Numba数组函数指针

Question

我最近发现numba确实支持函数指针：如何通过将setter函数传递给jitclass本身来更新带有其字符串名称的jitclass变量？。是否可以构造一个指向函数的这样的指针数组？

这是一个MCVE：

from numba import float64
from numba.experimental import jitclass

t = None

@jitclass(spec={'a': float64,
                'ptrs': t[:]})
class Test:
    def __init__(self):
        self.a = 0
        self.ptrs = np.array([self.x, self.y])
    def x(self):
        self.a += 1
    def y(self):
        self.a += 2
    def increment(self, n):
        self.ptrs[n](self)

t = Test()
print(t.a)     # Desired: 0
t.increment(0)
print(t.a)     # Desired: 1
t.increment(1)
print(t.a)     # Desired: 3

显然，对于t = None，这会引发一个错误，即使没有指示数组的[:]索引。

如果设置t = void，则在行t = Test()上出现错误

Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
  File "C:\Users\madphysicist\Anaconda3\lib\site-packages\numba\experimental\jitclass\base.py", line 122, in __call__
    return cls._ctor(*bind.args[1:], **bind.kwargs)
  File "C:\Users\madphysicist\Anaconda3\lib\site-packages\numba\core\dispatcher.py", line 439, in _compile_for_args
    raise e
  File "C:\Users\madphysicist\Anaconda3\lib\site-packages\numba\core\dispatcher.py", line 372, in _compile_for_args
    return_val = self.compile(tuple(argtypes))
  File "C:\Users\madphysicist\Anaconda3\lib\site-packages\numba\core\dispatcher.py", line 909, in compile
    cres = self._compiler.compile(args, return_type)
  File "C:\Users\madphysicist\Anaconda3\lib\site-packages\numba\core\dispatcher.py", line 79, in compile
    status, retval = self._compile_cached(args, return_type)
  File "C:\Users\madphysicist\Anaconda3\lib\site-packages\numba\core\dispatcher.py", line 93, in _compile_cached
    retval = self._compile_core(args, return_type)
  File "C:\Users\madphysicist\Anaconda3\lib\site-packages\numba\core\dispatcher.py", line 106, in _compile_core
    cres = compiler.compile_extra(self.targetdescr.typing_context,
  File "C:\Users\madphysicist\Anaconda3\lib\site-packages\numba\core\compiler.py", line 606, in compile_extra
    return pipeline.compile_extra(func)
  File "C:\Users\madphysicist\Anaconda3\lib\site-packages\numba\core\compiler.py", line 353, in compile_extra
    return self._compile_bytecode()
  File "C:\Users\madphysicist\Anaconda3\lib\site-packages\numba\core\compiler.py", line 415, in _compile_bytecode
    return self._compile_core()
  File "C:\Users\madphysicist\Anaconda3\lib\site-packages\numba\core\compiler.py", line 395, in _compile_core
    raise e
  File "C:\Users\madphysicist\Anaconda3\lib\site-packages\numba\core\compiler.py", line 386, in _compile_core
    pm.run(self.state)
  File "C:\Users\madphysicist\Anaconda3\lib\site-packages\numba\core\compiler_machinery.py", line 339, in run
    raise patched_exception
  File "C:\Users\madphysicist\Anaconda3\lib\site-packages\numba\core\compiler_machinery.py", line 330, in run
    self._runPass(idx, pass_inst, state)
  File "C:\Users\madphysicist\Anaconda3\lib\site-packages\numba\core\compiler_lock.py", line 35, in _acquire_compile_lock
    return func(*args, **kwargs)
  File "C:\Users\madphysicist\Anaconda3\lib\site-packages\numba\core\compiler_machinery.py", line 289, in _runPass
    mutated |= check(pss.run_pass, internal_state)
  File "C:\Users\madphysicist\Anaconda3\lib\site-packages\numba\core\compiler_machinery.py", line 262, in check
    mangled = func(compiler_state)
  File "C:\Users\madphysicist\Anaconda3\lib\site-packages\numba\core\typed_passes.py", line 463, in run_pass
    NativeLowering().run_pass(state)
  File "C:\Users\madphysicist\Anaconda3\lib\site-packages\numba\core\typed_passes.py", line 386, in run_pass
    lower.create_cpython_wrapper(flags.release_gil)
  File "C:\Users\madphysicist\Anaconda3\lib\site-packages\numba\core\lowering.py", line 242, in create_cpython_wrapper
    self.context.create_cpython_wrapper(self.library, self.fndesc,
  File "C:\Users\madphysicist\Anaconda3\lib\site-packages\numba\core\cpu.py", line 162, in create_cpython_wrapper
    builder.build()
  File "C:\Users\madphysicist\Anaconda3\lib\site-packages\numba\core\callwrapper.py", line 122, in build
    self.build_wrapper(api, builder, closure, args, kws)
  File "C:\Users\madphysicist\Anaconda3\lib\site-packages\numba\core\callwrapper.py", line 187, in build_wrapper
    obj = api.from_native_return(retty, retval, env_manager)
  File "C:\Users\madphysicist\Anaconda3\lib\site-packages\numba\core\pythonapi.py", line 1396, in from_native_return
