improve FriCAS interface #21231

mantepse · 2016-08-12T11:05:04Z

The FriCAS interface is currently very rudimentary. In particular, converting the results of a computation into sage types is available only in very few special cases.

Depends on #21209

CC: @sagetrac-bpage @hemmecke @dkrenn @dimpase

Component: interfaces

Keywords: FriCAS

Author: Martin Rubey

Branch/Commit: e281742

Reviewer: Bill Page, Emmanuel Charpentier

Issue created by migration from https://trac.sagemath.org/ticket/21231

The text was updated successfully, but these errors were encountered:

mantepse · 2016-08-12T11:08:31Z

comment:1

implement a method that (reliably!) yields a tuple (type, 1d algebra output, 2d algebra output, error message, etc) as strings. 1d algebra output should come in "unparsed inputform" as one very long string, I'd say. 2d algebra output should be empty if 1d output is available.

It should be able to deal with very long lines, too, possibly be using file io. This involves use of ioHook and some regexps.

implement a method that translates fricas output into sage types.

I think this could work as follows: we want to map fricas types to sage constructors, possibly recursively. Recall that what's really sent to fricas is something like "sage23 := [x<sup>n/y</sup>n for n in 1..3]", so that sage can access the result using the variable "sage23". This string is referred to as self._name

Now what I propose is a method which takes a parsed type, e.g., "Integer" or ("List", "Integer") or
("UnivariatePolynomial", "x", ("Fraction", "Integer"))

def to_sage(type):
    if type == "Integer":
         return to_sage_integer()
    elif type == "String":
         return to_sage_string()
    ...
    elif isinstance(type, tuple):
        if type[0] == "List":
            return to_sage_list(type[1])
        elif type[0] == "Fraction":
            return to_sage_fraction(type[1])
        ....

and so on.

So, if the type is ("List" ("Fraction" ("UnivariatePolynomial" "x" "Integer"))), this method would call

    to_sage_list(("Fraction" ("UnivariatePolynomial" "x" "Integer"))

which might be something like

def to_sage_list(self, type):
    n = fricas.eval("#" + self._name).to_sage_integer()
    return [fricas.eval(self._name + ".%n").to_sage(type) for n in range(1,n+1)]

mantepse · 2016-08-12T23:37:20Z

Commit: 5b7e6ab

mantepse · 2016-08-12T23:37:20Z

Branch: /u/mantepse/21231

mantepse · 2016-08-12T23:37:20Z

New commits:

`5fa5c04`	`Merge branch 'u/mantepse/generating_function_in_findstat_interface' of git://trac.sagemath.org/sage into develop`
`4b555c5`	`Merge branch 'develop' of git://github.com/sagemath/sage into develop`
`17f7dbe`	`Merge branch 'u/dkrenn/16137' of git://trac.sagemath.org/sage into develop`
`08494a2`	`Merge branch 'u/vdelecroix/16137' of git://trac.sagemath.org/sage into develop`
`f077211`	`Merge branch 'develop' of git://github.com/sagemath/sage into develop`
`5b7e6ab`	`initial version of new fricas interface`

mantepse · 2016-08-12T23:38:43Z

comment:4

This is very beta, but I'd be very happy to receive some comments.

mantepse · 2016-08-12T23:51:55Z

Author: Martin Rubey

mantepse · 2016-08-12T23:51:55Z

Changed commit from 5b7e6ab to d8a26e1

mantepse · 2016-08-12T23:51:55Z

New commits:

`d8a26e1`	`Merge remote-tracking branch 'origin/develop' into fricas-interface`

vbraun · 2016-08-13T21:04:13Z

Changed branch from /u/mantepse/21231 to u/mantepse/21231

vbraun · 2016-08-13T21:04:13Z

comment:6

Branch name doesn't have leading slash

New commits:

`b2ee8cc`	`make it build`

vbraun · 2016-08-13T21:04:13Z

Changed commit from d8a26e1 to b2ee8cc

mantepse · 2016-08-14T07:54:29Z

comment:7

Replying to @vbraun:

Branch name doesn't have leading slash

Howto? I used

git push --set-upstream trac HEAD:u/mantepse/21231

dimpase · 2016-08-14T09:56:42Z

comment:8

Replying to @mantepse:

Replying to @vbraun:

Branch name doesn't have leading slash

Howto? I used
git push --set-upstream trac HEAD:u/mantepse/21231

This is fine, this does not do anything to the webpage contents. You made a typo, this leading slash, while editing the latter.

