Chapter VII · Additional theorems in the differential and integral calculus
The binomial series by Cauchy's remainder
Cauchy's form of the remainder finishes the proof that for every between and , exactly where Lagrange's form gave up.
Why this matters
In §152 the binomial series
was proved for , but the proof broke down for negative . Lagrange’s remainder contains , and when and is large this factor could be enormous if happened to be close to ; knowing only , we could not rule that out. §168 closes the gap with the Cauchy form of the remainder from the previous lesson. It is a short section, but it is the payoff of the whole integral-remainder machinery, and the binomial series is used constantly later (Chapters VIII and IX).
The idea
Cauchy’s form of the remainder for contains the factor
This is the factor that rescues us. Whether is positive or negative, as long as the fraction is less than , so its power is less than whatever is. All the dangerous dependence on is in a quantity we can bound by ; what is left is an explicit number , which tends to by a ratio argument. Lagrange’s form has no such factor: its can only be bounded by , which grows faster than shrinks when .
Try this. Start at , . Compare the three families of dots: the true remainder, the bound Lagrange’s form guarantees, and Cauchy’s bound. Move slowly from to , watching the hollow orange dots; try other values of .
- true |Rₙ|
- Lagrange: the best bound when only 0 < θ < 1 is known
- Cauchy’s bound ρₙ
- size 1 (log₁₀ = 0)
What to notice
The true remainder always tends to for , and Cauchy’s bound follows it down, always a little above it. Lagrange’s bound tends to for but grows without limit for (watch it turn upwards as crosses ). A bound that grows proves nothing, even though the quantity it bounds is small.
Precise statements
Here for any real , and .
Let be rational and not a non-negative integer. Then for ,
that is, the remainder after terms tends to as . More precisely, where
and .
When is a non-negative integer the series stops after the term in and the formula holds for every . Once is defined for irrational (Chapter IX), the same proof works word for word. What happens at depends on ; that is a question about convergence of series (§222).
Proofs
Write Cauchy’s remainder for and regroup it so that every appearance of the unknown is in a factor we can bound independently of . What remains is , a fixed number for each , and a ratio argument shows .
Cauchy’s remainder for
. With and , Cauchy’s form (§167) gives, for some in ,
Regroup
Note that , and . So
The key factor is at most
is the same as , which is true because and . Both sides are positive, so , for positive and negative alike, and its -th power is at most .
Bound the last factor
lies between and , both positive. If the power is at most ; if it is at most ; if it is . Hence , and has disappeared.
The ratio of consecutive values is . Choose with ; then for all some , so (this is Examples XXVII, 13 in Hardy). Therefore .
The proof never uses the sign of except in the third step, and there only . That is the whole difference from §152.
Worked examples and exercises
Let and . Show that the best bound Lagrange’s form gives, knowing only , is , and that it tends to infinity when .
Hint
Hint
Solution
Since and decreases towards as , the factor can be as large as (nearly) , and nothing in the information excludes this. So the best bound is with .
The ratio . If then , so eventually grows geometrically and tends to infinity. (If , and Lagrange’s bound works; this is why §152 could have handled too.)
Use the series for to compute . How many terms does Cauchy’s bound guarantee for an error below ?
Hint
Hint
Solution
. With and :
So terms (, up to the term in ) are guaranteed to give to within . The true error after terms is about , and terms would already suffice: a bound is a guarantee, not an estimate.
Show that , and write down the expansion of for . Why are all its coefficients positive?
Hint
Solution
The numerator has negative factors , , so it is ; and . Hence the formula. Replacing by multiplies the -th term by , which cancels the sign:
All coefficients are positive because and all its derivatives are positive at ; each derivative of is a positive multiple of a negative power of . Here the theorem of this lesson (rather than Lagrange’s form) is needed precisely for , which corresponds to in the series for ; there all the terms are positive.
Show that as when , for any real .
Hint
Solution
Let . If is a non-negative integer, for . Otherwise no vanishes, and . Pick with . There is an with for , hence for , and (§72). So . The same argument applied to shows .
Check yourself
Lagrange’s form of the remainder, with only the information , proves the binomial series for every in .
False. For the best bound it gives grows without limit (first exercise). The series is still valid there; proving it needs Cauchy’s form, whose factor absorbs the dangerous dependence on .
The inequality (for ) holds exactly when:
Given , the inequality is , that is, , true for . It holds for negative too: that is the whole point of using Cauchy’s form.
What is the coefficient of in the expansion of ?
. So .
In the widget, Cauchy’s bound always lies above the true remainder . Why?
exactly, for Cauchy’s . Replacing the second factor by and the third by its largest possible value can only increase it. That is what “bound” means; the theorem says nothing about how close the bound is.
Hardy’s language
Summary
- For , Cauchy’s remainder is .
- For , the factor is at most , whatever and whatever the sign of ; this removes from the bound.
- The resulting tends to by a ratio argument, so on the whole interval .
- Lagrange’s form fails for not because the remainder is large but because its bound is too crude: the choice of remainder form matters.