1B - Changing base and finding exponents

1B - Changing base and finding exponents

How can you find the exponent in 3x=203^x=20 when 2020 is not a familiar power of 33? In 1B you will change logarithm bases, use reciprocal logarithms and calculate unknown exponents. You will also interpret those exponents as times in growth and decay models.

Changing the base

A logarithm can be calculated using another base, provided both logarithms in the resulting ratio use that same base.

For 3x=203^x=20, the exact answer is already named by x=log⁡320x=\log_3 20. A calculator may offer base-ten logarithms or natural logarithms directly. On calculators, the key labelled “log” usually means base 1010; the notation lg⁡x\lg x also means log⁡10x\log_{10}x. The key ln⁡x\ln x is the natural logarithm, whose base is the constant e≈2.71828e\approx2.71828.

The earlier standard graph used natural-number bases greater than 11. The algebraic laws and change-of-base formula also work for any real base that is positive and different from 11. A base of 11 fails because 1x1^x never changes; zero and negative bases do not define a real logarithm for every positive argument.

Let u=log⁡axu=\log_a x, so au=xa^u=x. Taking logarithms to another valid base bb and using the power law gives

ulog⁡ba=log⁡bx.u\log_b a=\log_b x.

Because a≠1a\ne1, the denominator log⁡ba\log_b a is nonzero, so we can divide by it.

Change of base

For x>0x>0, a>0a>0, b>0b>0, with a≠1a\ne1 and b≠1b\ne1,

log⁡ax=log⁡bxlog⁡ba.\log_a x=\frac{\log_b x}{\log_b a}.

The argument goes on top and the original base goes underneath. This formula is provided on the examination formulae sheet.

Either log⁡10\log_{10} or ln⁡\ln works. Mixing the two within one ratio does not apply the formula.

Worked example 1: using a calculator base

Question. Find log⁡820\log_8 20 to 33 significant figures.

Plan. Change both logarithms to base 1010, then round only the final value.

log⁡820=log⁡1020log⁡108use one common base=1.440642698…evaluate the whole ratio\begin{aligned} \log_8 20&=\frac{\log_{10}20}{\log_{10}8} &&\text{use one common base}\\ &=1.440642698\ldots &&\text{evaluate the whole ratio} \end{aligned}

Answer. 1.441.44, to 33 significant figures.

Check. 81<20<828^1<20<8^2, so the answer must lie between 11 and 22. Using the unrounded ratio in 8u8^u gives 2020.

Now change only the argument from 2020 to 44. A convenient base can give an exact answer without any calculator decimals.

Worked example 2: choosing a convenient base

Question. Find the exact value of log⁡84\log_8 4.

Plan. Use base 22, because both 44 and 88 are powers of 22.

log⁡84=log⁡24log⁡28=23.\log_8 4=\frac{\log_2 4}{\log_2 8}=\frac23.

Answer. 23\frac23.

Check. 82/3=(83)2=22=48^{2/3}=(\sqrt[3]{8})^2=2^2=4.

Changing base also lets you evaluate terms before combining an expression. For example, log⁡48=32\log_4 8=\frac32 and log⁡84=23\log_8 4=\frac23; their bases need not be forced into a sum law. Evaluate these exact values first, then use ordinary fraction arithmetic.

Reciprocal logarithms

Swapping a logarithm's base and argument gives its reciprocal when both numbers are valid bases.

For example, log⁡28=3\log_2 8=3 and log⁡82=13\log_8 2=\frac13. In change of base, choosing the new base to equal the argument gives the general reason:

log⁡ab=log⁡bblog⁡ba=1log⁡ba.\log_a b=\frac{\log_b b}{\log_b a}=\frac1{\log_b a}.

Reciprocal-base relation

For a>0a>0 and b>0b>0, with a≠1a\ne1 and b≠1b\ne1,

log⁡ab=1log⁡ba,(log⁡ab)(log⁡ba)=1.\log_a b=\frac1{\log_b a}, \qquad (\log_a b)(\log_b a)=1.

The reciprocal-base relation must be learnt; it is not separately provided on the examination formulae sheet.

An ordinary logarithm may have argument 11, but that argument cannot become a base. Thus log⁡21=0\log_2 1=0 is valid, while log⁡12\log_1 2 is not. You also cannot take the reciprocal of a zero logarithm.