dimpase · 2016-08-14T15:15:42Z

Dependencies: #21209

mantepse · 2016-08-14T15:28:09Z

comment:10

(it actually doesn't depend on #21209, since it works with any installation of FriCAS but never mind)

sagetrac-git · 2016-08-14T15:54:28Z

Changed commit from b2ee8cc to cbcb94f

sagetrac-git · 2016-08-14T15:54:28Z

Branch pushed to git repo; I updated commit sha1. New commits:

`cbcb94f`	`quick dirty fix for bug reported by Bill Page, add idea for better type handling`

sagetrac-bpage · 2016-08-15T01:59:48Z

Sage worksheet under 7.2

sagetrac-bpage · 2016-08-15T02:01:17Z

Attachment: sage-7.2-fricas.pdf.gz

Attachment: sage-7.4.beta0-fricas.pdf.gz

Sage worksheet under 7.4.beta0 (typeset output fails)

sagetrac-bpage · 2016-08-15T02:07:31Z

comment:12

In Sage Worksheet typeset output or FriCAS object fails with this patch on 7.4.beta0. See attached pdf files.

https://github.com/sagemath/sage-prod/files/10658907/sage-7.2-fricas.pdf.gz (typeset output OK)

https://github.com/sagemath/sage-prod/files/10658908/sage-7.4.beta0-fricas.pdf.gz (typeset output fails)

I am not sure whether or not this example worked under 7.4.beta0 without the patch.

sagetrac-git · 2016-08-15T09:37:57Z

Changed commit from cbcb94f to 5b4a95d

sagetrac-git · 2016-08-15T09:37:57Z

Branch pushed to git repo; I updated commit sha1. New commits:

`5b4a95d`	`switch to better type treatment`

sagetrac-git · 2016-08-16T00:07:49Z

Changed commit from 5b4a95d to 0ab0f92

sagetrac-git · 2016-08-16T00:07:49Z

Branch pushed to git repo; I updated commit sha1. New commits:

`0ab0f92`	`further improve translation to sage objects, make restart work, fix doctests`

mantepse · 2016-08-16T00:09:01Z

comment:15

I have a question about the _fricas_init_ and _fricas_ methods
defined in various files, in particular in
~/sage-develop/src/sage/rings/rational_field.py and in ~/sage-develop/src/sage/rings/real_mpfr.pyx, see below.

The question is: the doctests work without adapting the methods.
What are the methods supposed to do?

~/sage-develop/src/sage/rings/rational_field.py

    def _axiom_init_(self):
        r"""
        Return the axiom/fricas representation of `\QQ`.

        EXAMPLES::

           sage: axiom(QQ)    #optional - axiom # indirect doctest
           Fraction Integer
           sage: fricas(QQ)   #optional - fricas # indirect doctest
           Fraction(Integer)

        """
        return 'Fraction Integer'

    _fricas_init_ = _axiom_init_

in ~/sage-develop/src/sage/rings/real_mpfr.pyx

    def _axiom_(self, axiom):
        """
        Return ``self`` as a floating point number in Axiom.

        EXAMPLES::

            sage: R = RealField(100)
            sage: R(pi)
            3.1415926535897932384626433833
            sage: axiom(R(pi))  # optional - axiom # indirect doctest
            3.1415926535 8979323846 26433833
            sage: fricas(R(pi)) # optional - fricas
            3.1415926535_8979323846_26433833

        """
        prec = self.parent().prec()

        #Set the precision in Axiom
        old_prec = axiom('precision(%s)$Float'%prec)
        res = axiom('%s :: Float'%self.exact_rational())
        axiom.eval('precision(%s)$Float'%old_prec)

        return res

    _fricas_ = _axiom_

sagetrac-git · 2016-08-16T00:58:48Z

Changed commit from 0ab0f92 to d2d0fb8

sagetrac-git · 2016-10-10T10:25:49Z

Changed commit from 70c3ace to ab4214a

sagetrac-git · 2016-10-10T10:25:49Z

Branch pushed to git repo; I updated commit sha1 and set ticket back to needs_review. New commits:

`ab4214a`	`Merge branch 'develop' of git://trac.sagemath.org/sage into u/mantepse/fricas_interface`

dimpase · 2016-10-10T10:40:42Z

comment:97

the latest push created a mega-patch for no good reason. What was the reason for this?

mantepse · 2016-10-10T10:43:14Z

comment:98

I am sorry, I thought I am supposed to merge when a new develop appears. Could you tell me how to revert?

dimpase · 2016-10-10T10:55:11Z

comment:99

Replying to @mantepse:

I am sorry, I thought I am supposed to merge when a new develop appears. Could you tell me how to revert?

only if there are merge conflicts (which you would see by the Branch field in the ticket turning red).

