What Actually Goes on a Calculus 2 Final Cheat Sheet
Most students walk into their final with a three-page scribble of half-remembered formulas and a prayer. It rarely helps. The ones that do work are usually the ones that map techniques to problem types rather than just listing derivatives and integrals side by side. I spent years watching people fail because they memorized the wrong relationship between two methods, or because they had no way to quickly decide which convergence test to reach for when a series looked deceptively simple. The hardest part isn't collecting the formulas. It's organizing them so your brain can access the right one under time pressure. That is where a properly built Calculus 2 Final Cheat Sheet earns its weight in the room.
Calculus 2 Final Cheat Sheet
Build yours around decision trees, not just reference tables. On the front side, put the integration techniques with clear triggers. If the problem has a polynomial times an exponential, that is integration by parts. If you see a quadratic denominator that does not factor over the integers, partial fractions is probably your move, but only after you check whether the numerator degree is lower. If the integrand contains sqrt(a^2 - x^2), that screams trig substitution with x = a sin(theta). Write those triggers down next to the formulas, not separately. I once watched a student lose forty minutes on a midterm because the problem was int(x^2 / (x^2 + 4x + 8)) dx. They immediately reached for trig substitution because of the square-looking denominator, then got stuck rewriting theta back into x. The faster path was completing the square first, then splitting the numerator into 2x + 8 minus 8, which broke it into a straightforward log substitution plus a standard arctan form. If your cheat sheet includes a note that says complete the square before reaching for trig sub when the denominator is an irreducible quadratic, you save yourself that kind of detour.
Integration Techniques and When They Fail
u-substitution works when you can spot a function and its derivative sitting next to each other, even if a constant factor is hiding in front. The trap is chasing substitutions that look promising but leave you with a worse integral. If your u-sub makes the exponent more complicated, backtrack immediately. Integration by parts follows the LIATE rule as a starting heuristic: Logarithmic, Inverse trig, Algebraic, Trig, Exponential. Pick u from the earlier category. This breaks down when you have a product of two exponentials, or when repeated by parts cycles back to the original integral without a clean solve. In that cycling case, bring the integral back to the left side and solve algebraically, but only after verifying the antiderivative exists in closed form. Partial fractions requires the rational function to be proper first. If the numerator degree is greater or equal, do polynomial long division before anything else. I have seen people skip that step repeatedly and then argue the method should work. It will not. Once proper, factor the denominator completely over the reals. Repeated linear factors generate terms like A/(x-a) + B/(x-a)^2. Irreducible quadratics generate (Ax+B)/(quadratic). The coefficients come from either the cover-up method for simple linear factors or from solving a small linear system for repeated or quadratic cases.
Get the Full Details
Trig substitution has three standard forms. For sqrt(a^2 - x^2), use x = a sin(theta). For sqrt(a^2 + x^2), use x = a tan(theta). For sqrt(x^2 - a^2), use x = a sec(theta). After integrating in theta, draw a reference triangle to convert back to x. The triangle method is faster and less error-prone than trying to invert the substitution algebraically, especially when the answer involves multiple inverse trig terms.
Improper Integrals and Convergence Checks
Improper integrals fall into two categories: infinite bounds or unbounded integrands. For infinite bounds, compare against 1/x^p. The integral of 1/x^p from 1 to infinity converges only when p is strictly greater than 1. For unbounded integrands near a vertical asymptote at c, compare against 1/(x-c)^p. The integral converges only when p is strictly less than 1. These comparisons are the backbone of the direct comparison test and the limit comparison test. A detail people miss is that the comparison test requires non-negative integrands on the interval. If your function oscillates, you need absolute convergence first before you can use standard comparison arguments. The absolute convergence test here means checking whether the integral of the absolute value converges. If it does, the original converges. If it does not, the original may still converge conditionally, and you need to analyze it differently, usually through alternating series logic or Dirichlet-type arguments. I ran into a case recently where the integrand was cos(x)/sqrt(x) from 1 to infinity. At first glance it looks like a direct comparison candidate, but cos(x) changes sign, so absolute comparison fails. The fix was rewriting the integral as an alternating-type form and applying Dirichlet's test for integrals: 1/sqrt(x) decreases monotonically to zero, and the integral of cos(x) over any interval is bounded. That combination guarantees convergence without computing the value.
Series Convergence Tests as a Ranked List
When you face a series, run through these tests in order until one applies. First, check the nth-term divergence test. If the limit of a_n is not zero, the series diverges. This is the only test that gives a definitive divergence answer on its own. Do not call it the ratio test or anything else. It is its own thing, and it catches more divergent series than most students realize. Next, identify whether the series is geometric, p-series, telescoping, or alternating. Geometric series converge when |r| is strictly less than 1, and the sum is a_1 / (1 - r). P-series converge when p is strictly greater than 1. Telescoping series collapse when you write out the partial sums, and the answer is often just the first term minus the limit of the tail.
