What is 22? School algebra defines ap/q as a root of a power and stops there, yet calculus freely uses xπ, 10x for all real x, and the rule dxdxm=mxm−1 for irrational m. In §214 Hardy fills the gap with one line, ax=exloga, and shows that all the familiar rules survive. This also removes the restriction ”x rational” from many results of earlier chapters.
The idea
For a rational exponent we already know two things: ap/q is the positive q-th root of ap, and its logarithm is qploga. So for rational r,
ar=erloga.
The right-hand side makes sense for every real r, because ey is now defined for every real y. So we take it as the definition when r is irrational. Is that the “right” value? The test is continuity: if r1,r2,… are rationals tending to x, the elementary powers arn should tend to the new ax.
ExploreClose in on 22 with rational exponents
Try this.With a=2 and exponent 2, increase the number of decimal places from 0 to 8. Each row uses only roots and powers: 21.41=1002141. Then try a<1 and the exponent π.
xaˣ
ax=exloga
rational powers ap/q
a2
rational exponent p/q
ap/q=qap
distance from exloga
1
2.000000000
0.665
1.4
2.639015822
0.0261
1.41
2.657371628
0.00777
1.414
2.664749650
0.000394
Definition: a2=e2loga=e0.980258=2.665144143. Each extra decimal place in the exponent brings ap/q roughly ten times closer, because exloga is continuous (indeed differentiable) in x.
What to notice
The elementary values 1.41,1.414,1.4142,… give 21.41≈2.6574, 21.414≈2.6647, … and settle on 2.665144…=e2log2, each extra digit gaining about one more correct digit. Every rational point lies on the one smooth curve exloga: no other continuous way of filling in the irrational exponents exists. For a<1 the curve falls instead of rising.
Precise statements
DefinitionThe general powerHardy: §214
For a>0 and every real x,
ax=exloga.
For rational x this is the positive value of the elementary power. Consequently ax>0 and logax=xloga for all real x. The power ax with irrational x is defined only for positive a.
TheoremProperties of axHardy: §214 (1)–(5)
Let a>0 and b>0, and let x, y be real.
Laws of indices:axay=ax+y and (ax)y=axy; also (ab)x=axbx.
Growth: if a>1, ax→∞ as x→∞, faster than any power of x; if 0<a<1, ax→0 as x→∞, faster than any power of 1/x.
Derivative in x:dxdax=axloga.
Derivative in a:dadax=xax−1. In other words dtdtm=mtm−1 for t>0 and every real m.
A limit:xax−1→loga as x→0.
Proofs
ProofThe laws of indices
Strategy
Everything reduces, through the definition, to eu+v=euev and logeu=u.
axay=ax+y
axay=exlogaeyloga=e(x+y)loga=ax+y.
(ax)y=axy
By definition (ax)y=eylog(ax), and logax=logexloga=xloga. So (ax)y=exyloga=axy.
(ab)x=axbx
(ab)x=exlogab=exloga+xlogb=axbx.
∎
ProofGrowth
Strategy
Write ax as an exponential eγx with γ=loga and use the previous lesson.
a>1
Then γ=loga>0 and ax=eγx, which tends to ∞ faster than any power of x.
0<a<1
Then γ=loga<0, and eγx⋅xk→0 for every k: ax tends to 0 faster than every power x−k. The graph is that of (1/a)x reflected left to right, since ax=(1/a)−x.
∎
ProofThe derivatives and the limit
Strategy
Differentiate exloga by the chain rule, once treating x as the variable and once a.
In x
dxdexloga=exloga⋅loga=axloga.
In a
dadexloga=exloga⋅ax=xaxa−1=xax−1, using the law of indices.
The limit
xax−1=x−0ax−a0 is a difference quotient of ax at 0, so it tends to the derivative there, a0loga=loga.
∎
Part 4 extends the power rule, which Chapter VI proved for rational exponents, to every real exponent. Part 5 is Examples LXXXVI, 3 in a new notation (there (eαy−1)/y→α; take α=loga, so that eαy=ay), as Hardy remarks, and its sequence form n(a1/n−1)→loga is the subject of §216.
