GIKI Test common mistakes
GIKI Test · Understand the traps and how to avoid them.
When a circuit problem asks for current through a branch, the candidate applies the voltage-divider formula directly to resistors in parallel branches, or sums branch currents algebraically without accounting for direction at a node.
Voltage-divider and current-divider shortcuts are memorised as formulas, but candidates do not check the precondition that divider rules apply only to series strings (voltage) or parallel branches sharing two nodes (current). Under time pressure the formula is pattern-matched to any resistor network.
What to do instead
Before using any divider rule, redraw the network and label which elements are in series (same current) and which are in parallel (same voltage). If the precondition fails, fall back to nodal analysis: assign node voltages, write KCL at each unknown node, and solve. Practise by taking five mixed networks and solving each twice, once by divider rules where valid and once by nodal analysis, to see where the shortcut breaks.
The candidate computes a Thevenin or Norton equivalent but reports the open-circuit voltage with the load still attached, or forgets to deactivate independent sources (short voltage sources, open current sources) when finding the equivalent resistance.
The procedure has several ordered steps and candidates compress them, treating the load as part of the network. Deactivating sources is a convention that is easy to skip because it feels like altering the circuit rather than analysing it.
What to do instead
Use a fixed four-step checklist written on scratch paper: (1) remove the load and mark its terminals, (2) find V_oc across those terminals, (3) deactivate all independent sources, (4) find R_eq looking into the terminals. Verify with the short-circuit current: R_th must equal V_oc divided by I_sc. If the two disagree, an error exists in one of the steps.
In a transient RC or RL problem, the candidate writes the time constant as R times C or L over R using the wrong resistance, for example the source resistance instead of the resistance seen by the capacitor or inductor with sources deactivated.
The formula tau = RC is remembered as a single symbol pattern without the rule that R is the equivalent resistance facing the reactive element. Circuits with multiple resistors make the correct R non-obvious.
What to do instead
For every transient problem, first find the initial condition (capacitor voltage or inductor current cannot jump), then find the final steady state (capacitor open, inductor short), then compute tau by deactivating sources and finding the resistance seen from the reactive element's terminals. Check that the solution satisfies both the initial and final values before moving on.
The candidate analyses a recursive or nested-loop algorithm by counting the inner loop body once per outer iteration without multiplying by the inner loop's own bound, and concludes O(n) where the true cost is O(n^2) or O(n log n).
Candidates read code line by line rather than summing work over iteration bounds. Nested loops with dependent bounds, such as an inner loop running from i to n, are especially prone to this because the inner count changes each pass.
What to do instead
For each loop, write the number of executions as a summation with explicit bounds, then evaluate it, for example the sum from i=1 to n of (n - i + 1) equals n(n+1)/2. For divide-and-conquer code, write the recurrence and solve it with the master theorem or a recursion tree. Confirm the answer by counting operations for small n (n=4, n=8) and checking the growth matches the claimed bound.
When asked for the output of pointer or reference code in C or C++, the candidate traces the code assuming the pointer and the pointee are the same variable, so an assignment through a pointer is treated as reassigning the pointer itself.
Pointer notation compresses two distinct operations, changing the address stored and changing the value at that address, into visually similar syntax. Candidates who have not internalised the distinction trace memory incorrectly.
What to do instead
Draw a memory diagram with boxes for variables and arrows for pointers before executing each line. After every assignment, update the diagram and ask whether the arrow moved or the box's contents changed. Practise with swap functions (by value versus by pointer) and with arrays passed to functions, where the array name decays to a pointer.
In a statics problem, the candidate sets up equilibrium equations but omits a reaction component, for example treating a pin support as providing only a vertical reaction, or ignores a moment reaction at a fixed support.
Support types are learned as a table, but under exam pressure the candidate draws the free-body diagram from the physical sketch rather than from the support symbols, and an incomplete diagram yields an unsolvable or wrongly solved system.
What to do instead
Build a free-body diagram by listing every support and writing its reaction count from the symbol: roller gives one force normal to the surface, pin gives two force components, fixed gives two forces plus a moment. Count unknowns and compare with available equations (three in two dimensions) before solving. If unknowns exceed equations, the structure is statically indeterminate and the problem likely expects a different method.
The candidate answers a thermodynamics question about an isentropic process by applying the relation between pressure and volume with the wrong specific-heat ratio, using the value for air regardless of the gas stated, or applying the relation to a process that is not adiabatic and reversible.
The formula with gamma is memorised as the general ideal-gas relation, and the constant 1.4 for diatomic air is over-learned. Candidates do not check whether the gas is monatomic or polyatomic, or whether the process is actually isentropic.
What to do instead
Before using any isentropic relation, confirm two things: the process is adiabatic and reversible, and the gas's degrees of freedom give the correct gamma (5/3 for monatomic, 7/5 for diatomic, 4/3 for polyatomic). Derive the relation once from the first law plus the ideal-gas law so the assumptions are visible, and check the entropy change is zero as a consistency test.
In a phase-diagram or heat-treatment question, the candidate reads a composition from the phase boundaries but does not apply the lever rule, instead reporting the average alloy composition as the phase composition, or inverts the lever-rule arms.
The lever rule is a geometric construction on the tie line, and candidates confuse which segment corresponds to which phase. The average composition is also the most visually prominent number on the diagram, so it is mistakenly reported as a phase composition.
What to do instead
Draw the tie line at the temperature of interest, mark the alloy composition and both phase-boundary compositions on it, and write the fraction of each phase as the length of the opposite segment divided by the total tie-line length. Sanity-check that the fractions sum to one and that the alloy composition lies between the two phase compositions.
In a reaction-engineering or separation problem, the candidate writes a material balance in molar units for one stream and mass units for another, or forgets to convert a volumetric flow rate to molar flow using the correct density or ideal-gas relation.
Problem statements mix units deliberately, and candidates carry numbers forward without tracking units. Molar and mass balances look identical in form, so the mismatch is invisible until the final number is wrong.
What to do instead
Attach units to every number written, including in intermediate steps, and cancel them explicitly. Convert all flows to one basis (moles per time is usually safest) at the start, and state the basis on the first line of the solution. Check the final answer's units and order of magnitude against the feed stream before reporting.
In a structural or geotechnical problem, the candidate selects a design load combination or a factor of safety from memory without checking which limit state (strength or serviceability) the question describes, and applies strength factors to a deflection check.
Load factors and safety factors are memorised as numbers, but their pairing with limit states is a code-specific rule. Candidates who have not practised classifying the limit state default to whichever factor they remember best.
What to do instead
Before computing, classify the question as strength (collapse, yielding, bearing capacity) or serviceability (deflection, settlement, crack width), and write the limit state on the page. Use factored loads for strength and unfactored service loads for serviceability. If the code is not stated, state the code and edition you are assuming so the examiner can follow your reasoning.