Linear Approximation Cheatsheet

1. Definition

When xx is close to some reference value x0x_0 (in other words Δx0\Delta x \approx 0), you can approximate f(x)f(x) as the tangent line at that reference value.

Δxx₀x
f(x0)f(x_0)
1.349859
f~(x)\tilde f(x)on the tangent line
2.159774
f(x)f(x)on the curve
2.459603
error (f(x)f~(x)\left| f(x) - \tilde f(x) \right|)
0.299829
formformulaanalog
point–slope formf(x)f(x0)+f(x0)(xx0)f(x) \approx f(x_0) + f'(x_0)(x - x_0)yy0=m(xx0)y - y_0 = m(x - x_0)
slope–intercept formf(x+Δx)f(x)Δx+f(x)f(x + \Delta x) \approx f'(x)\,\Delta x + f(x)y=mx+by = mx + b

2. Cheatsheet

Each entry gives you ff and ff', then lets you pick the reference point. The formula, the table and the graph all follow that choice. Pick any x0x_0 to see the general form and drag the reference around.

2.1 Trigonometric

2.1.1sinx\sin x
sinx\sin x
\approx
sinx0+cosx0(xx0)\sin x_0 + \cos x_0\,(x - x_0)
reference
at 00
sinx\sin x
\approx
sin0+cos0(x0)\sin 0 + \cos 0\,(x - 0)
\approx
xx
f(x)f(x)f~(x)\tilde f(x)error
0+00 + 00.0000000.0000000.0000000.0000000.0000000.000000
0+11000 + \frac{1}{100}0.0100000.0100000.0100000.0100001.67×1071.67 \times 10^{-7}
0+1100 + \frac{1}{10}0.0998330.0998330.1000000.1000000.0001670.000167
0+120 + \frac{1}{2}0.4794260.4794260.5000000.5000000.0205740.020574
0+10 + 10.8414710.8414711.0000001.0000000.1585290.158529
sinx\sin xf~\tilde f
at π6\frac{\pi}{6}
sinx\sin x
\approx
sinπ6+cosπ6(xπ6)\sin\frac{\pi}{6} + \cos\frac{\pi}{6}\left(x - \frac{\pi}{6}\right)
\approx
12+32(xπ6)\frac{1}{2} + \frac{\sqrt{3}}{2}\left(x - \frac{\pi}{6}\right)
f(x)f(x)f~(x)\tilde f(x)error
π6+0\frac{\pi}{6} + 00.5000000.5000000.5000000.5000000.0000000.000000
π6+1100\frac{\pi}{6} + \frac{1}{100}0.5086350.5086350.5086600.5086602.51×1052.51 \times 10^{-5}
π6+110\frac{\pi}{6} + \frac{1}{10}0.5839600.5839600.5866030.5866030.0026420.002642
π6+12\frac{\pi}{6} + \frac{1}{2}0.8539860.8539860.9330130.9330130.0790270.079027
π6+1\frac{\pi}{6} + 10.9988860.9988861.3660251.3660250.3671390.367139
sinx\sin xf~\tilde f
at π4\frac{\pi}{4}
sinx\sin x
\approx
sinπ4+cosπ4(xπ4)\sin\frac{\pi}{4} + \cos\frac{\pi}{4}\left(x - \frac{\pi}{4}\right)
\approx
22+22(xπ4)\frac{\sqrt{2}}{2} + \frac{\sqrt{2}}{2}\left(x - \frac{\pi}{4}\right)
f(x)f(x)f~(x)\tilde f(x)error
π4+0\frac{\pi}{4} + 00.7071070.7071070.7071070.7071070.0000000.000000
π4+1100\frac{\pi}{4} + \frac{1}{100}0.7141420.7141420.7141780.7141783.55×1053.55 \times 10^{-5}
π4+110\frac{\pi}{4} + \frac{1}{10}0.7741670.7741670.7778170.7778170.0036500.003650
π4+12\frac{\pi}{4} + \frac{1}{2}0.9595500.9595501.0606601.0606600.1011110.101111
π4+1\frac{\pi}{4} + 10.9770610.9770611.4142141.4142140.4371520.437152
sinx\sin xf~\tilde f
at π2\frac{\pi}{2}
sinx\sin x
\approx
sinπ2+cosπ2(xπ2)\sin\frac{\pi}{2} + \cos\frac{\pi}{2}\left(x - \frac{\pi}{2}\right)
\approx
11
f(x)f(x)f~(x)\tilde f(x)error
π2+0\frac{\pi}{2} + 01.0000001.0000001.0000001.0000000.0000000.000000
π2+1100\frac{\pi}{2} + \frac{1}{100}0.9999500.9999501.0000001.0000005.00×1055.00 \times 10^{-5}
π2+110\frac{\pi}{2} + \frac{1}{10}0.9950040.9950041.0000001.0000000.0049960.004996
π2+12\frac{\pi}{2} + \frac{1}{2}0.8775830.8775831.0000001.0000000.1224170.122417
π2+1\frac{\pi}{2} + 10.5403020.5403021.0000001.0000000.4596980.459698
sinx\sin xf~\tilde f

At 0 this is the small-angle rule. At π/2 the curve is flat, so the approximation is just the constant 1.

