11

In this problem, I guess b is larger, but not know how to prove it without going to lengthy calculations. It is highly appreciated if anyone can give me a help.

Which number is larger

$$\begin{align} &\textrm{(a)}\quad 7^{94} &\quad\textrm{(b)}\quad 9^{91} \end{align}$$

Michael Lugo
  • 22,354
  • 3
    The quick way to do it, especially if it's multiple choice, is comparing the progressions $7 \cdot 7 \cdot 7 \cdot 7 \cdot 7 \cdot 7....$ with $9 \cdot 9 \cdot 9 \cdot 9....$.

    Intuitively, considering the "growth rate", it is clear that for large $n$ $$\prod_{i=1}^{n-3} 9> \prod_{i=1}^{n} 7$$

    – MathematicsStudent1122 May 16 '16 at 23:49
  • 5
    FSK, could you please add something about the fact that you weren't asking for the numerical values, but some quick way? I cannot help but feel bad about the second most upvoted answer. – Vincenzo Oliva May 17 '16 at 09:31

9 Answers9

71

The first is $7^{91}\times 343$. The second is $7^{91}\times(9/7)^{91}$. Since $\frac{9^3}{7^3}\gt 2$, it follows that $(9/7)^{91}$ is much much bigger than $343$.

André Nicolas
  • 507,029
  • I would not assume that everybody knows that $(9/7)^3>2$. I think most people could not check that without pen and paper. –  May 17 '16 at 13:26
  • 2
    @Pakk: True, I used a fairly tight estimate. But there is a tremendous amount of slack here, as long as we know that a smallish power of $9/7$ is greater than $2$, the inequality follows. If we use $\frac{9^3}{7^3}\gt 2$, then $(9/7)^{91}\gt 2^{30}\gt 1000^3$. – André Nicolas May 17 '16 at 16:02
  • @Pakk: My answer gives a similar approach using a looser estimate that's trivial to see, if you prefer zero explicit computation. – user21820 May 17 '16 at 16:05
  • $9^2/7^2=81/49>80/50=1.6 >\sqrt 2$ so $9^4/7^4>2$. – DanielWainfleet May 17 '16 at 19:04
  • 1
    @Pakk I tried to see it as $9^3 > 2 \cdot 7^3$, which is $729 > 2 \cdot 343=686$ – Cookie May 17 '16 at 23:02
  • $\left(\frac97\right)^3 > 1.26^3 = \left(\frac54 + \frac{1}{100}\right)^3 > \left(\frac54\right)^3 + 3\left(\frac54\right)^2\frac{1}{100} = 2$. – Théophile Aug 31 '22 at 15:33
  • 1
    @DanielWainfleet Alternatively: $\left(\frac97\right)^3=\frac97\left(\frac97\right)^2=\frac97\cdot\frac{81}{49}>\frac97\cdot\frac85=\frac{72}{35}>2$. – Théophile Aug 31 '22 at 17:51
33

$$7^{94} = 7^{10} 49 ^{42} < 7^{10} 54 ^{42} = 7^{10} 8^{14} 9^{63} < 9^{10} 9^{14} 9^{63} = 9^{87} < 9^{91} $$

9

André already nailed it, but here's another way. The following inequalities are equivalent:\begin{align}7^{94} &< (7+2)^{94-3} \\ 9^3&<(1+2/7)^{94} \\ 3\log3&<47\log(1+2/7),\end{align} and by the Maclaurin expansion of $\log(1+x)$, the latter follows from \begin{align}3\log3&<94\left(\frac17-\frac1{49}\right) \\ \log3<3&<2\cdot\frac{94}{49}.\end{align}

  • 7
    I guess the downvoter upvoted guestDiego's answer. – Vincenzo Oliva May 17 '16 at 09:27
  • 1
    This should be the preferred answer to the OPs question. The calculator based answer does not constitute an appropriate answer, despite 16 voters opinions. Well reasoned Vincenzo. –  May 27 '16 at 08:39
  • Quite incredible - I stand by my comment to you in May, this is excellent reasoning. Barry's answer is also suitable, but i stick to my vote! All the best, Sir. –  Aug 05 '16 at 08:51
8

$Log(9^{91})=91\cdot Log(9)=86.836068359$

$Log(7^{94})=94\cdot Log(7)=79.4392157613$.

