1

Taylor Series in 2 Variables

Recall from first year calculus that if 𝑓(𝑥) has an infinite number of derivatives
near a point 𝑥 = 𝑎, then we have:

′ (𝑎)(𝑥 𝑓′′(𝑎) 𝑓𝑛(𝑎)

𝑓(𝑥) = 𝑓(𝑎) + 𝑓 − 𝑎) + (𝑥 − 𝑎)2 + ⋯+ (𝑥 − 𝑎)𝑛 + 𝑅𝑛 (𝑥, 𝑎)
2! 𝑛!

𝑓 (𝑛+1) (𝑐)
where 𝑅𝑛 (𝑥, 𝑎) =
(𝑛+1)!
(𝑥 − 𝑎)𝑛+1 and 𝑐 is between 𝑥 and 𝑎.

This allows us to approximate the value of a function, 𝑓(𝑥), using an 𝑛-th

degree polynomial if we know the value of the function and its derivatives at
𝑥 = 𝑎. In addition, 𝑅𝑛 (𝑥, 𝑎) = error in the approximation, allows us to put an upper
bound on the error of this approximation.

𝑦 = 𝑓(𝑥)
𝑦 = 𝑇2 (𝑥)

𝑦 = 𝑇1 (𝑥)

where:
𝑇1 (𝑥 ) = 𝑓 (𝑎) + 𝑓′(𝑎)(𝑥 − 𝑎)
𝑓 ′′ (𝑎)
𝑇2 (𝑥 ) = 𝑓(𝑎) + 𝑓 ′ (𝑎)(𝑥 − 𝑎) + (𝑥 − 𝑎)2 .
2!
 2

Ex. Approximate the value of 𝑒 .02 using a second order Taylor polynomial around 𝑎 = 0.

𝑓 (𝑥 ) = 𝑒 𝑥 𝑓(0) = 1
𝑓′(𝑥 ) = 𝑒 𝑥 𝑓′(0) = 1
𝑓′′(𝑥 ) = 𝑒 𝑥 𝑓′′(0) = 1
𝑓′′′(𝑥 ) = 𝑒 𝑥 𝑓′′′(0) = 1

𝑥 ′( 𝑓 ′′ (0)
𝑒 = 𝑓(0) + 𝑓 0)(𝑥 ) + (𝑥 2 ) + 𝑅2 (𝑥, 0)
2!

𝑓 ′′′ (𝑐) 𝑒𝑐
where 𝑅2 (𝑥, 0) = (𝑥 − 0)3 = 𝑥 3 and 0 < 𝑐 < 𝑥.
3! 3!

𝑥 𝑥2 𝑒𝑐
𝑒 =1+𝑥+ + 3! 𝑥 3
2!

So now we can plug in 𝑥 = .02

.02 (.02)2 .0004

𝑒 ≈ 1 + .02 + = 1 + .02 + = 1.0202
2! 2

The error is no bigger than:

𝑒𝑐 3
|𝑅2 (.02, 0)| = (. 02)3 ≤ (. 000008) = .000004.
3! 3!
 3

Now we want to be able to approximate the value of 𝑓(𝑥, 𝑦) using polynomials in 𝑥

and 𝑦.

Second Order Taylor Formula:

Let 𝑓: 𝑈 ⊆ ℝ2 → ℝ have continuous partial derivatives of third order and let
(𝑥0 , 𝑦0 ) ∈ 𝑈. Then we have:

𝑓 (𝑥, 𝑦) = 𝑓 (𝑥0 , 𝑦0 ) + (𝑓𝑥 (𝑥0 , 𝑦0 ))(𝑥 − 𝑥0 ) + (𝑓𝑦 (𝑥0 , 𝑦0 )) (𝑦 − 𝑦0 )

1
+ [𝑓𝑥𝑥 (𝑥0 , 𝑦0 )(𝑥 − 𝑥0 )2 + 2𝑓𝑥𝑦 (𝑥0 , 𝑦0 )(𝑥 − 𝑥0 )(𝑦 − 𝑦0 ) +
2
𝑓𝑦𝑦 (𝑥0 , 𝑦0 )(𝑦 − 𝑦0 )2 ] + 𝑅2 (𝑥, 𝑦, 𝑥0 , 𝑦0 )

where:
2
1 𝜕3𝑓
𝑅2 (𝑥, 𝑦, 𝑥0 , 𝑦0 ) = ∑ ( (𝑐 ))(Δ𝑥𝑖 )(Δ𝑥𝑗 )(Δ𝑥𝑘 )
3 𝜕𝑥𝑖 𝜕𝑥𝑗 𝜕𝑥𝑘 𝑖𝑗𝑘
𝑖,𝑗,𝑘=1
𝑥1 = 𝑥
𝑥2 = 𝑦
Δ𝑥𝑖 = 𝑥 − 𝑥0 if 𝑖 = 1
= 𝑦 − 𝑦0 if 𝑖 = 2

