Propagating 0 seconds with cowell fails

[shreyasbapat]

Currently we can only use the basic mean_method function to propagate orbits. As the Orbit.Sample() can now accept other propagators as well, we must be able to use cowell method as well.
🐞 Problem
We can't use the cowell's method and other propagators to propagate the orbit because of differences in the parameters passed in the normal propagator while using Orbit.Sample().

🖥 Please paste the output of following commands

conda info -a

python -c "import poliastro.testing; poliastro.testing.test()"
python3 -c "import poliastro.testing; poliastro.testing.test()"
============================= test session starts ==============================
platform linux -- Python 3.6.3, pytest-3.4.1, py-1.5.2, pluggy-0.6.0
rootdir: /home/shreyas/poliastro, inifile: setup.cfg
plugins: cov-2.5.1
collected 141 items

poliastro/src/poliastro/tests/test_bodies.py ...... [ 4%]
poliastro/src/poliastro/tests/test_coordinates.py ... [ 6%]
poliastro/src/poliastro/tests/test_hyper.py ........... [ 14%]
poliastro/src/poliastro/tests/test_iod.py ......... [ 20%]
poliastro/src/poliastro/tests/test_jit.py ... [ 22%]
poliastro/src/poliastro/tests/test_maneuver.py ...... [ 26%]
poliastro/src/poliastro/tests/test_patched_conics.py .. [ 28%]
poliastro/src/poliastro/tests/test_plotting.py ......... [ 34%]
poliastro/src/poliastro/tests/test_plotting3d.py ...... [ 39%]
poliastro/src/poliastro/tests/test_stumpff.py ... [ 41%]
poliastro/src/poliastro/tests/test_twobody.py .. [ 42%]
poliastro/src/poliastro/tests/test_util.py ....... [ 47%]
poliastro/src/poliastro/tests/tests_neos/test_dastcom5.py ......... [ 53%]
poliastro/src/poliastro/tests/tests_neos/test_neos_neows.py ...... [ 58%]
poliastro/src/poliastro/tests/tests_twobody/test_angles.py ....... [ 63%]
poliastro/src/poliastro/tests/tests_twobody/test_decorators.py .. [ 64%]
poliastro/src/poliastro/tests/tests_twobody/test_orbit.py .............. [ 74%]
..... [ 78%]
poliastro/src/poliastro/tests/tests_twobody/test_propagation.py ........ [ 83%]
. [ 84%]
poliastro/src/poliastro/tests/tests_twobody/test_sample.py ............ [ 92%]
poliastro/src/poliastro/tests/tests_twobody/test_states.py .......... [100%]

========================= 141 passed in 21.47 seconds ==========================

🎯 Goal
To be able to Propagate the orbit using different propagators.

💡 Possible solutions

📋 Steps to solve the problem

• Comment below about what you've started working on.
• Add, commit, push your changes
• Submit a pull request and add this in comments - Addresses #<put issue number here>
• Ask for a review in comments section of pull request
• Celebrate your contribution to this project 🎉

[astrojuanlu]

Can you check if this is still the case after merging #322? Also, is it a specific problem of cowell or affects other propagators as well? Can you ellaborate?

[shreyasbapat]

Yes! It's still the case. Wait the names have changed and hence it works for mean_motion and kepler .But not for cowell.

---------------------------------------------------------------------------
RuntimeError                              Traceback (most recent call last)
<ipython-input-16-c620f81ccde0> in <module>()
----> 1 iss.sample(None,method=cowell)

~/poliastro/src/poliastro/twobody/orbit.py in sample(self, values, method)
    292         """
    293         if values is None:
--> 294             return self.sample(100, method)
    295 
    296         elif isinstance(values, int):

~/poliastro/src/poliastro/twobody/orbit.py in sample(self, values, method)
    308                 nu_values = np.insert(nu_values, 0, self.ecc)
    309 
--> 310             return self.sample(nu_values, method)
    311 
    312         elif hasattr(values, "unit") and values.unit in ('rad', 'deg'):

~/poliastro/src/poliastro/twobody/orbit.py in sample(self, values, method)
    312         elif hasattr(values, "unit") and values.unit in ('rad', 'deg'):
    313             values = self._generate_time_values(values)
--> 314         return (values, self._sample(values, method))
    315 
    316     def _sample(self, time_values, method=mean_motion):

~/poliastro/src/poliastro/twobody/orbit.py in _sample(self, time_values, method)
    317         values = np.zeros((len(time_values), 3)) * self.r.unit
    318         for ii, epoch in enumerate(time_values):
--> 319             rr = self.propagate(epoch, method).r
    320             values[ii] = rr
    321 

~/poliastro/src/poliastro/twobody/orbit.py in propagate(self, epoch_or_duration, method, rtol)
    254             time_of_flight = time.TimeDelta(epoch_or_duration)
    255 
--> 256         return propagate(self, time_of_flight, method=method, rtol=rtol)
    257 
    258     def sample(self, values=None, method=mean_motion):

~/poliastro/src/poliastro/twobody/propagation.py in propagate(orbit, time_of_flight, method, rtol, **kwargs)
    213                   time_of_flight.to(u.s).value,
    214                   rtol=rtol,
--> 215                   **kwargs)
    216     return orbit.from_vectors(orbit.attractor, r * u.km, v * u.km / u.s, orbit.epoch + time_of_flight)
    217 

~/poliastro/src/poliastro/twobody/propagation.py in cowell(k, r0, v0, tof, rtol, ad, callback, nsteps)
    102         r, v = rr.y[:3], rr.y[3:]
    103     else:
--> 104         raise RuntimeError("Integration failed")
    105 
    106     return r, v

RuntimeError: Integration failed

[shreyasbapat]

I realized that I could forget this hence I put it here.