Hence $ 9^{91}$ is bigger.

guestDiego
  • 4,006
  • 5
    This should be the selected answer. – Brian Risk May 17 '16 at 01:32
  • @BrianRisk did you upvote it? it's a fine answer but I don't agree it's the best answer. – djechlin May 17 '16 at 02:24
  • 28
    This doesn't aid the asker in finding a way to obtain solutions for such problems without lengthy calculations (which is what logs boil down to). "Plug this into a calculator thusly" << using simple concepts to give an elegant and concise answer which doesn't need any hard work at all. – Nij May 17 '16 at 03:33
  • 13
    Holy crap, MSE has become embarrassing. – Vincenzo Oliva May 17 '16 at 09:14
  • 2
    This is the best answer, raw computational power baby. just plug it in, no magic needed –  May 17 '16 at 11:28
  • 7
    @Loffen No insight, either. – G. Bach May 17 '16 at 11:41
  • 5
    I beg to disagree. The logarithm is an immensely powerful tool for calculations that involve powers. (My stats background may be showing through.) Showing how quickly this question can be answered by logging both expressions and looking at some log tables is a Good Answer. (Yes, I also upvoted André's answer.) – Stephan Kolassa May 17 '16 at 11:46
  • 2
    @StephanKolassa: And where are you getting your log-tables from? By your own reasoning, it is better to put the question straight to a calculator and get $2.75 \cdot 10^{79} < 6.86 \cdot 10^{86}$. – user21820 May 17 '16 at 12:08
  • 3
    @user21820: yes, of course log tables involve a lot of computation (and smart computation at that). And yes, of course this can be answered without logs. Nevertheless, I'll upvote an answer that shows the power of logs to answer this. – Stephan Kolassa May 17 '16 at 12:17
  • 3
    @StephanKolassa: You're missing the point. By your reasoning you should upvote an answer that shows the power of calculators to answer the question....... – user21820 May 17 '16 at 12:18
  • 3
    I suppose you'd use a calculator to evaluate those logs? Or maybe R or matlab or something like that on your laptop. But then again why not just do (7^94)>(9^91) [1] FALSE :) – curious_cat May 17 '16 at 12:54
  • 11
    This answer should simply be deleted, it's an embarassment for MSE. It currently has 32 upvotes, which tells much about the mathematical level of some MSE users. – Alex M. May 17 '16 at 13:05
  • Some of the following answer are very elegant (see Pak, or

    user21820 and others) The more they are connected to the particular numbers 7,9,94,91, the more elagant they are... Of course the solution with the logarithm is of a different kind. Using the Logarithm of 7 and 9 is not an unfair use of terribly powerful calculus devices (with all due respect to Neper)... Maybe the commentator above would prefer a further 'inexpensive' way to estimate the Log(9) in comparison to Log(7). This could be done... What is truly emabrassing is that one could find this answer even 'embarassing'.

    – guestDiego May 17 '16 at 17:56
  • 1
    @curious_cat. Simply because five centuries ago they still would have approved my answer! Is it the case to let you note what is the negative aspect of solutions which works ONLY for the particular numbers 7,9,91,94? – guestDiego May 17 '16 at 18:01
  • Of course, my prevoius comment was in no way a criticism of the other very smart solutions. This clarification is addressed only to the most polemic guys here! – guestDiego May 17 '16 at 18:20
  • @guestDiego Actually, I loved your solution. It seems very pragmatic. – curious_cat May 18 '16 at 03:21
  • 1
    I don't understand what is being argued about here - what is embarrassing and why? – Chill2Macht May 22 '16 at 01:37
  • 2
    Here's why: nothing about this answer is mathematical. It's literally telling the OP to put it into a calculator. Logs can be VERY useful (and can be used in many of these types of problems), but this answer just gave a computationally cheaper way to put these expressions in to the calculator (in an era that doesn't see the original computation as particularly expensive). – rnrstopstraffic May 26 '16 at 05:09
  • So the answer could have been valid in a previous era and it is no more now! If the question was about $7^{987}$ and $9^{970}$, the answer could have been considered valid still in this era? Without saying that most of the calculators give a result to $7^{94}$ just by using logarithms. – guestDiego May 26 '16 at 23:52
  • 1
    No, it isn't an illuminating answer in any era. It's just the icing on the cake that there isn't even a need to be economical with that calculation anymore. – rnrstopstraffic May 27 '16 at 04:47
  • 1
    May be it is not illuminating. In fact, some of the solutions below are more illuminating, just for the current example of course. If I added a low order Taylor expansion to calculate Log(7) and Log(9) without calculator or tables, would the method meet your criterion of adequacy? – guestDiego May 27 '16 at 13:12
6

$9^{91} \div 7^{94} = (\frac97)^{94} \div 9^3 > (1+\frac27)^{7 \times 13} \div 3^6 > (1+2)^{13} \div 3^6 = 3^7$ which is way bigger than $1$.

user21820
  • 57,693
  • 9
  • 98
  • 256
4

First notice that $3^9 = 19683 > 16807 = 7^5$ (this can be calculated manually). Thus $9^9 = (3^2)^9 = 3^{18} = (3^9)^2 > (7^5)^2 = 7^{10}$. It follows that $9^{91} > 9^{90} = (9^9)^{10} > (7^{10})^{10} = 7^{100} > 7^{94}$.

