# Cannot unify int64 and array(float64, 1d, C) for 'n.2'

**URL:** <https://numba.discourse.group/t/cannot-unify-int64-and-array-float64-1d-c-for-n-2/1850>\
**Category:** Support: What is this error message?\
**Created:** [March 19, 2023, 3:25pm UTC](https://numba.discourse.group/t/cannot-unify-int64-and-array-float64-1d-c-for-n-2/1850 "2023-03-19T15:25:42Z")\
**Posts on this page:** 5\
**Page:** 1

<div class="post-metadata">

**Author:** ![HenrYxZ](https://yyz2.discourse-cdn.com/free1/user_avatar/numba.discourse.group/henryxz/32/804_2.png) [@HenrYxZ](https://numba.discourse.group/u/HenrYxZ)\
**Post date:** [March 19, 2023, 3:25pm UTC](https://numba.discourse.group/t/cannot-unify-int64-and-array-float64-1d-c-for-n-2/1850/1 "2023-03-19T15:25:42Z")

</div>

I’m trying to make a simple Raytracer with Numba, but I have this error and can’t find the specific line that is causing it or how to solve it:

```auto
Traceback (most recent call last):
  File "C:\Users\Yoyo\projects\experiments\nrt.py", line 123, in <module>
    main()
  File "C:\Users\Yoyo\projects\experiments\nrt.py", line 115, in main
    render(arr, SX, SY, PROJ_DIST, CAM_POS, CAM_COORDS, RADIUS)
  File "C:\Users\Yoyo\projects\experiments\venv\lib\site-packages\numba\core\dispatcher.py", line 468, in _compile_for_args
    error_rewrite(e, 'typing')
  File "C:\Users\Yoyo\projects\experiments\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)
Cannot unify int64 and array(float64, 1d, C) for 'n.2', defined at C:\Users\Yoyo\projects\experiments\nrt.py (81)

File "nrt.py", line 81:
def render(arr, sx, sy, proj_dist, cam_pos, cam_coords, radius):
    <source elided>
            for n in range(AA_V_SAMPLES):
                for m in range(AA_H_SAMPLES):
                ^

During: typing of assignment at C:\Users\Yoyo\projects\experiments\nrt.py (81)

File "nrt.py", line 81:
def render(arr, sx, sy, proj_dist, cam_pos, cam_coords, radius):
    <source elided>
            for n in range(AA_V_SAMPLES):
                for m in range(AA_H_SAMPLES):
                ^

```

You can see my source code here: [experiments/nrt.py at main · HenrYxZ/experiments · GitHub](https://github.com/HenrYxZ/experiments/blob/main/nrt.py)

