convolutions

首先定义了三个坐标轴：

axes1, axes2, axes3 = all_axes = VGroup(*(
    Axes((-3, 3), (-1, 1), height=2.0, width=FRAME_WIDTH * 0.5 - 1)
    for x in range(3)
))

我发现 3b1b 特别喜欢这样定义一组 mobject:

seq1_tex, seq2_tex, sum_seq, prod_seq, conv_seq = seqs = VGroup(
    OldTex(f"a = {list(seq1)}", **kw),
    OldTex(f"b = {list(seq2)}", **kw),
    OldTex(f"a + b = {list(seq1 + seq2)}", **kw),
    OldTex(Rf"a \cdot b = {list(seq1 * seq2)}", **kw),
    OldTex(Rf"a * b = {list(np.convolve(seq1, seq2))}", **kw),
)

left_op_labels = VGroup(*map(Text, ["Addition", "Multiplication", "Convolution"]))

all_boxes = VGroup(*(SurroundingRectangle(m[2:]) for m in seqs[:3]))

以下是多组数据的简便方法：

for left_op_label, seq, graph_label in zip(left_op_labels, seqs[2:], graph_labels[2:]):

        self.play(
            *map(Write, box_arrows),
            FadeIn(boxes),
            *(
                Transform(
                    box.copy().set_stroke(width=0, opacity=0),
                    mystery_box.copy(),
                    path_arc=-PI / 3,
                    remover=True
                )
                for box in boxes
            )
        )

 

随后使用滞后出现的方法添加了图像