Did you know that an equation containing both 2^x and 2^(−x) can often be solved using ordinary quadratic factorisation? Set t = 2^x, so t > 0 and 2^(−x) = 1/t. Multiply through by t, solve the resulting quadratic, discard any non-positive t-values and then recover x using powers or logarithms.
For example, 2^x + 2^(−x) = 5/2 becomes t + 1/t = 5/2 and then 2t² − 5t + 2 = 0. This factors as (2t − 1)(t − 2) = 0, giving t = 1/2 or 2, and therefore x = −1 or 1. The guide explains why positive substitution roots are essential, when the equation has no solutions, how inequalities work and how symmetry can check the result.
Recognise two powers that are reciprocals
The key structure is a^x + a^(−x), where a > 0 and a ≠ 1. By the index law a^(−x) = 1/a^x. These two expressions are connected rather than independent variables.
A problem that appears to mix two exponentials can therefore be reduced to an expression involving one positive quantity and its reciprocal. That is the same algebraic idea behind solving a quadratic in a newly introduced variable.
For example, when a = 2, write t = 2^x. The companion term is 2^(−x) = 1/t. There is no need to introduce a separate unknown for the second exponential.
This first structural decision is more important than fast calculator work. If a student treats 2^x and 2^(−x) as unrelated, an ordinary hidden quadratic remains invisible.
The substituted variable must be positive
For every real x, 2^x > 0. It can be a very small positive fraction or a very large positive number, but it never equals zero or becomes negative.
Therefore t = 2^x satisfies t > 0. This restriction stays with the problem even after the exponential notation disappears.
It is legitimate to multiply the equation by t because t cannot be zero. It is not legitimate to keep t = −2 as a final substitution result, since no real x could produce 2^x = −2.
The distinction resembles a quadratic in x², except that x² allows zero while 2^x does not. Whenever a substitution is introduced, identify the exact range of the expression it represents.
Solve the opening equation in full
Start with 2^x + 2^(−x) = 5/2 and set t = 2^x > 0. The equation becomes t + 1/t = 5/2.
Multiply by 2t, a positive nonzero quantity. This gives 2t² + 2 = 5t, or 2t² − 5t + 2 = 0.
Factorise as (2t − 1)(t − 2) = 0. Thus t = 1/2 or t = 2. Both values satisfy t > 0.
Return to x. From 2^x = 1/2 = 2^(−1), obtain x = −1. From 2^x = 2, obtain x = 1. The complete real solution set is {−1,1}.
Check both answers in the original exponential equation
At x = 1, the left side is 2^1 + 2^(−1) = 2 + 1/2 = 5/2.
At x = −1, the terms exchange places: 2^(−1) + 2^1 = 1/2 + 2 = 5/2.
Both values reproduce the original right-hand side exactly. The fact that the two solutions are opposite numbers is a consequence of the symmetry of this particular sum.
Checking the original exponential equation is more informative than checking only the quadratic in t. It confirms that the return from t to x used the correct exponential base and did not lose a sign or introduce a forbidden substitution value.
Why two answers often come in opposite pairs
Define F(x) = a^x + a^(−x), with a > 0 and a ≠ 1. Then F(−x) = a^(−x) + a^x = F(x).
Thus F is an even function. Its graph is symmetric about the y-axis, and every nonzero solution of F(x) = K comes with the partner −x.
In the opening example, x = 1 and x = −1 are the two answers. The equations are symmetric because reversing x swaps the two positive exponential terms without changing their sum.
This gives an effective completeness check. If an unrestricted equation of this exact form produces only one nonzero solution, investigate whether the corresponding opposite solution was accidentally omitted.
The function has a minimum value of two
Put t = a^x > 0. The expression F(x) = t + 1/t is always at least 2.
One proof follows from (t − 1)² ≥ 0. Expanding gives t² − 2t + 1 ≥ 0. Divide by t > 0 to obtain t + 1/t ≥ 2.
Equality holds only when t = 1. For a valid base a ≠ 1, a^x = 1 occurs at x = 0.
Therefore the entire real range of a^x + a^(−x) is [2,∞). This determines whether the equation F(x) = K can have real solutions before any quadratic calculation.
Classify the equation using the target value
Consider a^x + a^(−x) = K for a > 0, a ≠ 1. If K < 2, no real solution exists because the left side never falls below 2.