Or here:

```python
from math import sqrt
from numba import njit
import numpy as np
from PIL import Image

import utils

SCREEN_WIDTH = 280
SCREEN_HEIGHT = 210
COLOR_CHANNELS = 3
AA_H_SAMPLES = 1
AA_V_SAMPLES = 1
AA_SAMPLES = AA_V_SAMPLES * AA_H_SAMPLES

RADIUS = 0.5
SPHERE_POS = np.array((0.0, 0.0, 0.0))
SX = 2.0
SY = 1.5
PROJ_DIST = 2.0
CAM_POS = np.array((0.0, 0.0, -2.0))
CAM_COORDS = np.array((
    (1.0, 0.0, 0.0), # n0
    (0.0, 1.0, 0.0), # n1
    (0.0, 0.0, 1.0) # n2
))
MAT_COL = np.array((128 / 255.0, 0.0, 128 / 255.0))
C0 = MAT_COL * 0.2 # Lowest possible material color
BG_COLOR = np.zeros(COLOR_CHANNELS)
LIGHT_DIR = utils.normalize(np.array([-0.3, 1.0, -0.5]))

rng = np.random.default_rng()

@njit
def project_pixel(sx, sy, n0, n1, h, w, pixel_coords, p00):
    """
    Get the projection point for a given screen pixel

    Args:
        sx:
        sy:
        n0:
        n1:
        h:
        w:
        pixel_coords (tuple[float]):
        p00 (ndarray):

    Returns:
        ndarray: The projection point in 3D coordinates
    """
    xp = (pixel_coords[0] / w) * sx
    yp = ((h - pixel_coords[1]) / h) * sy
    pp = p00 + n0 * xp + n1 * yp
    return pp

@njit
def intersect(pp, nr, pos, radius) -> float:
    diff = pp - pos
    b = np.dot(nr, diff)
    c = np.dot(diff, diff) - radius ** 2
    discriminant = b ** 2 - c
    if b > 0 or discriminant < 0:
        return -1.0
    t = -b - sqrt(discriminant)
    return t

@njit
def render(arr, sx, sy, proj_dist, cam_pos, cam_coords, radius):
    h, w, _ = arr.shape
    n0, n1, n2 = cam_coords
    p00 = cam_pos + (n0 * (-sx / 2) + n1 * (-sy / 2) + n2 * proj_dist)
    for j in range(h):
        for i in range(w):
            color = np.zeros(COLOR_CHANNELS)
            for n in range(AA_V_SAMPLES):
                for m in range(AA_H_SAMPLES):
                    offsets = rng.random(2)
                    pixel_coords = (
                        i + (m + offsets[0]) / AA_H_SAMPLES,
                        j + (n + offsets[1]) / AA_V_SAMPLES
                    )
                    # Project pixel to point in 3D coordinates
                    pp = project_pixel(
                        sx, sy, n0, n1, h, w, pixel_coords, p00
                    )
                    # Shoot ray
                    nr = utils.normalize(pp - cam_pos)
                    t = intersect(cam_pos, nr, SPHERE_POS, radius)
                    # print(t)
                    if t > 0:
                        p_hit = cam_pos + nr * t
                        n = utils.normalize(p_hit - SPHERE_POS)
                        color += np.maximum(np.dot(n, LIGHT_DIR) * MAT_COL, C0)
                    else:
                        color += BG_COLOR
            color = color / AA_SAMPLES
            final_color = np.empty_like(color)
            np.round(color * 255, 0, final_color)
            arr[j, i, 0] = int(final_color[0])
            arr[j, i, 1] = int(final_color[1])
            arr[j, i, 2] = int(final_color[2])

def main():
    arr = np.zeros(
        (SCREEN_HEIGHT, SCREEN_WIDTH, COLOR_CHANNELS), dtype=np.uint8
    )
    timer = utils.Timer()
    with timer:
        render(arr, SX, SY, PROJ_DIST, CAM_POS, CAM_COORDS, RADIUS)
    img = Image.fromarray(arr)
    img_path = "docs/nrt_output.png"
    img.save(img_path)
    print(f"Finished rendering and saved image in {img_path}")

if __name__ == ' __main__':
    main()

```

I guess it could be the last line of the render function where I try to cast a float 3D array to a 3D uint8 array (RGB)

---

<div class="post-metadata">

**Author:** ![HenrYxZ](https://yyz2.discourse-cdn.com/free1/user_avatar/numba.discourse.group/henryxz/32/804_2.png) [@HenrYxZ](https://numba.discourse.group/u/HenrYxZ)\
**Post date:** [March 19, 2023, 3:27pm UTC](https://numba.discourse.group/t/cannot-unify-int64-and-array-float64-1d-c-for-n-2/1850/2 "2023-03-19T15:27:03Z")

</div>

(It works fine in regular Python)

---

<div class="post-metadata">

**Author:** ![nelson2005](https://yyz2.discourse-cdn.com/free1/user_avatar/numba.discourse.group/nelson2005/32/47_2.png) [@nelson2005](https://numba.discourse.group/u/nelson2005)\
**Post date:** [March 19, 2023, 3:49pm UTC](https://numba.discourse.group/t/cannot-unify-int64-and-array-float64-1d-c-for-n-2/1850/3 "2023-03-19T15:49:01Z")

</div>

does the type of any of the local variables change during the course of running this program?

In particular, is ‘n’ always assigned the same type?

---

<div class="post-metadata">

**Author:** ![Rutger](https://yyz2.discourse-cdn.com/free1/user_avatar/numba.discourse.group/rutger/32/446_2.png) [@Rutger](https://numba.discourse.group/u/Rutger)\
**Post date:** [March 20, 2023, 8:09am UTC](https://numba.discourse.group/t/cannot-unify-int64-and-array-float64-1d-c-for-n-2/1850/4 "2023-03-20T08:09:58Z")

</div>

Yes as @nelson2005 already suggests, it’s probably the reuse of `n` at line 97:  
`n = utils.normalize(p_hit - SPHERE_POS)`

---

<div class="post-metadata">

**Author:** ![HenrYxZ](https://yyz2.discourse-cdn.com/free1/user_avatar/numba.discourse.group/henryxz/32/804_2.png) [@HenrYxZ](https://numba.discourse.group/u/HenrYxZ)\
**Post date:** [March 21, 2023, 11:35pm UTC](https://numba.discourse.group/t/cannot-unify-int64-and-array-float64-1d-c-for-n-2/1850/5 "2023-03-21T23:35:17Z")

</div>

Thank you guys, that was the problem