  • What... that's not a Suslin tree! :-) – Asaf Karagila May 17 '16 at 21:05
  • @AsafKaragila Not only is it not a Souslin tree, but the numbers are finite. When was the last time either of us solved a problem involving finite numbers? At least this way it doesn't require AC. – Ari Brodsky May 30 '16 at 23:04
  • Ha! Well, finite things are always harder than choiceless things, which are inherently harder than choice-y things anyway. :-P – Asaf Karagila May 30 '16 at 23:06
3

I voted for André's answer, but here's another approach, using a different bit of maths.

Note that $7^{94} = 7^3 \times 7^{91}$. $9^{91} = (7 \times \frac{9}{7})^{91}$, where $\alpha = \frac{9}{7} = 1 + \frac{2}{7} > 1$. So $$ \frac{7^{94}}{9^{91}} = \frac{7^3}{\alpha^{91}}. $$

What do we make of $\frac{7^3}{\alpha^{91}}$? Well, $7^3 = 49 \times 7 = 343$. Using the binomial theorem, and observing that positive ratios always diminish when the numerators (resp. denominators) are decreased (resp. increased), \begin{align*} \alpha^{91} &= \left(1 + \frac{2}{7}\right)^{91} \\ &> 1 + \frac{91}{1!} \times \frac{2}{7} + \frac{91 \times 90}{2!} \times \frac{2^2}{7^2} \\ &\quad= 1 + \frac{182}{7} + \frac{8190 \times 4}{98} \\ &\quad> 1 + 25 + \frac{4 \times 80 \times 100}{100} \\ &\qquad= 1 + 25 + 320 \\ &\qquad= 346 \\ &\qquad> 343. \end{align*} So $\alpha^{91} > 7^3$ and thus $9^{91} > 7^{94}$.

Wooster
  • 98
  • Okay, so I just spotted that I made a bit of a meal of the cancellation up top. I also wanted to add that I am positively teeming with news about the binomial theorem. ahem – Wooster May 17 '16 at 12:50
3

$$\begin{align} \left(9\over7\right)^3={729\over343}\gt2 &\implies\left(9\over7\right)^{15}\gt2^5\gt7\\ &\implies9^{15}\gt7^{16}\\ &\implies9^{90}\gt7^{96}\\ &\implies9^{91}\gt7^{94} \end{align}$$

Another proof, using the general inequality $\ln(1-x)\lt-x$ for $0\lt x\lt1$ and the numerical inequality $7\lt2^3\lt e^3$:

$$\ln(7^{94}/9^{91})=3\ln7+91\ln\left(1-{2\over9}\right)\lt3\cdot3-90\cdot{2\over9}=9-20\lt0$$

A third proof, presented in easily checkable, but almost completely unmotivated form:

$$\begin{align} 2^{47}7^{94} &=98^{47}\\ &\lt100^{47}\\ &=10^3\cdot10^3\cdot10^{88}\\ &\lt2^{10}\cdot2^{10}\cdot10000^{22}\\ &\lt20000^{22}\\ &\lt160^{44}\\ &\lt36(162)^{45}\\ &=2^{47}9^{91} \end{align}$$

And one more proof, this one based on the fact that $2^{10}=1024\gt1000=10^3$, which implies $\log2\gt0.3$ (where "log" here means log base $10$):

$$94\log7=47\log49\lt47(\log100-\log2)\lt47\cdot1.7=79.9$$ whereas

$$91\log9\gt91\log8=273\log2\gt273\cdot0.3=81.9$$

Full disclosure: I used a calculator for $47\cdot1.7=79.9$. But everything else I did by hand.

Added 5/25/15: At another question, proofs are given of the inequality $7^{19}\lt9^{17}$. (See in particular joriki's answer there.) It follows that

$$7^{94}\lt7^{95}\lt9^{85}\lt9^{91}$$

Barry Cipra
  • 79,832
0

Here are a couple of equivalent statements: \begin{eqnarray*} 7^{94}&<&9^{91}\\ 7^3&<&\left(\frac97\right)^{91}\\ 343&<&\left(1+\frac27\right)^{91} \end{eqnarray*} We can expand the latter expression using the binomial theorem to get $$\left(1+\frac27\right)^{91}>1+\binom{91}{1}\times\frac27+\binom{91}{2}\times\left(\frac27\right)^2=1+26+\frac{90\times26}{7}>343.$$

Servaes
  • 63,261
  • 7
  • 75
  • 163