If K = 2, the only solution is x = 0, where both exponential terms equal one. If K > 2, there are exactly two real solutions, opposite in sign.
The quadratic t² − Kt + 1 = 0 gives the same classification. Its discriminant is K² − 4, but the positivity restriction on t is still essential when interpreting algebraic roots.
The inequality F(x) ≥ 2 is the clearest reason for the full solution count. It avoids a misleading discriminant-only argument when K is negative.
Why a positive discriminant is not enough
Suppose K = −3. The transformed equation is t² + 3t + 1 = 0, whose discriminant is 9 − 4 = 5 > 0.
It has two real t-roots, namely (−3 ± √5)/2. Both are negative because √5 < 3.
But t = a^x must be positive. Therefore neither transformed root corresponds to a real x, and the original equation has no real solution.
This is a useful counterexample to the shortcut “positive discriminant means two exponential solutions”. The quadratic discriminant describes reality of t; the substitution’s range determines which t-values can come from a real exponent.
The boundary case K = 2 gives one solution
Solve 2^x + 2^(−x) = 2. Put t = 2^x > 0, multiply by t and rearrange to t² − 2t + 1 = 0.
This is (t − 1)² = 0, so t = 1 is a repeated quadratic root. Returning to x gives 2^x = 1 and hence x = 0.
There is one distinct real x-answer, not two separate answers both equal to zero. The repeated factor indicates the minimum level where the horizontal line y = 2 touches the even exponential-sum graph.
This interpretation agrees with the positive-square inequality and provides a graphical explanation of the boundary solution count.
An example that needs logarithms
Solve 2^x + 2^(−x) = 3. Set t = 2^x to obtain t + 1/t = 3.
Multiplying through gives t² − 3t + 1 = 0. Use the quadratic formula: t = (3 ± √5)/2. Both values are positive.
Therefore x = log₂[(3 + √5)/2] or x = log₂[(3 − √5)/2]. These are exact expressions.
The two t-values multiply to one, so their base-two logarithms are negatives of one another. Thus the solutions form an opposite pair, even though neither is an ordinary integer exponent.
Write the logarithmic answers more compactly
For K > 2, define T = (K + √(K² − 4))/2, which is greater than one.
The other positive quadratic root is 1/T, since the product of the roots of t² − Kt + 1 is one.
Hence the general solutions of a^x + a^(−x) = K can be expressed as x = ±log_a T, with a > 0 and a ≠ 1. The signs may appear in the opposite numerical order when the base is below one, but the set remains symmetric.
This compact form is optional. A school solution that presents the two logarithms separately is equally correct, provided both are real and the base is valid.
A base below one does not change the existence condition
Take a = 1/2. Then a^x + a^(−x) = (1/2)^x + (1/2)^(−x) = 2^(−x) + 2^x.
This is exactly the same function as 2^x + 2^(−x), because the two summands have exchanged places.
Therefore the minimum is still 2, the threshold classification is unchanged and the equation equal to 5/2 still has solutions x = ±1.
The base affects which transformed root corresponds to which sign of x, but not the set of values attainable by the symmetric reciprocal sum. This is a useful check on the method’s generality.
An ordinary exponential model is a different structure
The equation 2^x = 5/2 contains only one exponential term. It can be solved directly by x = log₂(5/2), without a quadratic substitution.
The equation 2^x + 2^(−x) = 5/2 contains two related terms and is not equivalent to 2^(x−x) = 5/2. Index laws do not allow adding exponential expressions by adding their exponents.
Specifically, a^p a^q = a^(p+q) applies to multiplication, not addition. The false rule a^p + a^q = a^(p+q) fails even for simple numerical powers.
Recognising when terms are multiplied, added or composed prevents the common algebraic mistake that turns a legitimate reciprocal-exponential problem into an unrelated equation.
A weighted reciprocal exponential equation
Consider 3·2^x + 12·2^(−x) = 12. Let t = 2^x > 0. The equation becomes 3t + 12/t = 12.
Multiply by t to obtain 3t² − 12t + 12 = 0. Divide by three: t² − 4t + 4 = 0, or (t − 2)² = 0.
The unique positive root is t = 2, so x = 1. This is an example where the coefficients of the two reciprocal terms are unequal and the minimum does not occur at x = 0.
The same substitution works, but the symmetry is now around a shifted centre rather than the y-axis. We should not expect answers ±x without first checking the weights.