For general positive-term series, the ratio test and root test are your heaviest tools. The ratio test computes lim |a_{n+1}/a_n|. If the limit is less than 1, convergent. Greater than 1, divergent. Equal to 1, inconclusive. The root test computes lim |a_n|^(1/n) and follows the same threshold rules. Use the ratio test when factorials and exponentials dominate. Use the root test when everything is raised to the nth power. The limit comparison test is the workhorse for rational functions and algebraic expressions. Compare a_n against 1/n^p where p is the difference in degrees between denominator and numerator. This test is more reliable than the direct comparison test because you do not need to establish inequalities across the entire tail. You only need the limit to exist and be finite and positive. The integral test connects series to improper integrals. It applies when f(x) is positive, continuous, and decreasing for x greater than or equal to some N. The series and the integral share convergence behavior. The practical tradeoff is that evaluating the integral can be harder than testing the series directly, so use this only when the integral is familiar or easier than manipulating the series terms.
Power Series and Taylor Expansions
Find the radius of convergence using the ratio test on the general term. The resulting limit in terms of |x - c| gives you an inequality that solves directly to R. Always check the endpoints separately. The ratio test tells you nothing about x = c plus or minus R, and endpoint behavior can change the interval from open to half-open to closed. Common Taylor series to memorize are e^x, sin(x), cos(x), 1/(1-x), and ln(1+x). Any other expansion you need can usually be derived by substitution, differentiation, or integration from one of these five. For example, the series for ln(x) centered at 1 comes from integrating the geometric series for 1/t term by term. The series for arcsin(x) comes from integrating the binomial expansion of (1-x^2)^(-1/2), which is messier but follows the same pattern. A practical edge case is multiplying two known series together. Do not multiply term by term like polynomials unless you actually need a Cauchy product up to a certain order. If you only need the first three nonzero terms, expand each series to sufficient order, distribute carefully, and collect like powers. Spending ten minutes writing out every coefficient is almost never worth it on a timed exam. Stop when the next power you would generate exceeds what the question asks for.
Operations on Convergent Series
Adding or subtracting two convergent series preserves convergence. Multiplying a convergent series by a constant preserves convergence. These are straightforward, but people forget that term-by-term differentiation and integration of power series preserve the radius of convergence while possibly changing endpoint behavior. The new series has the same R, but you must recheck the endpoints from scratch. The interval of convergence for a differentiated series is always at least as large as the original, but can shrink at the endpoints. The same is true for an integrated series, except integration tends to improve endpoint behavior rather than worsen it. I once had a student assume the integrated series inherited the original's divergence at an endpoint, when in fact the extra factor of 1/n from integration made that endpoint convergent. Always re-evaluate endpoints after any operation.
Numerical Integration Limits
Numerical methods on a final usually mean Riemann sums, the trapezoidal rule, or Simpson's rule. Remember that the trapezoidal rule overestimates when the function is concave up and underestimates when concave down. Simpson's rule requires an even number of subintervals and gives exact results for polynomials up to degree 3. Both error bounds exist, but you rarely need to compute them from scratch. Simpson's error bound involves the fourth derivative maximum, which means if the fourth derivative is large or unbounded on the interval, Simpson's rule may converge slowly or not at all. I encountered a problem where Simpson's rule was requested on an integral containing a logarithmic singularity at an endpoint. The theoretical error bound assumed bounded fourth derivatives, which did not hold here. The numeric output looked plausible at first glance, but refining the mesh revealed systematic drift. The workaround was splitting the interval away from the singularity, applying Simpson's rule on the smooth portion, and handling the singular piece with a substitution that removed the blowup before approximating.
Parametric and Polar Pitfalls
For parametric arc length, the formula is the integral of sqrt((dx/dt)^2 + (dy/dt)^2) dt. Do not cancel terms inside the square root unless you are certain the parameter range does not introduce absolute value issues. If dx/dt and dy/dt share a common factor that can be negative over part of the interval, the simplified integrand may integrate to a negative value over that subinterval, which breaks the arc length result. Polar area uses (1/2) int r^2 d theta. The curve may trace itself multiple times depending on the parameter range. Before setting up the integral, determine the minimal theta interval that traces the curve exactly once. For rose curves r = a sin(n theta) or r = a cos(n theta), if n is odd, the full curve traces in [0, pi]. If n is even, you need [0, 2 pi]. Using the wrong interval double-counts or undercounts area without any warning from the formula itself.
Putting It Together Under Exam Conditions
A functional cheat sheet is roughly one side of an 8.5 by 11 page when handwritten, organized into four zones: integration triggers, series test decision tree, standard series and their intervals of convergence, and geometry formulas for parametric and polar curves. Leave blank space in each zone for a single example that caused you trouble during study. Writing that example in your own words cements the pattern recognition faster than re-reading a solution. The real bottleneck is not knowing what is on the sheet. It is deciding in the first thirty seconds which zone to look at. Build your sheet around that decision speed, not around comprehensiveness. A shorter sheet that forces you to think correctly is better than a dense one that slows you down when you already know the content but panic at the layout. If your instructor allows a handwritten note, use that constraint to force prioritization. If they allow a printed sheet, do not fill it with everything you have ever seen. Fill it with the connections between methods, the edge cases you have personally struggled with, and the quick checks that prevent the most common errors. That is what actually moves the score.
