Init_pos(): RuntimeError: [lang_util.cpp:to_primitive_type@395] Assertion failure: primitive

[JerryYan97]

Describe the bug
Previously, I can run these code, but it doesn't work and get stuck at the init_pos() function. It looks like it is caused by this statement:

ti.Vector([i/N*0.4, j/W*0.2])

# sample code here
import taichi as ti
import numpy as np
import pymesh

# Used to create an obj mesh for pymesh, which is exactly same as the one used in PDJ demo3.
# This mesh will be read into PN.

N = 20  # internal of one edge
W = 10
NF = 2 * N * W  # 2 * N ** 2   # number of faces
NV = (N+1)*(W+1)  # (N + 1) ** 2 # number of vertices
dx = 1 / N  # 0.05

pos = ti.Vector.field(2, float, NV)
f2v = ti.Vector.field(3, int, NF)  # ids of three vertices of each face
# dirichlet_num = N + 1
# dirichlet = ti.Vector.field(2, float, W + 1)
# [i for i in range(11)]


@ti.kernel
def init_pos():
    # i: col; j: row;
    for i, j in ti.ndrange(N + 1, W + 1):
        k = i*(W+1)+j
        pos[k] = ti.Vector([i/N*0.4, j/W*0.2]) + ti.Vector([0.2, 0.4]) # 0.2, 0.4 - 0.6,0.6  0.02*0.02


@ti.kernel
def init_mesh():  # generate two triangles
    for i, j in ti.ndrange(N, W):
        k = (i * W + j) * 2  # tirangle index   w 2 n 3
        a = i * (W + 1) + j  # 0 0 = 0
        b = a + 1  # 1
        c = a + W + 2  # 12
        d = a + W + 1  # 11
        f2v[k + 0] = [a, d, c]
        f2v[k + 1] = [a, c, b]


init_pos()
init_mesh()

np_pos = pos.to_numpy()
np_f2v = f2v.to_numpy()

print(np.shape(np_pos))
print(np.shape(np_f2v))

f = open("demo3_mesh.obj", "w")

# Write all pos
for [x, y] in np_pos:
    f.write("v " + str(x) + " " + str(y) + " 0\n")

# Write all faces
for [i, j, k] in np_f2v:
    f.write("f " + str(i + 1) + "/" + str(i + 1) + " " + str(j + 1) + "/" + str(j + 1) + " " + str(k + 1) + "/" + str(k + 1) + "\n")

f.close()

/usr/bin/python3.8 /home/jiaruiyan/self_study/Taichi_PJD_3D/mesh_create.py
[Taichi] mode=release
[Taichi] preparing sandbox at /tmp/taichi-1envx3t5
[Taichi] version 0.6.39, llvm 10.0.0, commit 12e02bec, linux, python 3.8.5
[Taichi] materializing...
[E 10/11/20 10:27:05.815] [lang_util.cpp:to_primitive_type@395] Assertion failure: primitive