Find the minimum of a weighted sum
For A > 0 and B > 0, consider A a^x + B a^(−x). Set t = a^x > 0, giving At + B/t.
By the arithmetic–geometric mean inequality, At + B/t ≥ 2√(AB), with equality when At = B/t.
That equality gives t² = B/A and t = √(B/A). Therefore the minimum value is 2√(AB), attained at x = log_a√(B/A).
For A = 3, B = 12 and a = 2, the minimum is 2√36 = 12, achieved at x = log₂2 = 1. This matches the repeated-root example exactly.
When the right-hand side exceeds the weighted minimum
Suppose 3·2^x + 12·2^(−x) = 15. The minimum is 12, so two real solutions are expected.
Set t = 2^x > 0. We obtain 3t + 12/t = 15, or 3t² − 15t + 12 = 0.
Divide by three: t² − 5t + 4 = (t − 1)(t − 4) = 0. Both t = 1 and t = 4 are positive.
Thus x = 0 or x = 2. Their midpoint is x = 1, the location of the weighted minimum. The solutions are symmetric about that shifted centre, not about zero.
A difference of reciprocal exponentials
Now solve 2^x − 2^(−x) = 3/2. Substitute t = 2^x > 0 to obtain t − 1/t = 3/2.
Multiply by 2t: 2t² − 2 = 3t. Rearrange to 2t² − 3t − 2 = 0.
Factorise as (2t + 1)(t − 2) = 0. The transformed roots are t = −1/2 and t = 2, but the negative root is inadmissible.
Therefore the only real solution is x = 1. Unlike the positive sum, the difference t − 1/t is strictly increasing in t and can take every real value, giving one solution for each real right-hand side.
Why the difference has one root rather than an opposite pair
Define G(x) = 2^x − 2^(−x). Replacing x by −x gives G(−x) = 2^(−x) − 2^x = −G(x).
Thus G is an odd function, not an even function. A nonzero target value typically produces one x, while the opposite target produces the opposite x.
For example, G(1) = 3/2 and G(−1) = −3/2. They do not both solve G(x) = 3/2.
This contrasts with the reciprocal sum, where opposite inputs give the same output. Recognising even and odd symmetry provides an independent qualitative check on solution counts.
Use the method for an inequality
Solve 2^x + 2^(−x) ≤ 5/2. Set t = 2^x > 0 and multiply through by 2t.
The inequality becomes 2t² − 5t + 2 ≤ 0, or (2t − 1)(t − 2) ≤ 0.
Thus 1/2 ≤ t ≤ 2. Because base 2 is increasing, 1/2 ≤ 2^x ≤ 2 translates to −1 ≤ x ≤ 1.
The final interval includes both boundaries because equality holds at x = −1 and x = 1. The minimum occurs at zero, so the middle interval is exactly where the symmetric exponential sum remains below the chosen threshold.
The opposite inequality gives two outside intervals
Consider 2^x + 2^(−x) ≥ 5/2. The same substitution gives (2t − 1)(t − 2) ≥ 0.
For t > 0, the solution is 0 < t ≤ 1/2 or t ≥ 2. Converting back through the increasing function t = 2^x gives x ≤ −1 or x ≥ 1.
This is consistent with the graph: the reciprocal sum has its minimum at x = 0 and grows as the magnitude |x| grows.
An inequality can therefore yield either a centred interval or two outer intervals depending on whether the target condition is ≤ or ≥. Translate the complete t-solution set back to x rather than taking logarithms of only one boundary.
A common inequality trap: multiplying by t is safe here
When solving t + 1/t ≤ K, multiplying by t preserves the inequality direction because t = a^x is strictly positive.
This differs from multiplying a general rational inequality by x − 2, whose sign depends on x and could reverse the inequality.
The substitution supplies a known positivity condition that justifies the multiplication. A student should state t > 0 at the start, not merely rely on an unstated assumption.
Careful sign reasoning makes the quadratic inequality legitimate and ensures that its interval solution can be transferred back to exponential inputs without an unnoticed forbidden denominator.
Base monotonicity matters when returning from t to x
If t = 2^x, then t increases as x increases. Inequalities involving t can be converted to x without reversing their direction.
If t = (1/2)^x, the substituted quantity decreases as x increases, so converting a t-inequality into an x-inequality reverses the order.
