Closed RussLaher closed 4 years ago
RussLaher
commented
4 years ago Here is more version info:
Successfully installed astropy-4.0 Successfully installed barycorrpy-0.2.2.1
$ python --version Python 3.7.4
$ which python /home1/07040/rrlaher7/anaconda3/bin/python
RussLaher
commented
4 years ago I downgraded numpy on the TACC machine to 1.16.2 and reran the barycorrpy tests successfully.
shbhuk
commented
4 years ago Hi @RussLaher, Thanks for bringing this up. I will have to check what part of the Astropy Quantities routine is failing under the new Numpy version. I will attempt to fix this in the new version release - v0.3
shbhuk
commented
4 years ago Hi, I'm reopening this issue so that I can remember to look into this numpy update issue and figure out what is breaking..
njcuk9999
commented
4 years ago Also getting this, maybe the traceback will help solve this faster (we don't really want to have to downgrade packages if possible)
bkwargs = {'ra': 263.3741740568687, 'dec': -5.745214357217233, 'epoch': 2457205.9999942128,
'px': 21.353919213862373, 'pmra': -41.93841615099209, 'pmdec': -95.48427622624564, 'lat': 19.825252, 'longi': -155.468876, 'alt': 4204.0, 'rv': 0.0, 'leap_update': False}
barycorrpy.get_BC_vel(2458849.04505247, zmeas=0.0, **bkwargs)/scratch/bin/anaconda3/envs/aperoenv/lib/python3.7/site-packages/barycorrpy/barycorrpy.py in get_BC_vel(JDUTC, starname, hip_id, ra, dec, epoch, pmra, pmdec, px, rv, obsname, lat, longi, alt, zmeas, ephemeris, leap_dir, leap_update)
157 zmeas=zm,
158 loc=loc,
--> 159 ephemeris=ephemeris, leap_dir=leap_dir, leap_update=leap_update,**star_output)
160 vel.append(a[0])
161 warning.append(a[1])
/scratch/bin/anaconda3/envs/aperoenv/lib/python3.7/site-packages/barycorrpy/barycorrpy.py in BCPy(JDUTC, ra, dec, epoch, pmra, pmdec, px, rv, zmeas, loc, ephemeris, leap_dir, leap_update)
189 ##### NUTATION, PRECESSION, ETC. #####
190
--> 191 r_pint, v_pint = PINT.gcrs_posvel_from_itrf(loc, JDUTC, JDTT)
192
193 r_eci = r_pint[0] # [m]
/scratch/bin/anaconda3/envs/aperoenv/lib/python3.7/site-packages/barycorrpy/PINT_erfautils.py in gcrs_posvel_from_itrf(loc, toas, tts)
86
87 # Get dX and dY from IERS B in arcsec and convert to radians
---> 88 dX = np.interp(mjds, iers_tab['MJD'], iers_tab['dX_2000A']) * asec2rad
89 dY = np.interp(mjds, iers_tab['MJD'], iers_tab['dY_2000A']) * asec2rad
90
<__array_function__ internals> in interp(*args, **kwargs)
/scratch/bin/anaconda3/envs/aperoenv/lib/python3.7/site-packages/astropy/units/quantity.py in __array_function__(self, function, types, args, kwargs)
1496 function_helper = FUNCTION_HELPERS[function]
1497 try:
-> 1498 args, kwargs, unit, out = function_helper(*args, **kwargs)
1499 except NotImplementedError:
1500 return self._not_implemented_or_raise(function, types)
/scratch/bin/anaconda3/envs/aperoenv/lib/python3.7/site-packages/astropy/units/quantity_helper/function_helpers.py in interp(x, xp, fp, *args, **kwargs)
812 from astropy.units import Quantity
813
--> 814 (x, xp), _ = _quantities2arrays(x, xp)
815 if isinstance(fp, Quantity):
816 unit = fp.unit
/scratch/bin/anaconda3/envs/aperoenv/lib/python3.7/site-packages/astropy/units/quantity_helper/function_helpers.py in _quantities2arrays(unit_from_first, *args)
348 # as we want to allow arbitrary unit for 0, inf, and nan.
349 try:
--> 350 arrays = tuple((q._to_own_unit(arg)) for arg in args)
351 except TypeError:
352 raise NotImplementedError
/scratch/bin/anaconda3/envs/aperoenv/lib/python3.7/site-packages/astropy/units/quantity_helper/function_helpers.py in <genexpr>(.0)
348 # as we want to allow arbitrary unit for 0, inf, and nan.
349 try:
--> 350 arrays = tuple((q._to_own_unit(arg)) for arg in args)
351 except TypeError:
352 raise NotImplementedError
/scratch/bin/anaconda3/envs/aperoenv/lib/python3.7/site-packages/astropy/units/quantity.py in _to_own_unit(self, value, check_precision)
1337 try:
1338 as_quantity = Quantity(value)
-> 1339 _value = as_quantity.to_value(self.unit)
1340 except TypeError:
1341 # Could not make a Quantity. Maybe masked printing?