Why no other definition would do. Suppose F(x) is any function that agrees with the elementary ax at rational x and is continuous. Given irrational x, choose rationals rn→x. Then F(x)=limF(rn)=limernloga=exloga, because ey is continuous. So continuity alone forces ax=exloga, which is what the widget shows numerically.
Worked examples and exercises
Exercise●●○Hardy: Examples LXXXVIII, 31
Show that x1/x→1 as x→∞, that x1/x has its greatest value at x=e, and sketch its graph for x>0.
Hint
x1/x=e(logx)/x. What does (logx)/x do?
Hint
Differentiate g(x)=(logx)/x: g′(x)=(1−logx)/x2.
Solution
x1/x=eg(x) with g(x)=(logx)/x. By §208, g(x)→0 as x→∞; since ey is continuous, x1/x→e0=1.
g′(x)=(1−logx)/x2 is positive for x<e and negative for x>e, so g, and with it eg, is greatest at x=e: the maximum value is e1/e≈1.4447.
As x→+0, g(x)=(logx)/x→−∞ (a large negative number divided by a small positive one), so x1/x→0. The graph rises from 0, crosses 1 at x=1, peaks at (e,e1/e), and falls slowly back towards 1.
Exerciseeπ or πe?●●○Hardy: Examples LXXXVIII, 31
Without a calculator, decide which is larger, eπ or πe.
Hint
Raise both to the power 1/(eπ) and use the previous exercise.
Solution
eπ>πe if and only if e1/e>π1/π (raise both sides to the positive power 1/(eπ); t↦t1/(eπ) is increasing). The function x1/x has its strict maximum at x=e, and π=e, so π1/π<e1/e. Hence eπ>πe. (Numerically 23.14>22.46.)
Exercise●○○Hardy: Examples LXXXVIII, 32
Show that xx→1 as x→+0. Where does xx (for x>0) take its least value?
Hint
xx=exlogx and xlogx→0, so xx→e0=1 by continuity of the exponential.
dxdxx=dxdexlogx=xx(logx+1), which is negative for x<1/e and positive for x>1/e. So the least value is at x=1/e, where xx=e−1/e≈0.6922. Note that neither rule "dxdxm=mxm−1" (exponent fixed) nor "dxdax=axloga" (base fixed) applies alone; the derivative is the sum of the two.
ExerciseThe power rule for every real exponent●○○Hardy: §214 (4)
Show that dxdx2=2x2−1 for x>0, and find ∫12xπdx.
Hint
Part 4 of the theorem, with a renamed x. Then reverse it.
Solution
By part 4 with m=2: dxdx2=2x2−1. In the same way dxdπ+1xπ+1=xπ, so ∫12xπdx=π+12π+1−1≈4.14216.65≈4.02.
Check yourself
Check yourselfEnter a number
Give 22 to four decimal places.
Why
22=e2log2=e1.41421×0.69315=e0.98026=2.66514…. The widget’s rational approximations 21.4142 and 21.41421 agree with this to four places.
Check yourselfTrue or false?
(−8)2 is defined by Hardy’s definition, since (−8)1/3=−2 makes sense.
Why
False. The definition ax=exloga needs loga, so a>0. Some rational powers of negative numbers make sense ((−8)1/3=−2), but not all ((−8)1/2), and there is no continuous way to fill in the irrational exponents. Complex powers of negative numbers are the subject of Chapter X.
Check yourselfChoose one
What is dxd2x?
Why
2x=exlog2, so the derivative is exlog2⋅log2. The first option is the derivative with respect to the base (part 4), the wrong variable here. Only base e gives a function equal to its own derivative, since loge=1.
Check yourselfChoose one
For 0<a<1, as x→∞, ax
Why
ax=e−βx with β=−loga>0, and xke−βx→0 for every k. The exponent tends to −∞, but the power itself is always positive.
Hardy’s language
Summary
Summary
For a>0 and any real x: ax=exloga. It agrees with roots of powers for rational x, and it is the only continuous function that does.
logax=xloga; the laws of indices axay=ax+y, (ax)y=axy hold for all real exponents.
dxdax=axloga, and dtdtm=mtm−1 for every real m.
ax→∞ (if a>1) or →0 (if a<1) faster than any power; (ax−1)/x→loga as x→0.