Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Enable the use of different dtypes in observations #79

Merged
merged 3 commits into from
Dec 28, 2020
Merged
Show file tree
Hide file tree
Changes from all commits
Commits
File filter

Filter by extension

Filter by extension

Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
2 changes: 2 additions & 0 deletions CHANGELOG.md
Original file line number Diff line number Diff line change
@@ -1,5 +1,7 @@
# HEAD

- Add the ability to specify the size of the floating-point type used in `Observation` objects [PR#79](https://github.com/litebird/litebird_sim/pull/79)

- Simple bin map-maker [PR#73](https://github.com/litebird/litebird_sim/pull/73)

- Use SI units in class `SpinningScanningStrategy` (**breaking change**) [PR#69](https://github.com/litebird/litebird_sim/pull/69)
Expand Down
31 changes: 25 additions & 6 deletions docs/source/observations.rst
Original file line number Diff line number Diff line change
Expand Up @@ -48,12 +48,31 @@ object::
)

Note that the 2-D array ``obs.tod`` is created for you. Its shape is
``(n_detectors, n_samples)``. In full scale simulations it may get too large to
fit in memory. You can chunk it along the time or detector dimension
(or both) using ``n_blocks_det, n_blocks_time, comm`` at construction time or
the `set_n_blocks` method. The same chunking is applied also to any detector
information that you add with `detector_global_info` or `detector_info`. Note
note ``det_idx`` is added for you at construction.
``(n_detectors, n_samples)``, and the default type is
``numpy.float32``: the choice of a 32-bit type is usually good enough
for the purposes of LiteBIRD simulations, but if you need less/more
precision, you are free to use any of the floating-point types
provided by NumPy (``float16``, ``float32``, ``float64``,
``float128``) using the keyword ``dtype_tod``::

import numpy as np

obs = lbs.Observation(
detectors=2,
start_time=0.0,
sampling_rate_hz=5.0,
n_samples=5,
dtype_tod=np.float64, # Use a 64-bit floating point type
)


In full scale simulations the TOD may get too large to fit in memory.
You can chunk it along the time or detector dimension (or both) using
``n_blocks_det, n_blocks_time, comm`` at construction time or the
`set_n_blocks` method. The same chunking is applied also to any
detector information that you add with `detector_global_info` or
`detector_info`. Note note ``det_idx`` is added for you at
construction.

When you distribute the observation the first
``n_blocks_det x n_blocks_time`` MPI ranks are organized in a row-major grid
Expand Down
3 changes: 3 additions & 0 deletions litebird_sim/simulations.py
Original file line number Diff line number Diff line change
Expand Up @@ -25,6 +25,7 @@
import astropy.time
import astropy.units
import markdown
import numpy as np
import jinja2
import tomlkit

Expand Down Expand Up @@ -606,6 +607,7 @@ def create_observations(
n_blocks_det=1,
n_blocks_time=1,
root=0,
dtype_tod=np.float32,
):
"Create a set of Observation objects"

Expand Down Expand Up @@ -637,6 +639,7 @@ def create_observations(
n_blocks_time=n_blocks_time,
comm=(None if distribute else self.mpi_comm),
root=0,
dtype_tod=dtype_tod,
)
observations.append(cur_obs)

Expand Down
22 changes: 14 additions & 8 deletions test/test_simulations.py
Original file line number Diff line number Diff line change
Expand Up @@ -224,15 +224,21 @@ def test_duration_units_in_parameter_file():


def test_distribute_observation(tmp_path):
sim = lbs.Simulation(
base_path=tmp_path / "simulation_dir", start_time=1.0, duration_s=11.0
)
det = lbs.DetectorInfo("dummy", sampling_rate_hz=15)
obs_list = sim.create_observations(detectors=[det], num_of_obs_per_detector=5)
for dtype in (np.float16, np.float32, np.float64, np.float128):
sim = lbs.Simulation(
base_path=tmp_path / "simulation_dir", start_time=1.0, duration_s=11.0
)
det = lbs.DetectorInfo("dummy", sampling_rate_hz=15)
obs_list = sim.create_observations(
detectors=[det], num_of_obs_per_detector=5, dtype_tod=dtype
)

assert len(obs_list) == 5
assert int(obs_list[-1].get_times()[-1] - obs_list[0].get_times()[0]) == 10
assert sum([o.n_samples for o in obs_list]) == sim.duration_s * det.sampling_rate_hz
assert len(obs_list) == 5
assert int(obs_list[-1].get_times()[-1] - obs_list[0].get_times()[0]) == 10
assert (
sum([o.n_samples for o in obs_list])
== sim.duration_s * det.sampling_rate_hz
)


def test_distribute_observation_astropy(tmp_path):
Expand Down