/scratch/bin/anaconda3/envs/aperoenv/lib/python3.7/site-packages/astropy/units/quantity.py in to_value(self, unit, equivalencies)
732 except Exception:
733 # Short-cut failed; try default (maybe equivalencies help).
--> 734 value = self._to_value(unit, equivalencies)
735 else:
736 value = self.view(np.ndarray)
/scratch/bin/anaconda3/envs/aperoenv/lib/python3.7/site-packages/astropy/units/quantity.py in _to_value(self, unit, equivalencies)
663 equivalencies = self._equivalencies
664 return self.unit.to(unit, self.view(np.ndarray),
--> 665 equivalencies=equivalencies)
666
667 def to(self, unit, equivalencies=[]):
/scratch/bin/anaconda3/envs/aperoenv/lib/python3.7/site-packages/astropy/units/core.py in to(self, other, value, equivalencies)
985 return UNITY
986 else:
--> 987 return self._get_converter(other, equivalencies=equivalencies)(value)
988
989 def in_units(self, other, value=1.0, equivalencies=[]):
/scratch/bin/anaconda3/envs/aperoenv/lib/python3.7/site-packages/astropy/units/core.py in _get_converter(self, other, equivalencies)
916 pass
917
--> 918 raise exc
919
920 def _to(self, other):
/scratch/bin/anaconda3/envs/aperoenv/lib/python3.7/site-packages/astropy/units/core.py in _get_converter(self, other, equivalencies)
902 try:
903 return self._apply_equivalencies(
--> 904 self, other, self._normalize_equivalencies(equivalencies))
905 except UnitsError as exc:
906 # Last hope: maybe other knows how to do it?
/scratch/bin/anaconda3/envs/aperoenv/lib/python3.7/site-packages/astropy/units/core.py in _apply_equivalencies(self, unit, other, equivalencies)
886 raise UnitConversionError(
887 "{} and {} are not convertible".format(
--> 888 unit_str, other_str))
889
890 def _get_converter(self, other, equivalencies=[]):
UnitConversionError: '' (dimensionless) and 'd' (time) are not convertibleGoing into the offending function it seems that it is the np.interp function:
np.interp(mjds, iers_tab['MJD'], iers_tab['dX_2000A']) * asec2radwhere for me:
In [59]: mjds
Out[59]: array([58848.54505247])
In [60]: iers_tab['MJD']
Out[60]: <Quantity [37665., 37666., 37667., ..., 58748., 58749., 58750.] d>
In [61]: iers_tab['dX_2000A']
Out[61]: <Quantity [0.0e+00, 0.0e+00, 0.0e+00, ..., 2.0e-06, 1.1e-05, 1.9e-05] arcsec>So your iers_tab['MJD'] has units "d" and mjds has no units...hence the UnitConversionError I'm not sure how this worked in the past... unless now astropy is pushing units to iers_tab['MJD'] when it wasn't before or np.interp was stripping units in the past... (I would be more likely to believe astropy changed... but it downgrading numpy worked it would suggest the latter).
you guess you have two solutions (1) add units to mjds or (2) remove units from iers_tab['MJD']
njcuk9999
commented
4 years ago i.e. for me this worked in my local barycorrpy files:
In PINT_erfautils.py:
dX = np.interp(mjds, iers_tab['MJD'].value, iers_tab['dX_2000A'].value) * asec2rad
dY = np.interp(mjds, iers_tab['MJD'].value, iers_tab['dY_2000A'].value) * asec2radand
xp = np.interp(mjds, iers_tab['MJD'].value, iers_tab['PM_x'].value) * asec2rad
yp = np.interp(mjds, iers_tab['MJD'].value, iers_tab['PM_y'].value) * asec2radbut again I'm not sure how you make older versions compatible -- that depends why it worked previously
shbhuk
commented
4 years ago Hi, I am not sure what is going on with the numpy updated versions, it is breaking some other packages of mine too.. I'll need to check
shbhuk
commented
4 years ago Okay, so I ran two different virtual envs, where I experimented with the astropy and numpy versions. Case I: Astropy 4.0, Numpy 1.16.4 = Worked Case II: Astropy 4.0, Numpy 1.18.1 = Fails Case III: Astropy 3.2.3, Numpy 1.16.4 = Worked Case IV: Astropy 3.2.3, Numpy 1.18.1 = Worked
So its combination of the latest Numpy and Astropy updates that are causing this. Very interesting..
astrowright
commented
4 years ago I might post a bug report to the astropy folks—I suspect they're aware of the problem and can recommend a solution.
On Sun, Jan 19, 2020 at 4:31 PM Shubham notifications@github.com wrote:
Okay, so I ran two different virtual envs, where I experimented with the astropy and numpy versions. Case I: Astropy 4.0, Numpy 1.16.4 = Worked Case II: Astropy 4.0, Numpy 1.18.1 = Fails Case III: Astropy 3.2.3, Numpy 1.16.4 = Worked Case IV: Astropy 3.2.3, Numpy 1.18.1 = Worked
So its combination of the latest Numpy and Astropy updates that are causing this. Very interesting..