2.1.2cosx\cos x
cosx\cos x
\approx
cosx0sinx0(xx0)\cos x_0 - \sin x_0\,(x - x_0)
reference
at 00
cosx\cos x
\approx
cos0sin0(x0)\cos 0 - \sin 0\,(x - 0)
\approx
11
f(x)f(x)f~(x)\tilde f(x)error
0+00 + 01.0000001.0000001.0000001.0000000.0000000.000000
0+11000 + \frac{1}{100}0.9999500.9999501.0000001.0000005.00×1055.00 \times 10^{-5}
0+1100 + \frac{1}{10}0.9950040.9950041.0000001.0000000.0049960.004996
0+120 + \frac{1}{2}0.8775830.8775831.0000001.0000000.1224170.122417
0+10 + 10.5403020.5403021.0000001.0000000.4596980.459698
cosx\cos xf~\tilde f
at π6\frac{\pi}{6}
cosx\cos x
\approx
cosπ6sinπ6(xπ6)\cos\frac{\pi}{6} - \sin\frac{\pi}{6}\left(x - \frac{\pi}{6}\right)
\approx
3212(xπ6)\frac{\sqrt{3}}{2} - \frac{1}{2}\left(x - \frac{\pi}{6}\right)
f(x)f(x)f~(x)\tilde f(x)error
π6+0\frac{\pi}{6} + 00.8660250.8660250.8660250.8660250.0000000.000000
π6+1100\frac{\pi}{6} + \frac{1}{100}0.8609820.8609820.8610250.8610254.32×1054.32 \times 10^{-5}
π6+110\frac{\pi}{6} + \frac{1}{10}0.8117820.8117820.8160250.8160250.0042430.004243
π6+12\frac{\pi}{6} + \frac{1}{2}0.5202960.5202960.6160250.6160250.0957290.095729
π6+1\frac{\pi}{6} + 10.0471800.0471800.3660250.3660250.3188450.318845
cosx\cos xf~\tilde f
at π3\frac{\pi}{3}
cosx\cos x
\approx
cosπ3sinπ3(xπ3)\cos\frac{\pi}{3} - \sin\frac{\pi}{3}\left(x - \frac{\pi}{3}\right)
\approx
1232(xπ3)\frac{1}{2} - \frac{\sqrt{3}}{2}\left(x - \frac{\pi}{3}\right)
f(x)f(x)f~(x)\tilde f(x)error
π3+0\frac{\pi}{3} + 00.5000000.5000000.5000000.5000000.0000000.000000
π3+1100\frac{\pi}{3} + \frac{1}{100}0.4913150.4913150.4913400.4913402.49×1052.49 \times 10^{-5}
π3+110\frac{\pi}{3} + \frac{1}{10}0.4110440.4110440.4133970.4133970.0023540.002354
π3+12\frac{\pi}{3} + \frac{1}{2}0.0235970.0235970.0669870.0669870.0433910.043391
π3+1\frac{\pi}{3} + 10.458584-0.4585840.366025-0.3660250.0925590.092559
cosx\cos xf~\tilde f
at π2\frac{\pi}{2}
cosx\cos x
\approx
cosπ2sinπ2(xπ2)\cos\frac{\pi}{2} - \sin\frac{\pi}{2}\left(x - \frac{\pi}{2}\right)
\approx
π2x\frac{\pi}{2} - x
f(x)f(x)f~(x)\tilde f(x)error
π2+0\frac{\pi}{2} + 06.12×10176.12 \times 10^{-17}6.12×10176.12 \times 10^{-17}0.0000000.000000
π2+1100\frac{\pi}{2} + \frac{1}{100}0.010000-0.0100000.010000-0.0100001.67×1071.67 \times 10^{-7}
π2+110\frac{\pi}{2} + \frac{1}{10}0.099833-0.0998330.100000-0.1000000.0001670.000167
π2+12\frac{\pi}{2} + \frac{1}{2}0.479426-0.4794260.500000-0.5000000.0205740.020574
π2+1\frac{\pi}{2} + 10.841471-0.8414711.000000-1.0000000.1585290.158529
cosx\cos xf~\tilde f

Flat at 0, so the line there is the constant 1. At π/2 it is steepest and the approximation is a pure slope.

