diff --git a/chapter_appendix-mathematics-for-deep-learning/multivariable-calculus.md b/chapter_appendix-mathematics-for-deep-learning/multivariable-calculus.md
index 1cfbefafa6..2131b7a4a1 100644
--- a/chapter_appendix-mathematics-for-deep-learning/multivariable-calculus.md
+++ b/chapter_appendix-mathematics-for-deep-learning/multivariable-calculus.md
@@ -15,7 +15,7 @@ Now, let's take this and change $w_2$ a little bit to $w_2 + \epsilon_2$:
 
 $$
 \begin{aligned}
-L(w_1+\epsilon_1, w_2+\epsilon_2, \ldots, w_N) & \approx L(w_1, w_2+\epsilon_2, \ldots, w_N) + \epsilon_1 \frac{\partial}{\partial w_1} L(w_1, w_2+\epsilon_2, \ldots, w_N+\epsilon_N) \\
+L(w_1+\epsilon_1, w_2+\epsilon_2, \ldots, w_N) & \approx L(w_1, w_2+\epsilon_2, \ldots, w_N) + \epsilon_1 \frac{\partial}{\partial w_1} L(w_1, w_2+\epsilon_2, \ldots, w_N) \\
 & \approx L(w_1, w_2, \ldots, w_N) \\
 & \quad + \epsilon_2\frac{\partial}{\partial w_2} L(w_1, w_2, \ldots, w_N) \\
 & \quad + \epsilon_1 \frac{\partial}{\partial w_1} L(w_1, w_2, \ldots, w_N) \\