    out = self.from_native_value(typ, val, env_manager)
  File "C:\Users\madphysicist\Anaconda3\lib\site-packages\numba\core\pythonapi.py", line 1410, in from_native_value
    return impl(typ, val, c)
  File "C:\Users\madphysicist\Anaconda3\lib\site-packages\numba\experimental\jitclass\boxing.py", line 139, in _box_class_instance
    box_subclassed = _specialize_box(typ)
  File "C:\Users\madphysicist\Anaconda3\lib\site-packages\numba\experimental\jitclass\boxing.py", line 122, in _specialize_box
    fast_fget = fget.compile((typ,))
  File "C:\Users\madphysicist\Anaconda3\lib\site-packages\numba\core\dispatcher.py", line 909, in compile
    cres = self._compiler.compile(args, return_type)
  File "C:\Users\madphysicist\Anaconda3\lib\site-packages\numba\core\dispatcher.py", line 79, in compile
    status, retval = self._compile_cached(args, return_type)
  File "C:\Users\madphysicist\Anaconda3\lib\site-packages\numba\core\dispatcher.py", line 93, in _compile_cached
    retval = self._compile_core(args, return_type)
  File "C:\Users\madphysicist\Anaconda3\lib\site-packages\numba\core\dispatcher.py", line 106, in _compile_core
    cres = compiler.compile_extra(self.targetdescr.typing_context,
  File "C:\Users\madphysicist\Anaconda3\lib\site-packages\numba\core\compiler.py", line 606, in compile_extra
    return pipeline.compile_extra(func)
  File "C:\Users\madphysicist\Anaconda3\lib\site-packages\numba\core\compiler.py", line 353, in compile_extra
    return self._compile_bytecode()
  File "C:\Users\madphysicist\Anaconda3\lib\site-packages\numba\core\compiler.py", line 415, in _compile_bytecode
    return self._compile_core()
  File "C:\Users\madphysicist\Anaconda3\lib\site-packages\numba\core\compiler.py", line 395, in _compile_core
    raise e
  File "C:\Users\madphysicist\Anaconda3\lib\site-packages\numba\core\compiler.py", line 386, in _compile_core
    pm.run(self.state)
  File "C:\Users\madphysicist\Anaconda3\lib\site-packages\numba\core\compiler_machinery.py", line 339, in run
    raise patched_exception
  File "C:\Users\madphysicist\Anaconda3\lib\site-packages\numba\core\compiler_machinery.py", line 330, in run
    self._runPass(idx, pass_inst, state)
  File "C:\Users\madphysicist\Anaconda3\lib\site-packages\numba\core\compiler_lock.py", line 35, in _acquire_compile_lock
    return func(*args, **kwargs)
  File "C:\Users\madphysicist\Anaconda3\lib\site-packages\numba\core\compiler_machinery.py", line 289, in _runPass
    mutated |= check(pss.run_pass, internal_state)
  File "C:\Users\madphysicist\Anaconda3\lib\site-packages\numba\core\compiler_machinery.py", line 262, in check
    mangled = func(compiler_state)
  File "C:\Users\madphysicist\Anaconda3\lib\site-packages\numba\core\typed_passes.py", line 463, in run_pass
    NativeLowering().run_pass(state)
  File "C:\Users\madphysicist\Anaconda3\lib\site-packages\numba\core\typed_passes.py", line 386, in run_pass
    lower.create_cpython_wrapper(flags.release_gil)
  File "C:\Users\madphysicist\Anaconda3\lib\site-packages\numba\core\lowering.py", line 242, in create_cpython_wrapper
    self.context.create_cpython_wrapper(self.library, self.fndesc,
  File "C:\Users\madphysicist\Anaconda3\lib\site-packages\numba\core\cpu.py", line 162, in create_cpython_wrapper
    builder.build()
  File "C:\Users\madphysicist\Anaconda3\lib\site-packages\numba\core\callwrapper.py", line 122, in build
    self.build_wrapper(api, builder, closure, args, kws)
  File "C:\Users\madphysicist\Anaconda3\lib\site-packages\numba\core\callwrapper.py", line 187, in build_wrapper
    obj = api.from_native_return(retty, retval, env_manager)
  File "C:\Users\madphysicist\Anaconda3\lib\site-packages\numba\core\pythonapi.py", line 1396, in from_native_return
    out = self.from_native_value(typ, val, env_manager)
  File "C:\Users\madphysicist\Anaconda3\lib\site-packages\numba\core\pythonapi.py", line 1410, in from_native_value
    return impl(typ, val, c)
  File "C:\Users\madphysicist\Anaconda3\lib\site-packages\numba\core\boxing.py", line 397, in box_array
    np_dtype = numpy_support.as_dtype(typ.dtype)
  File "C:\Users\madphysicist\Anaconda3\lib\site-packages\numba\np\numpy_support.py", line 156, in as_dtype
    raise NotImplementedError("%r cannot be represented as a Numpy dtype"
NotImplementedError: Failed in nopython mode pipeline (step: nopython mode backend)
Failed in nopython mode pipeline (step: nopython mode backend)
none cannot be represented as a Numpy dtype