***********************************
* Taichi Compiler Stack Traceback *
***********************************
/home/jiaruiyan/.local/lib/python3.8/site-packages/taichi/core/../lib/taichi_core.so: taichi::Logger::error(std::__cxx11::basic_string<char, std::char_traits<char>, std::allocator<char> > const&, bool)
/home/jiaruiyan/.local/lib/python3.8/site-packages/taichi/core/../lib/taichi_core.so(+0x702f26) [0x7f66b8e03f26]
/home/jiaruiyan/.local/lib/python3.8/site-packages/taichi/core/../lib/taichi_core.so: taichi::lang::promoted_type(taichi::lang::DataType, taichi::lang::DataType)
/home/jiaruiyan/.local/lib/python3.8/site-packages/taichi/core/../lib/taichi_core.so: taichi::lang::TypeCheck::visit(taichi::lang::BinaryOpStmt*)
/home/jiaruiyan/.local/lib/python3.8/site-packages/taichi/core/../lib/taichi_core.so: taichi::lang::TypeCheck::visit(taichi::lang::Block*)
/home/jiaruiyan/.local/lib/python3.8/site-packages/taichi/core/../lib/taichi_core.so: taichi::lang::TypeCheck::visit(taichi::lang::Block*)
/home/jiaruiyan/.local/lib/python3.8/site-packages/taichi/core/../lib/taichi_core.so: taichi::lang::irpass::type_check(taichi::lang::IRNode*)
/home/jiaruiyan/.local/lib/python3.8/site-packages/taichi/core/../lib/taichi_core.so: taichi::lang::irpass::compile_to_offloads(taichi::lang::IRNode*, taichi::lang::CompileConfig const&, bool, bool, bool, bool)
/home/jiaruiyan/.local/lib/python3.8/site-packages/taichi/core/../lib/taichi_core.so: taichi::lang::irpass::compile_to_executable(taichi::lang::IRNode*, taichi::lang::CompileConfig const&, bool, bool, bool, bool, bool, bool, bool)
/home/jiaruiyan/.local/lib/python3.8/site-packages/taichi/core/../lib/taichi_core.so: taichi::lang::Kernel::lower(bool)
/home/jiaruiyan/.local/lib/python3.8/site-packages/taichi/core/../lib/taichi_core.so: taichi::lang::Program::compile(taichi::lang::Kernel&)
/home/jiaruiyan/.local/lib/python3.8/site-packages/taichi/core/../lib/taichi_core.so: taichi::lang::Kernel::compile()
/home/jiaruiyan/.local/lib/python3.8/site-packages/taichi/core/../lib/taichi_core.so: taichi::lang::Kernel::operator()(taichi::lang::Kernel::LaunchContextBuilder&)
/home/jiaruiyan/.local/lib/python3.8/site-packages/taichi/core/../lib/taichi_core.so(+0x4c6164) [0x7f66b8bc7164]
/home/jiaruiyan/.local/lib/python3.8/site-packages/taichi/core/../lib/taichi_core.so(+0x3d0d30) [0x7f66b8ad1d30]
/usr/bin/python3.8: PyCFunction_Call
/usr/bin/python3.8: _PyObject_MakeTpCall
/usr/bin/python3.8: ) [0x50a978]
/usr/bin/python3.8: PyObject_Call
/usr/bin/python3.8: ) [0x59d47c]
/usr/bin/python3.8: _PyObject_MakeTpCall
/usr/bin/python3.8: _PyEval_EvalFrameDefault
/usr/bin/python3.8: _PyEval_EvalCodeWithName
/usr/bin/python3.8: _PyFunction_Vectorcall
/usr/bin/python3.8: PyObject_Call
/usr/bin/python3.8: _PyEval_EvalFrameDefault
/usr/bin/python3.8: _PyEval_EvalCodeWithName
/usr/bin/python3.8: _PyFunction_Vectorcall
/usr/bin/python3.8: ) [0x59d17f]
/usr/bin/python3.8: PyObject_Call
/usr/bin/python3.8: _PyEval_EvalFrameDefault
/usr/bin/python3.8: _PyEval_EvalCodeWithName
/usr/bin/python3.8: _PyFunction_Vectorcall
/usr/bin/python3.8: _PyEval_EvalFrameDefault
/usr/bin/python3.8: _PyEval_EvalCodeWithName
/usr/bin/python3.8: PyEval_EvalCode
/usr/bin/python3.8: ) [0x67bc91]
/usr/bin/python3.8: ) [0x67bd0f]
/usr/bin/python3.8: PyRun_FileExFlags
/usr/bin/python3.8: PyRun_SimpleFileExFlags
/usr/bin/python3.8: Py_RunMain
/usr/bin/python3.8: Py_BytesMain
/lib/x86_64-linux-gnu/libc.so.6: __libc_start_main
/usr/bin/python3.8: _start

Internal Error occurred, check this page for possible solutions:
https://taichi.readthedocs.io/en/stable/install.html#troubleshooting
Traceback (most recent call last):
  File "/home/jiaruiyan/self_study/Taichi_PJD_3D/mesh_create.py", line 40, in <module>
    init_pos()
  File "/home/jiaruiyan/.local/lib/python3.8/site-packages/taichi/lang/kernel.py", line 574, in wrapped
    return primal(*args, **kwargs)
  File "/home/jiaruiyan/.local/lib/python3.8/site-packages/taichi/lang/kernel.py", line 502, in __call__
    return self.compiled_functions[key](*args)
  File "/home/jiaruiyan/.local/lib/python3.8/site-packages/taichi/lang/kernel.py", line 460, in func__
    t_kernel(launch_ctx)
RuntimeError: [lang_util.cpp:to_primitive_type@395] Assertion failure: primitive

Process finished with exit code 1

[JerryYan97]

It works when I change the init_pos() to this:

@ti.kernel
def init_pos():
    # i: col; j: row;
    for i, j in ti.ndrange(N + 1, W + 1):
        k = i*(W+1)+j
        pos[k] = ti.Vector([i * (1.0 / N * 0.4), j * (1.0 / W * 0.2)]) + ti.Vector([0.2, 0.4]) # 0.2, 0.4 - 0.6,0.6  0.02*0.02

It looks like variables in ti.kernel cannot divide other staffs.

[archibate]

Thank for reporting! This is actually a known bug introduced by #1906.
Before that, we may add ti.init() to the first line of program to fix the problem.

Also btw, does ti.Vector([float(i)/float(N)*0.4, float(j)/float(W)*0.2]) depress the error? It seems the type promotion for division is broken.

[JerryYan97]

Yes, it can depress the error.