Linfeng - First PR of c++ thermo using pybind11#3856
Linfeng - First PR of c++ thermo using pybind11#3856gitlinffff wants to merge 41 commits intoUnidata:mainfrom
Conversation
* vectorize the function using pybind11/numpy.h, achieving handling numpy array, float and dtype * add dewpoint, add constants.hpp, adding other functions WIP
…verting saturation vapor pressure WIP
…midity_from_mixing_ratio, specific_humidity_from_dewpoint
…multiple dimensions
…elProfileHelper wip
|
Linfeng Li seems not to be a GitHub user. You need a GitHub account to be able to sign the CLA. If you have already a GitHub account, please add the email address used for this commit to your account. You have signed the CLA already but the status is still pending? Let us recheck it. |
| requires = ["setuptools>=61", "setuptools-scm[toml]>=3.4", "scikit-build-core", "pybind11"] | ||
| build-backend = "scikit_build_core.build" |
There was a problem hiding this comment.
If you're using that build backend, you can drop setuptools from the build-requires list too.
| #set(CMAKE_CXX_STANDARD 14) | ||
| #set(CMAKE_CXX_STANDARD_REQUIRED ON) | ||
|
|
||
| find_package(Python COMPONENTS Interpreter Development.Module REQUIRED) |
There was a problem hiding this comment.
Is there a specific reason you need the interpreter? It looks like the module development component would be enough so far.
src/thermo.hpp
Outdated
| #include <pybind11/pybind11.h> | ||
| #include <pybind11/numpy.h> |
There was a problem hiding this comment.
These need to be the first includes in the file.
| #include <pybind11/pybind11.h> | ||
| #include <pybind11/numpy.h> | ||
| #include <pybind11/stl.h> |
There was a problem hiding this comment.
These need to be the first includes in the file.
src/thermo.cpp
Outdated
| #include <pybind11/pybind11.h> | ||
| #include <pybind11/numpy.h> |
There was a problem hiding this comment.
These need to be the first includes in the file
| #include <iostream> // for std::cerr | ||
| #include "math.hpp" | ||
|
|
||
| double lambert_wm1(double x, double tol, int max_iter) { |
There was a problem hiding this comment.
Would it be possible to use the SciPy LambertW (or the xsf function it wraps (C++ in a separate repo for people who don't want all of SciPy), or the Cython wrapper it might generate (it looks like it would be scipy.special._ufuncs.lambertw)? If we are very lucky, PyCapsule_Import() would get the correct function with little effort.
src/math.cpp
Outdated
|
|
||
| for (int i = 0; i < max_iter; ++i) { | ||
| double ew = std::exp(w); | ||
| double wew = w * ew; |
There was a problem hiding this comment.
It looks like spellcheck would prefer a name like w_ew
| "Calculate relative humidity from temperature and dewpoint.", | ||
| py::arg("temperature"), py::arg("dewpoint"), py::arg("phase")); | ||
|
|
||
| m.def("dry_lapse", &DryLapseVectorized, |
There was a problem hiding this comment.
Is there a specific reason this doesn't follow the pattern of py::vectorize(DryLapse)? That seems to be what the current implementation does.
…'structured binding'
…ler and an array with one element
4 participants
c++ functions I created at this stage:
MoistAirGasConstant
MoistAirSpecificHeatPressure
WaterLatentHeatVaporization
WaterLatentHeatSublimation
RelativeHumidityFromDewPoint
DryLapse
DryLapseProfile
CaldlnTdlnP
MoistLapse
MoistLapseProfile
LCL
_CheckPressure
_SaturationVaporPressureLiquid
_SaturationVaporPressureSolid
SaturationVaporPressure
DewPoint
MixingRatio
SaturationMixingRatio
SpecificHumidityFromMixingRatio
SpecificHumidityFromDewPoint
VirtualTemperature
lambert_wm1