Just revert the last commit (and push again, with -f switch). Not sure how to do this using git trac if you must use it...
Let me know if you get stuck on this.

sagetrac-git · 2016-10-10T11:08:59Z

Changed commit from ab4214a to 70c3ace

sagetrac-git · 2016-10-10T11:08:59Z

Branch pushed to git repo; I updated commit sha1. This was a forced push. New commits:

mantepse · 2016-10-10T11:10:48Z

comment:101

Replying to @dimpase:

Thank you!

dimpase · 2016-10-10T11:22:26Z

comment:102

OK, good, back to the positively reviewed patch.

jdemeyer · 2016-10-12T06:20:24Z

comment:103

You should not use a bare except: (see the end of src/sage/symbolic/integration/external.py). Moreover, I would suggest to replace

     try:
        return result.sage()
     except:
        raise ValueError("Unable to parse: {}".format(result))

by

return result.sage()

sagetrac-git · 2016-10-12T06:31:12Z

Branch pushed to git repo; I updated commit sha1. New commits:

`e281742`	`suggestion by reviewer`

sagetrac-git · 2016-10-12T06:31:12Z

Changed commit from 70c3ace to e281742

mantepse · 2016-10-12T06:32:19Z

comment:105

I agree - I only adapted this part of the code.

mantepse · 2016-10-12T08:22:39Z

comment:106

could you set this back to positive review? I would really like to get this into sage soon, because there will be trivial, but tedious merge conflicts all the time otherwise. I'd be grateful...

jdemeyer · 2016-10-12T08:38:30Z

comment:107

Well, if you don't set it to needs_review, I don't know that it's ready for review.

vbraun · 2016-10-21T07:04:20Z

Changed branch from u/mantepse/fricas_interface to e281742

mantepse added this to the sage-7.4 milestone Aug 12, 2016

mantepse added c: interfaces labels Aug 12, 2016

mantepse added the t: enhancement label Aug 12, 2016

EmmanuelCharpentier mannequin added the s: positive review label Oct 3, 2016

sagetrac-git mannequin added s: needs review and removed s: positive review labels Oct 10, 2016

dimpase added s: positive review and removed s: needs review labels Oct 10, 2016

jdemeyer added s: needs work and removed s: positive review labels Oct 12, 2016

jdemeyer added s: needs review and removed s: needs work labels Oct 12, 2016

vbraun removed the s: positive review label Oct 21, 2016

vbraun closed this as completed in e2c09d5 Oct 21, 2016

jdemeyer mentioned this issue Oct 29, 2016

Avoid bare "except:" statements #21687

Closed

improve FriCAS interface #21231

improve FriCAS interface #21231

Comments

mantepse commented Aug 12, 2016

mantepse commented Aug 12, 2016

mantepse commented Aug 12, 2016

mantepse commented Aug 12, 2016

mantepse commented Aug 12, 2016

mantepse commented Aug 12, 2016

mantepse commented Aug 12, 2016

mantepse commented Aug 12, 2016

mantepse commented Aug 12, 2016

vbraun commented Aug 13, 2016

vbraun commented Aug 13, 2016

vbraun commented Aug 13, 2016

mantepse commented Aug 14, 2016

dimpase commented Aug 14, 2016

dimpase commented Aug 14, 2016

mantepse commented Aug 14, 2016

sagetrac-git mannequin commented Aug 14, 2016

sagetrac-git mannequin commented Aug 14, 2016

sagetrac-bpage mannequin commented Aug 15, 2016

sagetrac-bpage mannequin commented Aug 15, 2016

sagetrac-bpage mannequin commented Aug 15, 2016

sagetrac-git mannequin commented Aug 15, 2016

sagetrac-git mannequin commented Aug 15, 2016

sagetrac-git mannequin commented Aug 16, 2016

sagetrac-git mannequin commented Aug 16, 2016

mantepse commented Aug 16, 2016

sagetrac-git mannequin commented Aug 16, 2016

sagetrac-git mannequin commented Oct 10, 2016

sagetrac-git mannequin commented Oct 10, 2016

dimpase commented Oct 10, 2016

mantepse commented Oct 10, 2016

dimpase commented Oct 10, 2016

sagetrac-git mannequin commented Oct 10, 2016

sagetrac-git mannequin commented Oct 10, 2016

mantepse commented Oct 10, 2016

dimpase commented Oct 10, 2016

jdemeyer commented Oct 12, 2016

sagetrac-git mannequin commented Oct 12, 2016

sagetrac-git mannequin commented Oct 12, 2016

mantepse commented Oct 12, 2016

mantepse commented Oct 12, 2016

jdemeyer commented Oct 12, 2016

vbraun commented Oct 21, 2016