2.1.3tanx\tan x
tanx\tan x
\approx
tanx0+sec2x0(xx0)\tan x_0 + \sec^2 x_0\,(x - x_0)
reference
at 00
tanx\tan x
\approx
tan0+sec20(x0)\tan 0 + \sec^2 0\,(x - 0)
\approx
xx
f(x)f(x)f~(x)\tilde f(x)error
0+00 + 00.0000000.0000000.0000000.0000000.0000000.000000
0+11000 + \frac{1}{100}0.0100000.0100000.0100000.0100003.33×1073.33 \times 10^{-7}
0+1100 + \frac{1}{10}0.1003350.1003350.1000000.1000000.0003350.000335
0+120 + \frac{1}{2}0.5463020.5463020.5000000.5000000.0463020.046302
0+10 + 11.5574081.5574081.0000001.0000000.5574080.557408
tanx\tan xf~\tilde f
at π6\frac{\pi}{6}
tanx\tan x
\approx
tanπ6+sec2π6(xπ6)\tan\frac{\pi}{6} + \sec^2\frac{\pi}{6}\left(x - \frac{\pi}{6}\right)
\approx
13+43(xπ6)\frac{1}{\sqrt{3}} + \frac{4}{3}\left(x - \frac{\pi}{6}\right)
f(x)f(x)f~(x)\tilde f(x)error
π6+0\frac{\pi}{6} + 00.5773500.5773500.5773500.5773500.0000000.000000
π6+1100\frac{\pi}{6} + \frac{1}{100}0.5907610.5907610.5906840.5906847.79×1057.79 \times 10^{-5}
π6+110\frac{\pi}{6} + \frac{1}{10}0.7193560.7193560.7106840.7106840.0086720.008672
π6+12\frac{\pi}{6} + \frac{1}{2}1.6413461.6413461.2440171.2440170.3973290.397329
π6+1\frac{\pi}{6} + 121.17180521.1718051.9106841.91068419.26112119.261121
tanx\tan xf~\tilde f
at π4\frac{\pi}{4}
tanx\tan x
\approx
tanπ4+sec2π4(xπ4)\tan\frac{\pi}{4} + \sec^2\frac{\pi}{4}\left(x - \frac{\pi}{4}\right)
\approx
1+2(xπ4)1 + 2\left(x - \frac{\pi}{4}\right)
f(x)f(x)f~(x)\tilde f(x)error
π4+0\frac{\pi}{4} + 01.0000001.0000001.0000001.0000000.0000000.000000
π4+1100\frac{\pi}{4} + \frac{1}{100}1.0202031.0202031.0200001.0200000.0002030.000203
π4+110\frac{\pi}{4} + \frac{1}{10}1.2230491.2230491.2000001.2000000.0230490.023049
π4+12\frac{\pi}{4} + \frac{1}{2}3.4082233.4082232.0000002.0000001.4082231.408223
π4+1\frac{\pi}{4} + 14.588038-4.5880383.0000003.0000007.5880387.588038
tanx\tan xf~\tilde f

Near 0, tan x flattens to x, the same as sin x. It steepens fast as you approach π/2.

2.1.4secx\sec x
secx\sec x
\approx
secx0+secx0tanx0(xx0)\sec x_0 + \sec x_0 \tan x_0\,(x - x_0)
reference
at 00
secx\sec x
\approx
sec0+sec0tan0(x0)\sec 0 + \sec 0\tan 0\,(x - 0)
\approx
11
f(x)f(x)f~(x)\tilde f(x)error
0+00 + 01.0000001.0000001.0000001.0000000.0000000.000000
0+11000 + \frac{1}{100}1.0000501.0000501.0000001.0000005.00×1055.00 \times 10^{-5}
0+1100 + \frac{1}{10}1.0050211.0050211.0000001.0000000.0050210.005021
0+120 + \frac{1}{2}1.1394941.1394941.0000001.0000000.1394940.139494
0+10 + 11.8508161.8508161.0000001.0000000.8508160.850816
secx\sec xf~\tilde f
at π6\frac{\pi}{6}
secx\sec x
\approx
secπ6+secπ6tanπ6(xπ6)\sec\frac{\pi}{6} + \sec\frac{\pi}{6}\tan\frac{\pi}{6}\left(x - \frac{\pi}{6}\right)
\approx
23+23(xπ6)\frac{2}{\sqrt{3}} + \frac{2}{3}\left(x - \frac{\pi}{6}\right)
f(x)f(x)f~(x)\tilde f(x)error
π6+0\frac{\pi}{6} + 01.1547011.1547011.1547011.1547010.0000000.000000
π6+1100\frac{\pi}{6} + \frac{1}{100}1.1614641.1614641.1613671.1613679.70×1059.70 \times 10^{-5}
π6+110\frac{\pi}{6} + \frac{1}{10}1.2318581.2318581.2213671.2213670.0104900.010490
π6+12\frac{\pi}{6} + \frac{1}{2}1.9219831.9219831.4880341.4880340.4339490.433949
π6+1\frac{\pi}{6} + 121.19540821.1954081.8213671.82136719.37404119.374041
secx\sec xf~\tilde f
at π4\frac{\pi}{4}
secx\sec x
\approx
secπ4+secπ4tanπ4(xπ4)\sec\frac{\pi}{4} + \sec\frac{\pi}{4}\tan\frac{\pi}{4}\left(x - \frac{\pi}{4}\right)
\approx
2+2(xπ4)\sqrt{2} + \sqrt{2}\left(x - \frac{\pi}{4}\right)
f(x)f(x)f~(x)\tilde f(x)error
π4+0\frac{\pi}{4} + 01.4142141.4142141.4142141.4142140.0000000.000000
π4+1100\frac{\pi}{4} + \frac{1}{100}1.4285701.4285701.4283561.4283560.0002150.000215
π4+110\frac{\pi}{4} + \frac{1}{10}1.5798251.5798251.5556351.5556350.0241910.024191
π4+12\frac{\pi}{4} + \frac{1}{2}3.5518993.5518992.1213202.1213201.4305791.430579
π4+1\frac{\pi}{4} + 14.695752-4.6957522.8284272.8284277.5241807.524180
secx\sec xf~\tilde f

Flat at 0 like cos, so the line there is the constant 1.