and 𝑐𝑖𝑗𝑘 lies somewhere on the line segment between (𝑥, 𝑦) and (𝑥0 , 𝑦0 ).
 4

Ex. Compute the second order Taylor formula for 𝑓 (𝑥, 𝑦) = 𝑒 𝑥 cos 𝑦 about the
point (𝑥0 , 𝑦0 ) = (0,0) and approximate the value of 𝑒 .02 cos(.04).

𝑓(𝑥, 𝑦) = 𝑒 𝑥 cos 𝑦 𝑓(0,0) = 1

𝑓𝑥 = 𝑒 𝑥 cos 𝑦 𝑓𝑥 (0,0) = 1
𝑓𝑦 = −𝑒 𝑥 sin 𝑦 𝑓𝑦 (0,0) = 0

𝑓𝑥𝑥 = 𝑒 𝑥 cos 𝑦 𝑓𝑥𝑥 (0,0) = 1

𝑓𝑥𝑦 = −𝑒 𝑥 sin 𝑦 𝑓𝑥𝑦 (0,0) = 0
𝑓𝑦𝑦 = −𝑒 𝑥 cos 𝑦 𝑓𝑦𝑦 (0,0) = −1

𝑒 𝑥 cos 𝑦 = 1 + 1(𝑥 − 0) + 0(𝑦 − 0)

1
+ [1(𝑥 − 0)2 + 2(0)(𝑥 − 0)(𝑦 − 0) − 1(𝑦 − 0)2 ] +𝑅2 (𝑥, 𝑦, 0, 0)
2
1 1
= 1 + 𝑥 + 𝑥 2 − 𝑦 2 + 𝑅2 (𝑥, 𝑦, 0,0).
2 2

Thus we have:
1 1
𝑒 𝑥 cos 𝑦 ≈ 1 + 𝑥 + 𝑥 2 − 𝑦 2
2 2

1 1
𝑒 .02 cos(.04) ≈ 1 + .02 + (. 02)2 − (. 04)2
2 2
= 1 + .02 + .0002 − .0008 = 1.0194.
 5

Ex. Determine the second order Taylor formula for 𝑓 (𝑥, 𝑦) = 𝑥 cos 𝑦 about the
𝜋
point (1, ) and approximate:
4
𝜋 𝜋
𝑓 (1.1, + .2) = (1.1) cos( + .2).
4 4

𝜋 √2
𝑓(𝑥, 𝑦) = 𝑥 cos 𝑦 𝑓 (1, ) =
4 2
𝜋 √2
𝑓𝑥 = cos 𝑦 𝑓𝑥 (1, ) =
4 2
𝜋 √2
𝑓𝑦 = −𝑥 sin 𝑦 𝑓𝑦 (1, ) = −
4 2

𝜋
𝑓𝑥𝑥 = 0 𝑓𝑥𝑥 (1, ) = 0
4
𝜋 √2
𝑓𝑥𝑦 = − sin 𝑦 𝑓𝑥𝑦 (1, ) = −
4 2
𝜋 √2
𝑓𝑦𝑦 = −𝑥 cos 𝑦 𝑓𝑦𝑦 (1, ) = −
4 2

√2 √2 √2 𝜋
𝑓 (𝑥, 𝑦) = + (𝑥 − 1) − (𝑦 − )
2 2 2 4
1 √2 𝜋 √2 𝜋 2
+ [0(𝑥 − 1)2 + 2 (− ) (𝑥 − 1) (𝑦 − ) + (− ) (𝑦 − ) ]
2 2 4 2 4
𝜋
+𝑅2 (𝑥, 𝑦, 1, )
4
√2 √2 √2 𝜋
= + (𝑥 − 1) − (𝑦 − )
2 2 2 4
1 𝜋 √2 𝜋 2 𝜋
+ [−√2(𝑥 − 1) (𝑦 − ) + (− ) (𝑦 − ) ] + 𝑅2 (𝑥, 𝑦, 1, ).
2 4 2 4 4

𝜋
(1.1) cos ( + .2)
4
√2 √2 √2 1 √2
≈ + ( . 1) − (. 2) + [−√2(. 1)(. 2) + (− ) (. 2)2 ]
2 2 2 2 2
√2 √2
= (. 9) − (. 04) ( ) ≈ .6081.
2 2