Worked example 3: expressing a reciprocal logarithm

Question. Given p=log⁡37p=\log_3 7, express log⁡73\log_7 3 in terms of pp.

Plan. Swap the base and argument using the reciprocal relation.

log⁡73=1log⁡37=1p.\log_7 3=\frac1{\log_3 7}=\frac1p.

Answer. 1/p1/p.

Check. Since 3p=73^p=7, raising both sides to the power 1/p1/p gives 71/p=37^{1/p}=3, verifying the required logarithm.

Now let the unknown be the base itself. The reciprocal relation moves that unknown into the argument of a fixed-base logarithm.

Worked example 4: solving for a base

Question. Solve 2log⁡x3=12\log_x 3=1.

Plan. Require x>0x>0, x≠1x\ne1, isolate the logarithm and swap its base and argument.

log⁡x3=12divide by 2log⁡3x=2take the reciprocal logarithmx=32=9reverse the logarithm\begin{aligned} \log_x 3&=\frac12 &&\text{divide by }2\\ \log_3 x&=2 &&\text{take the reciprocal logarithm}\\ x&=3^2=9 &&\text{reverse the logarithm} \end{aligned}

Answer. x=9x=9.

Check. 91/2=39^{1/2}=3, so 2log⁡93=2×12=12\log_9 3=2\times\frac12=1. The base 99 satisfies both restrictions.

Reciprocal logarithms are different from logarithms of reciprocals: log⁡a(1/b)=−log⁡ab\log_a(1/b)=-\log_a b, whereas log⁡ba=1/log⁡ab\log_b a=1/\log_a b. The first changes the argument; the second swaps the base and argument.

Finding an unknown exponent

Taking logarithms turns an unknown exponent into a multiplier that algebra can isolate.

In ax=ba^x=b, the left side must be positive, so b>0b>0. For a>0a>0, a≠1a\ne1, the power law gives

log⁡10(ax)=log⁡10bxlog⁡10a=log⁡10bx=log⁡10blog⁡10a\begin{aligned} \log_{10}(a^x)&=\log_{10}b\\ x\log_{10}a&=\log_{10}b\\ x&=\frac{\log_{10}b}{\log_{10}a} \end{aligned}

This is also x=log⁡abx=\log_a b. The logarithm method is useful when the two sides cannot easily be written as powers of one familiar base.

Worked example 5: an exponent with no simple exact value

Question. Solve 3x=203^x=20. Give an exact logarithmic expression, then a value to 33 significant figures.

Plan. Take logarithms of both sides and use the power law to isolate xx.

xlog⁡103=log⁡1020bring the exponent outsidex=log⁡1020log⁡103divide by a nonzero logarithm=2.726833028…\begin{aligned} x\log_{10}3&=\log_{10}20 &&\text{bring the exponent outside}\\ x&=\frac{\log_{10}20}{\log_{10}3} &&\text{divide by a nonzero logarithm}\\ &=2.726833028\ldots \end{aligned}

Answer. Exactly x=log⁡1020log⁡103x=\frac{\log_{10}20}{\log_{10}3}; to 33 significant figures, x≈2.73x\approx2.73.

Check. 32=9<20<27=333^2=9<20<27=3^3, so 2<x<32<x<3. Substituting the unrounded value gives 3x=203^x=20.

Keep the same base and result, but replace the exponent by 2x−12x-1. The logarithm finds that entire exponent before you solve for xx.

Worked example 6: an expression in the exponent

Question. Solve 32x−1=203^{2x-1}=20. Give your answer to 33 significant figures.

Plan. Apply the previous calculation to 2x−12x-1, then solve the linear equation without rounding early.

(2x−1)log⁡103=log⁡1020take logarithms and use the power law2x−1=log⁡1020log⁡103x=12(1+log⁡1020log⁡103)=1.863416514…\begin{aligned} (2x-1)\log_{10}3&=\log_{10}20 &&\text{take logarithms and use the power law}\\ 2x-1&=\frac{\log_{10}20}{\log_{10}3}\\ x&=\frac12\left(1+\frac{\log_{10}20}{\log_{10}3}\right)\\ &=1.863416514\ldots \end{aligned}

Answer. x≈1.86x\approx1.86, to 33 significant figures.

Check. The unrounded value gives 2x−1=2.726833028…2x-1=2.726833028\ldots, so the original power equals 2020. Substitution of the rounded answer gives a value close to 2020, rather than exactly 2020.

