Fix relational between Cartesian Product and Finite Sets #16764

Closed

oscarbenjamin mentioned this pull request Sep 1, 2019

Mul.is_imaginary fails for infinite values #17556

Closed

oscarbenjamin mentioned this pull request Nov 22, 2019

multiple generators [exp(-x), exp(-x**2)] when trying to solve(exp(-x - x ** 2), x) #17949

Closed

oscarbenjamin deleted the ext_pos_finite branch January 9, 2020 01:53

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Make positive -> finite with new facts like is_extended_positive #16666

Make positive -> finite with new facts like is_extended_positive #16666

oscarbenjamin commented Apr 16, 2019 •

edited

Loading

sympy-bot commented Apr 16, 2019 •

edited

Loading

oscarbenjamin commented Apr 17, 2019

oscarbenjamin commented Apr 17, 2019

oscarbenjamin Apr 17, 2019

codecov bot commented Apr 17, 2019 •

edited

Loading

oscarbenjamin commented Apr 17, 2019

oscarbenjamin commented Apr 18, 2019

oscarbenjamin commented Apr 18, 2019

oscarbenjamin commented Apr 18, 2019

oscarbenjamin commented Apr 18, 2019

oscarbenjamin commented Apr 22, 2019

oscarbenjamin commented May 27, 2019

smichr commented May 27, 2019

oscarbenjamin commented May 27, 2019

smichr commented May 27, 2019

asmeurer commented May 28, 2019

asmeurer commented May 28, 2019

oscarbenjamin commented May 28, 2019

oscarbenjamin commented May 28, 2019

asmeurer commented May 28, 2019

ShubhamKJha commented May 28, 2019

ShubhamKJha commented May 28, 2019 •

edited

Loading

jksuom commented May 28, 2019

smichr commented May 28, 2019

oscarbenjamin commented May 28, 2019

oscarbenjamin commented May 28, 2019

Make positive -> finite with new facts like is_extended_positive #16666

Make positive -> finite with new facts like is_extended_positive #16666

Conversation

oscarbenjamin commented Apr 16, 2019 • edited Loading

References to other Issues or PRs

Brief description of what is fixed or changed

Other comments

Release Notes

sympy-bot commented Apr 16, 2019 • edited Loading

oscarbenjamin commented Apr 17, 2019

oscarbenjamin commented Apr 17, 2019

oscarbenjamin Apr 17, 2019

Choose a reason for hiding this comment

codecov bot commented Apr 17, 2019 • edited Loading

Codecov Report

oscarbenjamin commented Apr 17, 2019

oscarbenjamin commented Apr 18, 2019

oscarbenjamin commented Apr 18, 2019

oscarbenjamin commented Apr 18, 2019

oscarbenjamin commented Apr 18, 2019

oscarbenjamin commented Apr 22, 2019

oscarbenjamin commented May 27, 2019

smichr commented May 27, 2019

oscarbenjamin commented May 27, 2019

smichr commented May 27, 2019

asmeurer commented May 28, 2019

asmeurer commented May 28, 2019

oscarbenjamin commented May 28, 2019

oscarbenjamin commented May 28, 2019

asmeurer commented May 28, 2019

ShubhamKJha commented May 28, 2019

ShubhamKJha commented May 28, 2019 • edited Loading

jksuom commented May 28, 2019

smichr commented May 28, 2019

oscarbenjamin commented May 28, 2019

oscarbenjamin commented May 28, 2019

oscarbenjamin commented Apr 16, 2019 •

edited

Loading

sympy-bot commented Apr 16, 2019 •

edited

Loading

codecov bot commented Apr 17, 2019 •

edited

Loading

ShubhamKJha commented May 28, 2019 •

edited

Loading