2.1.5cotx\cot x
cotx\cot x
\approx
cotx0csc2x0(xx0)\cot x_0 - \csc^2 x_0\,(x - x_0)
reference
at π4\frac{\pi}{4}
cotx\cot x
\approx
cotπ4csc2π4(xπ4)\cot\frac{\pi}{4} - \csc^2\frac{\pi}{4}\left(x - \frac{\pi}{4}\right)
\approx
12(xπ4)1 - 2\left(x - \frac{\pi}{4}\right)
f(x)f(x)f~(x)\tilde f(x)error
π4+0\frac{\pi}{4} + 01.0000001.0000001.0000001.0000000.0000000.000000
π4+1100\frac{\pi}{4} + \frac{1}{100}0.9801970.9801970.9800000.9800000.0001970.000197
π4+110\frac{\pi}{4} + \frac{1}{10}0.8176290.8176290.8000000.8000000.0176290.017629
π4+12\frac{\pi}{4} + \frac{1}{2}0.2934080.2934080.0000000.0000000.2934080.293408
π4+1\frac{\pi}{4} + 10.217958-0.2179581.000000-1.0000000.7820420.782042
cotx\cot xf~\tilde f
at π2\frac{\pi}{2}
cotx\cot x
\approx
cotπ2csc2π2(xπ2)\cot\frac{\pi}{2} - \csc^2\frac{\pi}{2}\left(x - \frac{\pi}{2}\right)
\approx
π2x\frac{\pi}{2} - x
f(x)f(x)f~(x)\tilde f(x)error
π2+0\frac{\pi}{2} + 06.12×10176.12 \times 10^{-17}6.12×10176.12 \times 10^{-17}0.0000000.000000
π2+1100\frac{\pi}{2} + \frac{1}{100}0.010000-0.0100000.010000-0.0100003.33×1073.33 \times 10^{-7}
π2+110\frac{\pi}{2} + \frac{1}{10}0.100335-0.1003350.100000-0.1000000.0003350.000335
π2+12\frac{\pi}{2} + \frac{1}{2}0.546302-0.5463020.500000-0.5000000.0463020.046302
π2+1\frac{\pi}{2} + 11.557408-1.5574081.000000-1.0000000.5574080.557408
cotx\cot xf~\tilde f

There is no reference point at 0 — cot blows up there. Away from 0 it behaves like anything else.

2.1.6cscx\csc x
cscx\csc x
\approx
cscx0cscx0cotx0(xx0)\csc x_0 - \csc x_0 \cot x_0\,(x - x_0)
reference
at π4\frac{\pi}{4}
cscx\csc x
\approx
cscπ4cscπ4cotπ4(xπ4)\csc\frac{\pi}{4} - \csc\frac{\pi}{4}\cot\frac{\pi}{4}\left(x - \frac{\pi}{4}\right)
\approx
22(xπ4)\sqrt{2} - \sqrt{2}\left(x - \frac{\pi}{4}\right)
f(x)f(x)f~(x)\tilde f(x)error
π4+0\frac{\pi}{4} + 01.4142141.4142141.4142141.4142140.0000000.000000
π4+1100\frac{\pi}{4} + \frac{1}{100}1.4002811.4002811.4000711.4000710.0002100.000210
π4+110\frac{\pi}{4} + \frac{1}{10}1.2917111.2917111.2727921.2727920.0189190.018919
π4+12\frac{\pi}{4} + \frac{1}{2}1.0421561.0421560.7071070.7071070.3350490.335049
π4+1\frac{\pi}{4} + 11.0234771.0234772.22×1016-2.22 \times 10^{-16}1.0234771.023477
cscx\csc xf~\tilde f
at π2\frac{\pi}{2}
cscx\csc x
\approx
cscπ2cscπ2cotπ2(xπ2)\csc\frac{\pi}{2} - \csc\frac{\pi}{2}\cot\frac{\pi}{2}\left(x - \frac{\pi}{2}\right)
\approx
11
f(x)f(x)f~(x)\tilde f(x)error
π2+0\frac{\pi}{2} + 01.0000001.0000001.0000001.0000000.0000000.000000
π2+1100\frac{\pi}{2} + \frac{1}{100}1.0000501.0000501.0000001.0000005.00×1055.00 \times 10^{-5}
π2+110\frac{\pi}{2} + \frac{1}{10}1.0050211.0050211.0000001.0000000.0050210.005021
π2+12\frac{\pi}{2} + \frac{1}{2}1.1394941.1394941.0000001.0000000.1394940.139494
π2+1\frac{\pi}{2} + 11.8508161.8508161.0000001.0000000.8508160.850816
cscx\csc xf~\tilde f

Same as cot — undefined at 0, fine everywhere else. At π/2 it bottoms out, so the line is flat.