Keep the exact ratio or the full calculator value until the last step. In particular, dividing by a logarithm is ordinary division: it is not permission to divide the logarithms' arguments.

Use a permitted scientific calculator for the numerical logarithms. Examination calculators must not have facilities for built-in symbolic algebra, symbolic differentiation or integration, or retrieval of text or formulae.

Exponential equations in context

In a growth or decay model, the unknown exponent often represents elapsed time or a number of repeated steps.

For a model Q=Q0atQ=Q_0a^t, Q0Q_0 is the initial amount. A growth rate of 8%8\% per time unit gives multiplier a=1.08a=1.08; a decrease of 15%15\% gives a=0.85a=0.85, because 85%85\% remains. First isolate the power by dividing by Q0Q_0, then use logarithms.

Worked example 7: finding a growth time

Question. A population is modelled by N=600(1.08)tN=600(1.08)^t, where tt is time in years. According to this model, when does the population reach 900900? Give your answer to 22 decimal places.

Plan. Divide by the initial population, then solve for the continuous time tt.

900=600(1.08)tset the model equal to the target(1.08)t=1.5divide by 600t=ln⁡1.5ln⁡1.08take logarithms=5.268446244…\begin{aligned} 900&=600(1.08)^t &&\text{set the model equal to the target}\\ (1.08)^t&=1.5 &&\text{divide by }600\\ t&=\frac{\ln1.5}{\ln1.08} &&\text{take logarithms}\\ &=5.268446244\ldots \end{aligned}

Answer. The model reaches 900900 after approximately 5.275.27 years.

Check. After 55 years the model gives about 881.60881.60, and after 66 years about 952.12952.12. The crossing time should lie between them.

Now use decay and restrict the observation times to whole hours. The logarithm gives the boundary time, but the required reading must occur at a permitted time after that boundary.

[DIAGRAM: asset_name: A decay crossing and the first whole-hour reading; asset_slug: decay-threshold; caption: The continuous model crosses 20 grams between hours 8 and 9, so hour 9 is the first whole-hour reading below the threshold; description: A decreasing curve M equals 80 times 0.85 to the t is plotted for nonnegative time t in hours. A horizontal dashed threshold is at M equals 20 grams. The continuous intersection occurs at about t equals 8.530, between marked whole-hour readings at hour 8 with mass about 21.799 grams and hour 9 with mass about 18.529 grams.]
The continuous model crosses 20 grams between hours 8 and 9, so hour 9 is the first whole-hour reading below the threshold

Worked example 8: interpreting a decay threshold

Question. A sample's mass is modelled by M=80(0.85)tM=80(0.85)^t grams after tt hours. Its mass is recorded at the end of each whole hour. Find the first reading below 2020 grams.

Plan. Find the time when the model equals 2020, then check the neighbouring whole-hour readings.

20=80(0.85)t(0.85)t=14t=ln⁡(1/4)ln⁡0.85=8.530048564…\begin{aligned} 20&=80(0.85)^t\\ (0.85)^t&=\frac14\\ t&=\frac{\ln(1/4)}{\ln0.85}\\ &=8.530048564\ldots \end{aligned}

The model is decreasing, so it is below 2020 after this crossing time. The first whole-hour time after 8.530…8.530\ldots is 99.

Answer. The first reading below 2020 grams is at the end of hour 99.

Check. 80(0.85)8≈21.79980(0.85)^8\approx21.799 grams, above 2020; 80(0.85)9≈18.52980(0.85)^9\approx18.529 grams, below 2020. Thus the previous reading does not yet meet the condition.

Both logarithms in the decay ratio are negative, so their quotient gives a positive time. Do not round a boundary time to the nearest integer when the question asks for the first reading that meets a condition. If a boundary happens at an exact integer time, the words “below” or “above” exclude equality.

Problem-solving

Solve the equality to locate a threshold, use growth or decay to choose the correct side, then check neighbouring permitted times. State the result in the context's units.

Apply the same reasoning to a value recorded once per year.

Quick Recap

Change of base makes any valid logarithm accessible through a consistent pair of calculator logarithms. Swapping base and argument gives a reciprocal; taking logarithms turns an unknown exponent into a multiplier. The opening equation has exact answer x=log⁡320=ln⁡20ln⁡3x=\log_3 20=\frac{\ln20}{\ln3}, approximately 2.732.73. In a model, interpret that exponent as a time or step number before deciding how to round.