For example, 1/2 ≤ (1/2)^x ≤ 2 gives −1 ≤ x ≤ 1, but the individual endpoint comparisons reverse directions before being combined.
The same final symmetric interval arises for this particular reciprocal sum. Nonetheless the reasoning should reflect the chosen exponential base’s monotonicity rather than copying an increasing-base rule automatically.
Do not confuse algebraic equality with parameter uniqueness
The equation A a^x + B a^(−x) = C with A,B > 0 may have zero, one or two real solutions depending on C relative to 2√(AB).
If C is below the minimum, none exist. At the minimum, the transformed quadratic has a repeated positive root and gives one x. Above the minimum, two positive t-roots give two x-values.
This classification comes from the positive-variable quadratic and the weighted AM–GM bound. A discriminant calculation can confirm it, since the quadratic At² − Ct + B has discriminant C² − 4AB.
But for negative C, a discriminant may be non-negative while the quadratic roots are non-positive. The positivity of t still has to be checked.
A domain check for logarithms in the final step
To recover x from a^x = t, write x = log_a t. The logarithm requires t > 0, base a > 0 and a ≠ 1.
These are precisely the assumptions already present in the exponential model. A negative quadratic root is therefore not a permissible logarithm argument.
For example, if a transformed equation yields t = −1/2, writing log₂(−1/2) as a real answer is invalid. The proper response is to reject that t-root.
This shows why the substituted-variable restriction is more than a formal line. It protects the real domain of the operation required to return to the original unknown x.
A compact exam solution routine
First look for a repeated base with exponents x and −x. Set t = a^x and state t > 0, so a^(−x) = 1/t.
Second substitute and multiply by t to obtain a quadratic equation or inequality. Solve it with factorisation or the quadratic formula.
Third discard every non-positive t-value. Fourth recover x through exact powers or logarithms, reversing inequalities where necessary for a base between zero and one.
Finally verify each candidate in the original equation. For a positive reciprocal sum, use its minimum and symmetry as reasonableness checks. This sequence is often shorter than any direct manipulation of the original exponentials.
Practice tasks with checked answers
Question one: solve 2^x + 2^(−x) = 5/2. Answer: x = −1 or 1.
Question two: solve 2^x + 2^(−x) = 2. Answer: x = 0 only.
Question three: solve 2^x + 2^(−x) = 3. Answer: x = ±log₂[(3 + √5)/2].
Question four: solve 2^x − 2^(−x) = 3/2. Answer: x = 1.
Question five: solve 3·2^x + 12·2^(−x) = 12. Answer: x = 1 only.
Question six: solve 3·2^x + 12·2^(−x) = 15. Answer: x = 0 or 2.
Question seven: solve 2^x + 2^(−x) ≤ 5/2. Answer: −1 ≤ x ≤ 1.
For every task, show the positive substitution and verify whether each quadratic root can represent a real exponential value.
What parents and tutors should diagnose
A student who recognises a quadratic in t but keeps a negative t-root needs practice with the range of exponential functions. Another may know the range but multiply 1/t incorrectly and require a short rational-algebra repair.
A third may obtain valid positive t-values but fail to convert them back to x using logarithms. A fourth may misread ≤ versus ≥ and shade the wrong quadratic sign interval.
These are distinct prerequisites. The tutor should identify the first invalid transformation, correct that skill and then test it on a new reciprocal-exponential example with changed coefficients.
Independent transfer matters more than copying the same pair of answers. The student should know why the positive sum has a minimum and why the negative transformed root in a difference equation is not a real exponential solution.
Syllabus context and further reading
Reciprocal-exponential equations connect indices, substitution, quadratic factorisation, logarithms, inequalities and function range. More advanced weighted and symmetric versions can serve as enrichment depending on the student’s examination pathway.
Consult the appropriate Singapore SEAB listings for 2026 GCE O-Level Additional Mathematics, 2027 SEC G3 or 2027 SEC G2, as relevant.
Related guides include How to Spot a Hidden Quadratic in an Exponential Equation, How to Solve a Reciprocal Quartic Using x + 1/x and Why a Logarithmic Inequality Reverses When Its Base Is Below One.
Explore the Additional Mathematics archive, or read the eduKate small-group learning approach and contact eduKate Singapore for tutorial suitability. Properly taught kids shine a bright light into the future.