2.2 Exponents and logarithms

2.2.1exe^x
exe^x
\approx
ex0+ex0(xx0)e^{x_0} + e^{x_0}\,(x - x_0)
reference
at 00
exe^x
\approx
e0+e0(x0)e^0 + e^0\,(x - 0)
\approx
1+x1 + x
f(x)f(x)f~(x)\tilde f(x)error
0+00 + 01.0000001.0000001.0000001.0000000.0000000.000000
0+11000 + \frac{1}{100}1.0100501.0100501.0100001.0100005.02×1055.02 \times 10^{-5}
0+1100 + \frac{1}{10}1.1051711.1051711.1000001.1000000.0051710.005171
0+120 + \frac{1}{2}1.6487211.6487211.5000001.5000000.1487210.148721
0+10 + 12.7182822.7182822.0000002.0000000.7182820.718282
exe^xf~\tilde f
at 11
exe^x
\approx
e1+e1(x1)e^1 + e^1\,(x - 1)
\approx
exe\,x
f(x)f(x)f~(x)\tilde f(x)error
1+01 + 02.7182822.7182822.7182822.7182820.0000000.000000
1+11001 + \frac{1}{100}2.7456012.7456012.7454652.7454650.0001360.000136
1+1101 + \frac{1}{10}3.0041663.0041662.9901102.9901100.0140560.014056
1+121 + \frac{1}{2}4.4816894.4816894.0774234.0774230.4042660.404266
1+11 + 17.3890567.3890565.4365645.4365641.9524921.952492
exe^xf~\tilde f
at 22
exe^x
\approx
e2+e2(x2)e^2 + e^2\,(x - 2)
\approx
e2(x1)e^2\left(x - 1\right)
f(x)f(x)f~(x)\tilde f(x)error
2+02 + 07.3890567.3890567.3890567.3890560.0000000.000000
2+11002 + \frac{1}{100}7.4633177.4633177.4629477.4629470.0003710.000371
2+1102 + \frac{1}{10}8.1661708.1661708.1279628.1279620.0382080.038208
2+122 + \frac{1}{2}12.18249412.18249411.08358411.0835841.0989101.098910
2+12 + 120.08553720.08553714.77811214.7781125.3074255.307425
exe^xf~\tilde f

At x₀ = 1 the two terms collapse: e + e(x − 1) = e·x.

2.2.2ln(1+x)\ln(1+x)
ln(1+x)\ln(1+x)
\approx
ln(1+x0)+xx01+x0\ln(1+x_0) + \dfrac{x - x_0}{1 + x_0}
reference
at 00
ln(1+x)\ln(1+x)
\approx
ln(1+0)+x01+0\ln(1+0) + \dfrac{x - 0}{1 + 0}
\approx
xx
f(x)f(x)f~(x)\tilde f(x)error
0+00 + 00.0000000.0000000.0000000.0000000.0000000.000000
0+11000 + \frac{1}{100}0.0099500.0099500.0100000.0100004.97×1054.97 \times 10^{-5}
0+1100 + \frac{1}{10}0.0953100.0953100.1000000.1000000.0046900.004690
0+120 + \frac{1}{2}0.4054650.4054650.5000000.5000000.0945350.094535
0+10 + 10.6931470.6931471.0000001.0000000.3068530.306853
ln(1+x)\ln(1+x)f~\tilde f
at 11
ln(1+x)\ln(1+x)
\approx
ln(1+1)+x11+1\ln(1+1) + \dfrac{x - 1}{1 + 1}
\approx
ln2+x12\ln 2 + \frac{x - 1}{2}
f(x)f(x)f~(x)\tilde f(x)error
1+01 + 00.6931470.6931470.6931470.6931470.0000000.000000
1+11001 + \frac{1}{100}0.6981350.6981350.6981470.6981471.25×1051.25 \times 10^{-5}
1+1101 + \frac{1}{10}0.7419370.7419370.7431470.7431470.0012100.001210
1+121 + \frac{1}{2}0.9162910.9162910.9431470.9431470.0268560.026856
1+11 + 11.0986121.0986121.1931471.1931470.0945350.094535
ln(1+x)\ln(1+x)f~\tilde f
at 22
ln(1+x)\ln(1+x)
\approx
ln(1+2)+x21+2\ln(1+2) + \dfrac{x - 2}{1 + 2}
\approx
ln3+x23\ln 3 + \frac{x - 2}{3}
f(x)f(x)f~(x)\tilde f(x)error
2+02 + 01.0986121.0986121.0986121.0986120.0000000.000000
2+11002 + \frac{1}{100}1.1019401.1019401.1019461.1019465.54×1065.54 \times 10^{-6}
2+1102 + \frac{1}{10}1.1314021.1314021.1319461.1319460.0005440.000544
2+122 + \frac{1}{2}1.2527631.2527631.2652791.2652790.0125160.012516
2+12 + 11.3862941.3862941.4319461.4319460.0456510.045651
ln(1+x)\ln(1+x)f~\tilde f

Written as ln(1 + x) so that 0 is a usable reference point. Plain ln x has none.