[astrojuanlu]

This looks like a convergence problem with the cowell method, rather than an API problem:

In [25]: iss.propagate(0 * u.s, method=cowell)
/home/juanlu/.miniconda36/envs/poliastro36/lib/python3.6/site-packages/scipy/integrate/_ode.py:1095: UserWarning: dop853: step size becomes too small
  self.messages.get(istate, unexpected_istate_msg)))
---------------------------------------------------------------------------
RuntimeError                              Traceback (most recent call last)
<ipython-input-25-5ec3bbe8e578> in <module>()
----> 1 iss.propagate(0 * u.s, method=cowell)

~/Development/poliastro/poliastro-library/src/poliastro/twobody/orbit.py in propagate(self, epoch_or_duration, method, rtol)
    254             time_of_flight = time.TimeDelta(epoch_or_duration)
    255 
--> 256         return propagate(self, time_of_flight, method=method, rtol=rtol)
    257 
    258     def sample(self, values=None, method=mean_motion):

~/Development/poliastro/poliastro-library/src/poliastro/twobody/propagation.py in propagate(orbit, time_of_flight, method, rtol, **kwargs)
    213                   time_of_flight.to(u.s).value,
    214                   rtol=rtol,
--> 215                   **kwargs)
    216     return orbit.from_vectors(orbit.attractor, r * u.km, v * u.km / u.s, orbit.epoch + time_of_flight)
    217 

~/Development/poliastro/poliastro-library/src/poliastro/twobody/propagation.py in cowell(k, r0, v0, tof, rtol, ad, callback, nsteps)
    102         r, v = rr.y[:3], rr.y[3:]
    103     else:
--> 104         raise RuntimeError("Integration failed")
    105 
    106     return r, v

RuntimeError: Integration failed

of course, sending a value of 0 seconds should return just the same position :) An easy issue to fix!

[shreyasbapat]

Yep! True!

[astrojuanlu]

I think that iss.sample(method=cowell) won't raise the error after this was merged:

#330

However, iss.propagate(0 * u.s, method=cowell) or iss.sample([0] * u.s, method=cowell) should do.

[shreyasbapat]

Working on it!

[astrojuanlu]

@shreyasbapat C'mon, leave some easy issues for the newcomers 😜

[shreyasbapat]

@Juanlu001 Not Working Now 😉

[astrojuanlu]

I still want some newcomer to fix this issue, so I'm removing the milestone.

[astrojuanlu]

Is this still the case with the new integrator @nikita-astronaut?

[nikita-astronaut]

This still fails, but the error is different: the new integrator does not do anything and returns nothing :) I will fix this in the next PR

[nikita-astronaut]

Hey, @Juanlu001

After the last PR, 'iss.propagate(0 * u.s, method=cowell)' forks nice. But

>>> iss.sample(np.array([0]) * u.s, method=cowell) Traceback (most recent call last): File "<stdin>", line 1, in <module> File "/home/astronaut/anaconda3/lib/python3.6/site-packages/poliastro/twobody/orbit.py", line 334, in sample return values, self._sample(values, method) File "/home/astronaut/anaconda3/lib/python3.6/site-packages/poliastro/twobody/orbit.py", line 339, in _sample values, _ = cowell(self, (time_values - self.epoch).to(u.s).value) File "/home/astronaut/anaconda3/lib/python3.6/site-packages/astropy/time/core.py", line 1411, in __rsub__ return -out TypeError: bad operand type for unary -: 'Time'

[nikita-astronaut]

Seems like providing of time list from outside of orbit.sample works incorrect, but we don't use it so tests do not fall.

The reason is that now one needs to explicitly cast u.s to Time instance. Am I right?

[astrojuanlu]

Seems like providing of time list from outside of orbit.sample works incorrect, but we don't use it so tests do not fall.

Then we need more tests!

The reason is that now one needs to explicitly cast u.s to Time instance. Am I right?

poliastro, not the user :)

[astrojuanlu]

Let's try to fix this one before the next release.

[astrojuanlu]

Well... Or maybe the next after that!

[astrojuanlu]

I will fix this myself.

[shreyasbapat]

You said this was a beginner issue :/

[astrojuanlu]

The original issue was, but the problem changed and we continued discussing it here:

#328 (comment)

I need to think carefully on the API and which cases do we accept, and also there's one doctest failing because of it (see #454), so I prefer to take care of it myself.