然后，我尝试将t设置为指向x和y的函数指针数组。

from numba import float64, intp, void, deferred_type
tt = deferred_type()
t = void(tt, intp)

来自'ptrs': t[:]})的错误是：

Traceback (most recent call last):
  File "<stdin>", line 7, in <module>
TypeError: 'Signature' object is not subscriptable

最后，我尝试了t = void。这给了t = Test()行编译过程中的错误。

Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
  File "C:\Users\madphysicist\Anaconda3\lib\site-packages\numba\experimental\jitclass\base.py", line 122, in __call__
    return cls._ctor(*bind.args[1:], **bind.kwargs)
  File "C:\Users\madphysicist\Anaconda3\lib\site-packages\numba\core\dispatcher.py", line 420, in _compile_for_args
    error_rewrite(e, 'typing')
  File "C:\Users\madphysicist\Anaconda3\lib\site-packages\numba\core\dispatcher.py", line 361, in error_rewrite
    raise e.with_traceback(None)
numba.core.errors.TypingError: Failed in nopython mode pipeline (step: nopython frontend)
Failed in nopython mode pipeline (step: nopython frontend)
NameError: name 'np' is not defined
During: resolving callee type: jitclass.Test#1bbda3d32e0<a:float64,ptrs:array(none, 1d, A)>
During: typing of call at <string> (3)

During: resolving callee type: jitclass.Test#1bbda3d32e0<a:float64,ptrs:array(none, 1d, A)>
During: typing of call at <string> (3)


File "<string>", line 3:
<source missing, REPL/exec in use?>

那么，我如何创建一个函数指针数组呢？

Answer 1

还不支持将Numba函数存储到Numpy数组中，而是。至少在0.54.1版本(和以前的版本)中，没有处于nopython模式。

一种实验检查方法是简单地编辑文档代码并将结果存储在数组中，然后让Numba推断相应的类型。Numba失败的原因是不支持数组的dtype。

另一种检查方法是读取存储库中的support.py代码。此文件包含当前支持的类型。我列出了以下内容：

Basic types:
    - bool
    - int8
    - int16
    - int32
    - int64
    - uint8
    - uint16
    - uint32
    - uint64
    - float32
    - float64
    - complex64
    - complex128

Advanced types:
    - Datetime
    - Timedelta
    - CharSeq (mark as 'S' in Numpy)
    - UnicodeCharSeq (mark as 'U' in Numpy)
    - Numpy structured types (containing other ones like Numpy does)

Generic type:
    - object (unsupported in nopython mode since the type inference cannot work with it)

Other (unknown) types:
    - EnumMember
    - NumberClass
    - NestedArray