2.2.3(1+x)r(1+x)^r
r =
1+x\sqrt{1+x}
\approx
1+x0+121+x0(xx0)\sqrt{1+x_0} + \dfrac{1}{2\sqrt{1+x_0}}(x - x_0)
reference
at 00
(1+x)r(1+x)^r
\approx
(1+0)r+r(1+0)r1(x0)(1+0)^r + r(1+0)^{r-1}(x - 0)
\approx
1+rx1 + rx
f(x)f(x)f~(x)\tilde f(x)error
0+00 + 01.0000001.0000001.0000001.0000000.0000000.000000
0+11000 + \frac{1}{100}1.0049881.0049881.0050001.0050001.24×1051.24 \times 10^{-5}
0+1100 + \frac{1}{10}1.0488091.0488091.0500001.0500000.0011910.001191
0+120 + \frac{1}{2}1.2247451.2247451.2500001.2500000.0252550.025255
0+10 + 11.4142141.4142141.5000001.5000000.0857860.085786
(1+x)r(1+x)^rf~\tilde f
at 11
(1+x)r(1+x)^r
\approx
(1+1)r+r(1+1)r1(x1)(1+1)^r + r(1+1)^{r-1}(x - 1)
\approx
2r+r2r1(x1)2^r + r\,2^{r-1}(x - 1)
f(x)f(x)f~(x)\tilde f(x)error
1+01 + 01.4142141.4142141.4142141.4142140.0000000.000000
1+11001 + \frac{1}{100}1.4177451.4177451.4177491.4177494.41×1064.41 \times 10^{-6}
1+1101 + \frac{1}{10}1.4491381.4491381.4495691.4495690.0004310.000431
1+121 + \frac{1}{2}1.5811391.5811391.5909901.5909900.0098510.009851
1+11 + 11.7320511.7320511.7677671.7677670.0357160.035716
(1+x)r(1+x)^rf~\tilde f
11+x\dfrac{1}{\sqrt{1+x}}
\approx
11+x012(1+x0)3/2(xx0)\dfrac{1}{\sqrt{1+x_0}} - \dfrac{1}{2(1+x_0)^{3/2}}(x - x_0)
reference
at 00
(1+x)r(1+x)^r
\approx
(1+0)r+r(1+0)r1(x0)(1+0)^r + r(1+0)^{r-1}(x - 0)
\approx
1+rx1 + rx
f(x)f(x)f~(x)\tilde f(x)error
0+00 + 01.0000001.0000001.0000001.0000000.0000000.000000
0+11000 + \frac{1}{100}0.9950370.9950370.9950000.9950003.72×1053.72 \times 10^{-5}
0+1100 + \frac{1}{10}0.9534630.9534630.9500000.9500000.0034630.003463
0+120 + \frac{1}{2}0.8164970.8164970.7500000.7500000.0664970.066497
0+10 + 10.7071070.7071070.5000000.5000000.2071070.207107
(1+x)r(1+x)^rf~\tilde f
at 11
(1+x)r(1+x)^r
\approx
(1+1)r+r(1+1)r1(x1)(1+1)^r + r(1+1)^{r-1}(x - 1)
\approx
2r+r2r1(x1)2^r + r\,2^{r-1}(x - 1)
f(x)f(x)f~(x)\tilde f(x)error
1+01 + 00.7071070.7071070.7071070.7071070.0000000.000000
1+11001 + \frac{1}{100}0.7053460.7053460.7053390.7053396.60×1066.60 \times 10^{-6}
1+1101 + \frac{1}{10}0.6900660.6900660.6894290.6894290.0006360.000636
1+121 + \frac{1}{2}0.6324560.6324560.6187180.6187180.0137370.013737
1+11 + 10.5773500.5773500.5303300.5303300.0470200.047020
(1+x)r(1+x)^rf~\tilde f
11+x\dfrac{1}{1+x}
\approx
11+x01(1+x0)2(xx0)\dfrac{1}{1+x_0} - \dfrac{1}{(1+x_0)^2}(x - x_0)
reference
at 00
(1+x)r(1+x)^r
\approx
(1+0)r+r(1+0)r1(x0)(1+0)^r + r(1+0)^{r-1}(x - 0)
\approx
1+rx1 + rx
f(x)f(x)f~(x)\tilde f(x)error
0+00 + 01.0000001.0000001.0000001.0000000.0000000.000000
0+11000 + \frac{1}{100}0.9900990.9900990.9900000.9900009.90×1059.90 \times 10^{-5}
0+1100 + \frac{1}{10}0.9090910.9090910.9000000.9000000.0090910.009091
0+120 + \frac{1}{2}0.6666670.6666670.5000000.5000000.1666670.166667
