Tensorflow array printers

[Upabjojr]

References to other Issues or PRs

Brief description of what is fixed or changed

Other comments

Release Notes

• printing
  • additions to the tensorflow printers.

[sympy-bot]

✅

Hi, I am the SymPy bot (v134). 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:

• printing
  • additions to the tensorflow printers. (#15462 by @Upabjojr)

This will be added to https://github.com/sympy/sympy/wiki/Release-Notes-for-1.4.

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 .-->


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


#### 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 -->

* printing
  * additions to the tensorflow printers.

<!-- END RELEASE NOTES -->

Update

The release notes on the wiki have been updated.

[asmeurer]

This seems to modify the numpy printers too. I don't see any tests added for tensorflow.

[asmeurer]

I guess they are changes from https://github.com/sympy/sympy/pull/15452/files. GitHub isn't noticing that the commit is in master now.

[Upabjojr]

GitHub isn't noticing that the commit is in master now.

Yeah, maybe I should merge the master branch.

[Upabjojr]

@asmeurer I use this trick to call tensorflow. Tensorflow requires the input data to be declared as placeholders, there was no way I could put this in a lambda as I need to define some local variables.

Furthermore the objects in sympy.codegen are of little use to me, as they all look to be optimized for C and Fortran code generation.

Probably this can be done in a better way, but I'd keep this trick as a temporary solution to concentrate on the code generation.

[asmeurer]

That is fine. Even lambdify doesn't use a lambda anymore.

Regarding codegen, what is missing? We would like for it to be as language agnostic as possible, but so far only C and Fortran have been in mind, so some of the abstractions may be wrong. CC @bjodah

[Upabjojr]

From what I've seen the function prototypes need to define the type already, which is the same for all variables of the function. This is a bit of a rigid concept, I think we should rather have the possibility of defining the type for each variable.

Anyway, there are other problems beyond functions that may arise, so I'll stick with this solution.

[Upabjojr]

It looks like there is a test that randomly fails on Python 3.4:

_______ sympy/solvers/tests/test_solvers.py:test_high_order_multivariate _______
Traceback (most recent call last):
  File "/home/travis/virtualenv/python3.4.6/lib/python3.4/site-packages/sympy-1.4.dev0-py3.4.egg/sympy/core/evalf.py", line 1306, in evalf
    rf = evalf_table[x.func]
KeyError: <class 'sympy.polys.rootoftools.ComplexRootOf'>

[asmeurer]

Can you open a new issue for the failure, along with the test header (which includes the random seeds).

[asmeurer]

Should this be tf, assuming the commented out tests are uncommented?

[Upabjojr]

Still uncertain, we probably need both tensorflow and numpy.

[Upabjojr]

Can you open a new issue for the failure, along with the test header (which includes the random seeds).

I've restarted the build.

[Upabjojr]

So, apparently the current generated function for command

lambdify((M, N), M*N, 'tensorflow')

is

def _lambdifygenerated(M, N):
    return (M*N)

which means it does not use tensorflow at all.

This has to be fixed.

[asmeurer]

Does tensorflow still not use operator overloading?

[Upabjojr]

I wasn't referring to operator overloading. Tensorflow uses placeholders as input streams in the computational graph. The generated lambda does not perform the computational immediately. But maybe it's the correct way to go.

[asmeurer]

That's probably the most useful. The user can always perform the final calculation on the result.