— You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub https://github.com/shbhuk/barycorrpy/issues/31?email_source=notifications&email_token=AASHFHJDDFSXAXDVW47YGELQ6TBEFA5CNFSM4KEIRGVKYY3PNVWWK3TUL52HS4DFVREXG43VMVBW63LNMVXHJKTDN5WW2ZLOORPWSZGOEJK5EAA#issuecomment-576049664, or unsubscribe https://github.com/notifications/unsubscribe-auth/AASHFHJPVYCBZGS2OMPJMCTQ6TBEFANCNFSM4KEIRGVA .
Jason T Wright Professor of Astronomy & Astrophysics Deputy Director, Center for Exoplanets and Habitable Worlds NEID Project Scientist http://neid.psu.edu/ he/him/his https://sites.psu.edu/astrowright/ https://sites.psu.edu/astrowright/ @Astro_Wright
shbhuk
commented
4 years ago So, @njcuk9999 iers_tab['mjd'] has had units of days in both Astropy 3.2.3 and 4.0, so that's not the issue. The numpy documentation for np.interp does not seem to have changed.
What baffles me is that the combination of Astropy 3.2.3, Numpy 1.18.1, seems to work. I have checked the inputs with this sample script
from astropy.utils.iers import conf,IERS_A, IERS_A_URL, IERS_B, IERS_B_URL, IERS
from astropy.utils.data import download_file
from astropy.time import Time
import astropy.table as table
import numpy as np
import astropy
# Load IERS table
iers_b_file = download_file(IERS_B_URL, cache=True)
iers_b = IERS_B.open(iers_b_file)
IERS.iers_table = iers_b
iers_tab = IERS.iers_table
mjd_astropy = iers_tab['MJD']
JDUTC = Time(2458000, format='jd', scale='utc')
JDTT = Time(2458000, format='jd', scale='tt')
toas = JDUTC
if toas.isscalar:
ttoas = Time([toas])
N = len(ttoas)
mjds = np.asarray(ttoas.mjd)
PM = iers_tab['PM_x'].value
# Check version and interpolation
print(astropy.__version__,':', np.__version__)
print(mjds, mjd_astropy, PM)
print(np.interp(mjds, mjd_astropy.value, PM))
print(np.interp(mjds, mjd_astropy, PM))I'm not sure what exactly is causing this emergent error, however for now I will include an update with v0.3 where I shall modify the code to use just the values from the Astropy quantities for the interpolation in PINT_erfautils
shbhuk
commented
4 years ago As Jason @astrowright suggested, I think the issue might be this update under Astropy -
Astropy 4.0, Units: : Point 3
I installed python3 via Anaconda3-2019.10-Linux-x86_64.sh on TACC machine Lonestar5:
Operating System: SUSE Linux Enterprise Server 12 SP3 CPE OS Name: cpe:/o:suse:sles:12:sp3 Kernel: Linux 4.4.103-6.38-default Architecture: x86-64
which includes astropy-4.0, astroquery-0.3.10, barycorrpy-0.2.2.1, jplephem-2.12, and astroscrappy-1.0.8.
cd ~/barycorrpy/ export PATH=/home1/07040/rrlaher7/anaconda3/bin:$PATH
$ python -m unittest discover |& grep -i error ERROR: test_JDUTC_BJDTDB (barycorrpy.tests.test_barycorrpy.Barycorrpy_tests) AttributeError: 'numpy.ndarray' object has no attribute 'to_value' astropy.units.core.UnitConversionError: 'Unit(dimensionless)' is not a scaled version of 'Unit("d")' astropy.units.core.UnitConversionError: '' (dimensionless) and 'd' (time) are not convertible ERROR: test_flux_weighting (barycorrpy.tests.test_barycorrpy.Barycorrpy_tests) AttributeError: 'numpy.ndarray' object has no attribute 'to_value' astropy.units.core.UnitConversionError: 'Unit(dimensionless)' is not a scaled version of 'Unit("d")' astropy.units.core.UnitConversionError: '' (dimensionless) and 'd' (time) are not convertible ERROR: test_hip_id (barycorrpy.tests.test_barycorrpy.Barycorrpy_tests) AttributeError: 'numpy.ndarray' object has no attribute 'to_value' astropy.units.core.UnitConversionError: 'Unit(dimensionless)' is not a scaled version of 'Unit("d")' astropy.units.core.UnitConversionError: '' (dimensionless) and 'd' (time) are not convertible ERROR: test_hip_id_astropy_obs (barycorrpy.tests.test_barycorrpy.Barycorrpy_tests) AttributeError: 'numpy.ndarray' object has no attribute 'to_value' astropy.units.core.UnitConversionError: 'Unit(dimensionless)' is not a scaled version of 'Unit("d")' astropy.units.core.UnitConversionError: '' (dimensionless) and 'd' (time) are not convertible ERROR: test_inputs (barycorrpy.tests.test_barycorrpy.Barycorrpy_tests) AttributeError: 'numpy.ndarray' object has no attribute 'to_value' astropy.units.core.UnitConversionError: 'Unit(dimensionless)' is not a scaled version of 'Unit("d")' astropy.units.core.UnitConversionError: '' (dimensionless) and 'd' (time) are not convertible FAILED (errors=5)
Any suggestions on how to proceed?