0+10 + 10.5000000.5000000.0000000.0000000.5000000.500000
(1+x)r(1+x)^rf~\tilde f
at 11
(1+x)r(1+x)^r
\approx
(1+1)r+r(1+1)r1(x1)(1+1)^r + r(1+1)^{r-1}(x - 1)
\approx
2r+r2r1(x1)2^r + r\,2^{r-1}(x - 1)
f(x)f(x)f~(x)\tilde f(x)error
1+01 + 00.5000000.5000000.5000000.5000000.0000000.000000
1+11001 + \frac{1}{100}0.4975120.4975120.4975000.4975001.24×1051.24 \times 10^{-5}
1+1101 + \frac{1}{10}0.4761900.4761900.4750000.4750000.0011900.001190
1+121 + \frac{1}{2}0.4000000.4000000.3750000.3750000.0250000.025000
1+11 + 10.3333330.3333330.2500000.2500000.0833330.083333
(1+x)r(1+x)^rf~\tilde f
(1+x)2(1+x)^2
\approx
(1+x0)2+2(1+x0)(xx0)(1+x_0)^2 + 2(1+x_0)(x - x_0)
reference
at 00
(1+x)r(1+x)^r
\approx
(1+0)r+r(1+0)r1(x0)(1+0)^r + r(1+0)^{r-1}(x - 0)
\approx
1+rx1 + rx
f(x)f(x)f~(x)\tilde f(x)error
0+00 + 01.0000001.0000001.0000001.0000000.0000000.000000
0+11000 + \frac{1}{100}1.0201001.0201001.0200001.0200001.00×1041.00 \times 10^{-4}
0+1100 + \frac{1}{10}1.2100001.2100001.2000001.2000000.0100000.010000
0+120 + \frac{1}{2}2.2500002.2500002.0000002.0000000.2500000.250000
0+10 + 14.0000004.0000003.0000003.0000001.0000001.000000
(1+x)r(1+x)^rf~\tilde f
at 11
(1+x)r(1+x)^r
\approx
(1+1)r+r(1+1)r1(x1)(1+1)^r + r(1+1)^{r-1}(x - 1)
\approx
2r+r2r1(x1)2^r + r\,2^{r-1}(x - 1)
f(x)f(x)f~(x)\tilde f(x)error
1+01 + 04.0000004.0000004.0000004.0000000.0000000.000000
1+11001 + \frac{1}{100}4.0401004.0401004.0400004.0400001.00×1041.00 \times 10^{-4}
1+1101 + \frac{1}{10}4.4100004.4100004.4000004.4000000.0100000.010000
1+121 + \frac{1}{2}6.2500006.2500006.0000006.0000000.2500000.250000
1+11 + 19.0000009.0000008.0000008.0000001.0000001.000000
(1+x)r(1+x)^rf~\tilde f
(1+x)3(1+x)^3
\approx
(1+x0)3+3(1+x0)2(xx0)(1+x_0)^3 + 3(1+x_0)^2(x - x_0)
reference
at 00
(1+x)r(1+x)^r
\approx
(1+0)r+r(1+0)r1(x0)(1+0)^r + r(1+0)^{r-1}(x - 0)
\approx
1+rx1 + rx
f(x)f(x)f~(x)\tilde f(x)error
0+00 + 01.0000001.0000001.0000001.0000000.0000000.000000
0+11000 + \frac{1}{100}1.0303011.0303011.0300001.0300000.0003010.000301
0+1100 + \frac{1}{10}1.3310001.3310001.3000001.3000000.0310000.031000
0+120 + \frac{1}{2}3.3750003.3750002.5000002.5000000.8750000.875000
0+10 + 18.0000008.0000004.0000004.0000004.0000004.000000
(1+x)r(1+x)^rf~\tilde f
at 11
(1+x)r(1+x)^r
\approx
(1+1)r+r(1+1)r1(x1)(1+1)^r + r(1+1)^{r-1}(x - 1)
\approx
2r+r2r1(x1)2^r + r\,2^{r-1}(x - 1)
f(x)f(x)f~(x)\tilde f(x)error
1+01 + 08.0000008.0000008.0000008.0000000.0000000.000000
1+11001 + \frac{1}{100}8.1206018.1206018.1200008.1200000.0006010.000601
1+1101 + \frac{1}{10}9.2610009.2610009.2000009.2000000.0610000.061000
1+121 + \frac{1}{2}15.62500015.62500014.00000014.0000001.6250001.625000
1+11 + 127.00000027.00000020.00000020.0000007.0000007.000000
(1+x)r(1+x)^rf~\tilde f

