The animation engine for agents: blazingly fast math and 3D videos with Manim's Python API, on a Rust GPU core.
Python
import math
import numpy as np
import manimgx as m
class TeacherScene(m.Scene):
def construct(self) -> None:
title = m.Tex("1D exponential vs 2D linear system").to_edge(m.UP, buff=0.3)
self.play(m.Write(title))
axes = m.Axes(
x_range=[0, 3, 1],
y_range=[0, 8, 2],
x_length=3.5,
y_length=2.5,
tips=False,
axis_config={"include_numbers": False, "stroke_width": 1.5},
).shift(m.LEFT * 3.5 + m.DOWN * 0.5)
curve = axes.plot(lambda x: math.exp(x), color=m.BLUE, x_range=[0, 3])
l_label = (
m.MathTex("\\dot x = rx", color=m.BLUE)
.scale(0.8)
.next_to(axes, m.UP, buff=0.2)
)
self.play(m.Create(axes), m.Write(l_label))
self.play(m.Create(curve), run_time=1.5)
plane = m.NumberPlane(
x_range=[-2.5, 2.5, 1],
y_range=[-2.0, 2.0, 1],
x_length=3.5,
y_length=2.5,
background_line_style={"stroke_opacity": 0.3, "stroke_color": m.GREY_B},
).shift(m.RIGHT * 3.0 + m.DOWN * 0.5)
r_label = (
m.MathTex("\\dot{\\mathbf v} = M\\,\\mathbf v", color=m.YELLOW)
.scale(0.8)
.next_to(plane, m.UP, buff=0.2)
)
self.play(m.Create(plane), m.Write(r_label))
traj_pts: list[np.ndarray] = [np.array([0.3, 0.0, 0.0])]
for _ in range(120):
p = traj_pts[-1]
v = np.array([0.3 * p[0] - 0.5 * p[1], 0.5 * p[0] + 0.3 * p[1], 0.0])
traj_pts.append(p + 0.04 * v)
spiral = m.VMobject(stroke_color=m.YELLOW, stroke_width=3.0)
spiral.set_points_as_corners(
[plane.coords_to_point(p[0], p[1]) for p in traj_pts]
)
self.play(m.Create(spiral), run_time=2.0)
self.wait(2.0)