Improve directed infinity #809

mmatera · 2023-03-06T22:31:18Z

This is based on one part of #766 related to how DirectedInfinity is handled. To simplify this, I remove part of the eval_Times code handling this kind of expression and replace it with upvalues rules. Also, some of the new pytests in #766 were included here.

mmatera · 2023-03-06T22:52:52Z

mathics/eval/plot.py

@@ -102,15 +100,10 @@ def quiet_f(*args):

            return quiet_f
    expr: Optional[Type[BaseElement]] = Expression(SymbolN, expr).evaluate(evaluation)
-    quiet_expr = Expression(


This came up when I run the doctests with the other changes: In WMA, when the points are evaluated, *Plot hides all the warnings, not just "Power::infy". This could go in a different PR, but the natural tests needs of the changes introduced here. (I also tested with functions like F[x_]:=(Open["nonexistingfile"];x^2) but it is just not so nice.

rocky · 2023-03-06T23:13:38Z

CHANGES.rst

@@ -3,6 +3,25 @@
 CHANGES
 =======

+
+API


Usually API refers the external API - things that a person using core would have to worry about.

I suspect this is just about internal organization and under the general category of we are moving evaluation functions into its own module, separate from the eval methods that function application finds.

rocky · 2023-03-06T23:20:54Z

mathics/builtin/arithmetic.py

    }

    formats = {
        "DirectedInfinity[1]": "HoldForm[Infinity]",
-        "DirectedInfinity[-1]": "HoldForm[-Infinity]",
+        "DirectedInfinity[-1]": "PrecedenceForm[-HoldForm[Infinity], 390]",


We should be getting numbers like 390 from the operator tables of Mathics scanner.

Times has Precedence 400, this needs to have a slightly lower value to ensure the parenthesis.

There is something then incomplete here. What comes in contact with what? Is there a value in Robert's table that is correct here?

And finally, if there is something that needs to be adjusted from the normal value of some other operator, a clearer way to do that would be to use that value and add or subtract from it. Something like Precedence["Times"] - xxx

Maybe what is missing is better documentation of what PrecedenceForm means.
https://reference.wolfram.com/language/ref/PrecedenceForm.html

In WMA, the right value for the precedence would be exactly the same as Times (400):

In[1]:= Times[b, PrecedenceForm[3 A ,400]] Out[1]= b (3 A) In[2]:= Times[b, PrecedenceForm[3 A ,401]] Out[2]= b 3 A

In Mathics, this does not works in the same way:

In[1]:= Times[b, PrecedenceForm[3 A ,400]] Out[1]= b (3 A) In[2]:= Times[b, PrecedenceForm[3 A ,401]] Out[2]= b (3 A) In[2]:= Times[b, PrecedenceForm[3 A ,399]] Out[2]= b (3 A)

but

In[1]:= Times[b, PrecedenceForm[A ,400]] Out[1]= b A In[2]:= Times[b, PrecedenceForm[3 A ,401]] Out[2]= b A In[2]:= Times[b, PrecedenceForm[3 A ,399]] Out[2]= b (A)

So maybe we need to adjust this before worrying about the exact value for the precedence in this rule.

This is good and important information. Give me some time to study this and compare with the existing tables and Robert's tables.

Notice also that Robert's tables have several values depending on the context. I guess that this is related to the use of PrecedenceForm in some internal the format rules in WMA.

Given that this is a much larger issue, and I think it is not going to be solved in the way proposed above, let's remove this change. I created #811 to capture the discussion above, and track this problem.

rocky · 2023-03-06T23:21:11Z

mathics/builtin/arithmetic.py

        "DirectedInfinity[]": "HoldForm[ComplexInfinity]",
-        "DirectedInfinity[DirectedInfinity[z_]]": "DirectedInfinity[z]",
-        "DirectedInfinity[z_?NumericQ]": "HoldForm[z Infinity]",
+        "DirectedInfinity[z_]": "PrecedenceForm[z HoldForm[Infinity], 390]",


390 - see above.

rocky · 2023-03-06T23:23:05Z

test/builtin/arithmetic/test_basic.py

@@ -164,7 +177,7 @@ def test_multiply(str_expr, str_expected, msg):
        ("I^(2/3)", "(-1) ^ (1 / 3)", None),
        # In WMA, the next test would return ``-(-I)^(2/3)``
        # which is less compact and elegant...
-        ("(-I)^(2/3)", "(-1) ^ (-1 / 3)", None),
+        #        ("(-I)^(2/3)", "(-1) ^ (-1 / 3)", None),


Do I have it correct that this was working before, and not it is not? If that it the case, what's the insight here. (And if not, what is the insight here.)

This works in #766 but not here. The issue comes from Power.eval_check.

rocky · 2023-03-12T15:45:20Z

mathics/builtin/arithmetic.py

@@ -208,6 +208,13 @@ class Abs(_MPMathFunction):
    summary_text = "absolute value of a number"
    sympy_name = "Abs"

+    def eval(self, x, evaluation):
+        "%(name)s[x_]"
+        result = eval_abs(x)


Should follow convention using capital letter. Also it would be nice to start removing %(name)s.

I will be putting in a PR to merge into this to correct such things.

CHANGES.rst ----------- this is a continuation of prior work, not an external API change arithmetic.py ------------- * eval_{abs,sign} -> eval_{Abs,Sign}; * remove MakeBoxes hack - will handle by a more pervasive go-over of Precedence. * generic style things changes. - Split long URL lines - reduce use of %(name)s - order class definitions test_basic.py ------------- Isolate failing DirectedInfinity tests

Improve directed infinity

rocky · 2023-03-12T22:03:48Z

After CI clears - LGTM

mmatera added 2 commits March 6, 2023 16:20

Improve rules for DirectedInfinity

9c6b54b

adding tests

8194c2c

mmatera commented Mar 6, 2023

View reviewed changes

rocky reviewed Mar 6, 2023

View reviewed changes

rocky reviewed Mar 12, 2023

View reviewed changes

rocky mentioned this pull request Mar 12, 2023

Go over Operator Precedence #811

Open

rocky and others added 2 commits March 12, 2023 15:15

Merge pull request #812 from Mathics3/improve_DirectedInfinity-rocky

8f1dc7e

Improve directed infinity

mmatera merged commit df2a9d1 into master Mar 12, 2023

mmatera deleted the improve_DirectedInfinity branch March 12, 2023 22:34

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Improve directed infinity #809

Improve directed infinity #809

mmatera commented Mar 6, 2023

mmatera Mar 6, 2023

rocky Mar 6, 2023 •

edited

Loading

rocky Mar 6, 2023

mmatera Mar 6, 2023

rocky Mar 6, 2023 •

edited

Loading

mmatera Mar 7, 2023 •

edited

Loading

rocky Mar 7, 2023

mmatera Mar 7, 2023

rocky Mar 12, 2023

rocky Mar 6, 2023

rocky Mar 6, 2023

mmatera Mar 6, 2023

rocky Mar 12, 2023

rocky commented Mar 12, 2023

@@ @@ -3,6 +3,25 @@ @@
               CHANGES
               =======
+              API

Improve directed infinity #809

Improve directed infinity #809

Conversation

mmatera commented Mar 6, 2023

Choose a reason for hiding this comment

rocky Mar 6, 2023 • edited Loading

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

rocky Mar 6, 2023 • edited Loading

Choose a reason for hiding this comment

mmatera Mar 7, 2023 • 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

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

rocky commented Mar 12, 2023

rocky Mar 6, 2023 •

edited

Loading

rocky Mar 6, 2023 •

edited

Loading

mmatera Mar 7, 2023 •

edited

Loading