The workhorse. r = 1/2 gives √(1+x), and r = −1 gives 1/(1+x).

2.2.4axa^x
a =
axa^x
\approx
ax0+ax0lna(xx0)a^{x_0} + a^{x_0}\ln a\,(x - x_0)
reference
at 00
axa^x
\approx
a0+a0lna(x0)a^0 + a^0\ln a\,(x - 0)
\approx
1+xlna1 + x\ln a
f(x)f(x)f~(x)\tilde f(x)error
0+00 + 01.0000001.0000001.0000001.0000000.0000000.000000
0+11000 + \frac{1}{100}1.0069561.0069561.0069311.0069312.41×1052.41 \times 10^{-5}
0+1100 + \frac{1}{10}1.0717731.0717731.0693151.0693150.0024590.002459
0+120 + \frac{1}{2}1.4142141.4142141.3465741.3465740.0676400.067640
0+10 + 12.0000002.0000001.6931471.6931470.3068530.306853
axa^xf~\tilde f
at 11
axa^x
\approx
a1+a1lna(x1)a^1 + a^1\ln a\,(x - 1)
\approx
a+alna(x1)a + a\ln a\,(x - 1)
f(x)f(x)f~(x)\tilde f(x)error
1+01 + 02.0000002.0000002.0000002.0000000.0000000.000000
1+11001 + \frac{1}{100}2.0139112.0139112.0138632.0138634.82×1054.82 \times 10^{-5}
1+1101 + \frac{1}{10}2.1435472.1435472.1386292.1386290.0049170.004917
1+121 + \frac{1}{2}2.8284272.8284272.6931472.6931470.1352800.135280
1+11 + 14.0000004.0000003.3862943.3862940.6137060.613706
axa^xf~\tilde f
axa^x
\approx
ax0+ax0lna(xx0)a^{x_0} + a^{x_0}\ln a\,(x - x_0)
reference
at 00
axa^x
\approx
a0+a0lna(x0)a^0 + a^0\ln a\,(x - 0)
\approx
1+xlna1 + x\ln a
f(x)f(x)f~(x)\tilde f(x)error
0+00 + 01.0000001.0000001.0000001.0000000.0000000.000000
0+11000 + \frac{1}{100}1.0110471.0110471.0109861.0109866.06×1056.06 \times 10^{-5}
0+1100 + \frac{1}{10}1.1161231.1161231.1098611.1098610.0062620.006262
0+120 + \frac{1}{2}1.7320511.7320511.5493061.5493060.1827450.182745
0+10 + 13.0000003.0000002.0986122.0986120.9013880.901388
axa^xf~\tilde f
at 11
axa^x
\approx
a1+a1lna(x1)a^1 + a^1\ln a\,(x - 1)
\approx
a+alna(x1)a + a\ln a\,(x - 1)
f(x)f(x)f~(x)\tilde f(x)error
1+01 + 03.0000003.0000003.0000003.0000000.0000000.000000
1+11001 + \frac{1}{100}3.0331403.0331403.0329583.0329580.0001820.000182
1+1101 + \frac{1}{10}3.3483703.3483703.3295843.3295840.0187860.018786
1+121 + \frac{1}{2}5.1961525.1961524.6479184.6479180.5482340.548234
1+11 + 19.0000009.0000006.2958376.2958372.7041632.704163
axa^xf~\tilde f
axa^x
\approx
ax0+ax0lna(xx0)a^{x_0} + a^{x_0}\ln a\,(x - x_0)
reference
at 00
axa^x
\approx
a0+a0lna(x0)a^0 + a^0\ln a\,(x - 0)
\approx
1+xlna1 + x\ln a
f(x)f(x)f~(x)\tilde f(x)error
0+00 + 01.0000001.0000001.0000001.0000000.0000000.000000
0+11000 + \frac{1}{100}1.0232931.0232931.0230261.0230260.0002670.000267
0+1100 + \frac{1}{10}1.2589251.2589251.2302591.2302590.0286670.028667
0+120 + \frac{1}{2}3.1622783.1622782.1512932.1512931.0109851.010985
0+10 + 110.00000010.0000003.3025853.3025856.6974156.697415
axa^xf~\tilde f
at 11
axa^x
\approx
a1+a1lna(x1)a^1 + a^1\ln a\,(x - 1)
\approx
a+alna(x1)a + a\ln a\,(x - 1)
f(x)f(x)f~(x)\tilde f(x)error
1+01 + 010.00000010.00000010.00000010.0000000.0000000.000000
1+11001 + \frac{1}{100}10.23293010.23293010.23025910.2302590.0026710.002671
1+1101 + \frac{1}{10}12.58925412.58925412.30258512.3025850.2866690.286669
1+121 + \frac{1}{2}31.62277731.62277721.51292521.51292510.10985110.109851
1+11 + 1100.000000100.00000033.02585133.02585166.97414966.974149
axa^xf~\tilde f

Covers e^x too. Set a = e, so ln a = 1.

3. Algebra

You do not have to differentiate a messy function at all. You can build its approximation out of the entries above instead.

Write L(f)L(f) for the linear approximation of ff near 00.

The useful fact is that you can approximate first and combine second. For a product:

L(L(f)L(g))=L(fg)L\big(L(f)\,L(g)\big) = L(fg)

So the LL of a product is just the linear approximation of the product of the linear approximations. You never have to touch fgfg itself.

Here is the full set:

rule
sumL(f+g)=L(f)+L(g)L(f+g) = L(f) + L(g)
scaleL(kf)=kL(f)L(kf) = k\,L(f)
productL(L(f)L(g))=L(fg)L\big(L(f)\,L(g)\big) = L(fg)
composition, when g(0)=0g(0)=0L(L(f)L(g))=L(fg)L\big(L(f)\circ L(g)\big) = L(f\circ g)

Applied to a polynomial, LL does nothing but chop off the high powers. So the outer LL on each left-hand side is the cut — the x2x^2 term a product or composition produces along the way is not trustworthy, since the inputs were only accurate to first order.

The g(0)=0g(0) = 0 condition on the last line is not a deep restriction — it falls out of how LL is defined here. L(f)L(f) only carries information about ff near 00, so plugging in g(x)g(x) is only valid where gg actually lands near 00. Centre LL at a different reference point and the condition moves with it: substitution is valid whenever the inside function's value at your reference point equals the outside function's reference point.

Linear Approximation Cheatsheet - Stefan Wullems