Numba compilation fails with nested jitclasses

Question

The code below fails, and I've exhausted all of my guesses as to why.

import numba

import numpy as np
from numba import int32, float64, types, deferred_type
from numba.experimental import jitclass

import constants


p_spec = [
    ('x', float64),
    ('y', float64),
]


@jitclass(p_spec)
class Particle(object):
    def __init__(self, x, y):
        self.x = x
        self.y = y


particle_type = deferred_type()
particle_type.define(Particle.class_type.instance_type)

ps_spec = [
    ('size', int32),
    ('particles', types.ListType(particle_type)),
]


@jitclass(ps_spec)
class Particles(object):
    def __init__(self, size):
        self.size = size
        self.particles = numba.typed.List([Particle(v[0], v[1]) for v in zip(np.random.uniform(0, constants.WIDTH, size),
                                                                             np.random.uniform(0, constants.HEIGHT, size))])


p = Particles(20)

The error is :

Traceback (most recent call last):
  File "C:\Users\User\PycharmProjects\Emergence\Particles2.py", line 40, in <module>
    p = Particles(20)
  File "C:\Users\User\PycharmProjects\Emergence\venv\lib\site-packages\numba\experimental\jitclass\base.py", line 124, in __call__
    return cls._ctor(*bind.args[1:], **bind.kwargs)
  File "C:\Users\User\PycharmProjects\Emergence\venv\lib\site-packages\numba\core\dispatcher.py", line 468, in _compile_for_args
    error_rewrite(e, 'typing')
  File "C:\Users\User\PycharmProjects\Emergence\venv\lib\site-packages\numba\core\dispatcher.py", line 409, 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: native lowering)
Cannot cast ListType[instance.jitclass.Particle#1f2c6ecb850<x:float64,y:float64>] to DeferredType#2142231399584: %"inserted.data.4" = insertvalue {i8*, i8*} %"inserted.meminfo.4", i8* %".714", 1
During: lowering "(self).particles = $60call_method.28" at C:\Users\User\PycharmProjects\Emergence\Particles2.py (36)
During: resolving callee type: jitclass.Particles#1f2c6f5efa0<size:int32,particles:DeferredType#2142231399584>
During: typing of call at <string> (3)

During: resolving callee type: jitclass.Particles#1f2c6f5efa0<size:int32,particles:DeferredType#2142231399584>
During: typing of call at <string> (3)


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

I'm new to jiclass, but isn't this complaining about the `list`, not so much the nested classes? I think `self.particles = numba.typed.List([Particle....])` is what you need. — Mikael Öhman, Oct 08 '22 at 17:41
Thanks @MikaelÖhman - I tried that but it did not seem to work. I even used a deferred type. — Amir Afghani, Oct 08 '22 at 17:44
In your updated code, you forgot to add the `ListType` to you deferred ps_spec. t's a `types.ListType(particle_type)`. — Mikael Öhman, Oct 08 '22 at 17:49

score 0 · Answer 1 · answered Oct 08 '22 at 18:05

0

This seems to work but I'm not sure why or if this is the 'best' answer.

ps_spec = [
    ('size', int32),
    ('particles', types.ListType(Particle.class_type.instance_type)),
]

answered Oct 08 '22 at 18:05

Amir Afghani

37,814
16
84
124

Numba compilation fails with nested jitclasses

1 Answers1