A numeral is a written expression representing a number. The number five can be written as the decimal numeral $5$, the binary numeral $101_2$, or the Roman numeral V. The expressions differ while the represented integer is the same.

This distinction matters when an algorithm examines numerical digits. A palindromic number, for example, is defined through its representation in a specified radix, rather than through the appearance of an abstract number. The integer five is a palindrome as the one-digit decimal numeral $5$ and as the binary numeral $101_2$, whereas ten is written $10$ in decimal and $1010_2$ in binary, neither of which is a palindrome.

Sulfate is the anion $\mathrm{S}{\mathrm{O}}_4^{2-}$, with sulfur formally in oxidation state $+6$ and an approximately tetrahedral oxygen arrangement. It forms salts with many cations.

Its charge determines the required cation ratio, while the particular ions and solid structure determine solubility. Sulfate should be distinguished from sulfite, which contains fewer oxygen atoms and has different redox chemistry.

Online text: raw.githubusercontent.com/kevin-wayne/algs4/master/src/main/java/edu/princeton/cs/algs4/Digraph.java

Authored accompanying software by Robert Sedgewick and Kevin Wayne, identified by the implementation bylines. The book publication year is 2011; copyright 2002–2025 is not an implementation publication date. Implementation locators refer to the inspected source file, not to the textbook pages.

Cited passages: Digraph.java, addEdge, adj and outdegree: one entry in the tail's collection; indegree is maintained separately; Digraph.java, class documentation and addEdge: parallel edges permitted; a directed self-loop adds one outgoing entry and increments indegree once; Digraph.java, class documentation and adj: obtaining the collection versus iterating in time proportional to outdegree.

Online text: raw.githubusercontent.com/kevin-wayne/algs4/master/src/main/java/edu/princeton/cs/algs4/Graph.java

Authored accompanying software by Robert Sedgewick and Kevin Wayne, identified by the implementation bylines. The book publication year is 2011; copyright 2002–2025 is not an implementation publication date. Implementation locators refer to the inspected source file, not to the textbook pages.

Cited passages: Graph.java, class documentation and fields: vertex-indexed adjacency collections; addEdge: entries at both endpoints; Graph.java, class documentation: parallel edges and two-entry undirected self-loop convention; addEdge applies both insertions even when v equals w.

Online text: www.stat.cmu.edu/~cshalizi/uADA/15/lectures/20.pdf

Archived chapter draft dated 1 April 2015. This timestamp identifies the inspected draft, not a print publication date. The parent is an author-hosted draft textbook, not a claim that Cambridge University Press published this version.

Cited passages: §21.1.3, printed p. 443, equations (21.2)-(21.3), and §21.1.7, p. 446: choose a label, then sample its component; §21.1.5, printed pp. 444-445: observational equivalence and label degeneracy; §21.1.6, printed pp. 445-446: probabilistic clustering and evaluation on future data.

The adrenal medulla is the central adrenal region containing chromaffin cells that release catecholamine signals into the circulation. Sympathetic neural input links it to rapid systemic responses. Circulating hormone release is distinct from local neurotransmitter release at a conventional peripheral synapse.

Write $n=|V|$ and $m=|E|$. Suppose each vertex identifier, list descriptor, link and optional edge weight occupies a constant number of machine words, and indexed array access and identifier comparison take constant time. There are $n$ list descriptors and $m$ directed, or $2m$ undirected, neighbour entries. Hence adjacency lists use $O(n+m)$ machine words in Big O notation. A full adjacency matrix has $n^2$ entries, regardless of $m$. These are counts under a storage model, not fixed byte sizes.^[ref]

Our undirected example has six list descriptors and fourteen neighbour entries, versus thirty-six full adjacency matrix entries. The numbers $20$ and $36$ count different kinds of logical items: they do not prove a particular byte saving. A linked list also needs links; an array may have unused capacity. Storing only half of a symmetric matrix changes its exact count while leaving quadratic growth. If identifiers or weights require growing numbers of machine words, their size must be included in the space complexity.

An unsorted adjacency list answers "does the edge $u\to v$ exist?" by linear search through $\mathrm{Adj}[u]$. In the worst case it inspects all $d$ entries, giving $O(1+d)$ time, or $O(d)$ for $d\ge 1$. Even if $v$ is absent, an uninspected entry could still contain it, so arbitrary lists require the whole scan. On our example, looking for $CF$ in $C$'s list compares $F$ with $A,B,D$ and fails after three comparisons. A full adjacency matrix instead answers from its single $C,F$ entry in $O(1)$ time.^[ref]

Different promises change the query. A sorted array of neighbour identifiers permits binary search in $O(\log (d+1))$ time, but maintaining order after insertion may require shifting entries. A per-vertex hash table permits expected constant-time membership under the hashing and load assumptions of that data structure; it does not give that guarantee to an arbitrary adjacency list. A stored list length can likewise make a vertex degree query constant-time even though neighbour enumeration still has to visit its entries.

The advantage grows with scale. For a graph path on $n\ge 2$ vertices, $m=n-1$, so there are $2n-2$ undirected neighbour entries and $n$ descriptors, versus $n^2$ full adjacency matrix entries. Such a family is a sparse graph with $m=O(n)$. For a complete graph, $m=n(n-1)/2$, so the undirected neighbour entries already number $n(n-1)$: the adjacency list advantage in growth rate disappears. Choose by the workload: following existing edges favours adjacency lists; many arbitrary pair-membership queries can favour a full adjacency matrix. The representation organizes the same graph for different questions.

Neighbour enumeration reports the neighbours of a specified vertex. For a simple graph, scan its adjacency list and report each entry. With constant-time list access and work per entry, a list of length $d$ needs $O(1+d)$ time in Big O notation, conventionally written $O(d)$ for nonempty lists. Use vertex degree for an undirected graph and outdegree for a directed graph. Obtaining the list reference can take constant time; consuming it requires visiting every entry.^[ref]

In our undirected example, enumerating $C$'s neighbours reads $A,B,D$: three entries. Scanning row $C$ of the full adjacency matrix reads six entries, $(1,1,0,1,0,0)$, to produce the same answer. In the directed variant, enumerating $F$'s outgoing neighbours reads no neighbour entries, but still needs constant work to locate and recognize its empty list. For a multigraph, scanning lists enumerates incident edges with multiplicity; reporting only distinct neighbours additionally requires removing repetitions.

For a short time complexity argument, each reported entry requires at least one output operation, and one scan suffices. Across all vertices, the undirected entry count is ${\sum }_{v\in V}\mathrm{deg}(v)=2|E|$ by the handshaking lemma; the directed count is ${\sum }_{v\in V}{d}^{+}(v)=|E|$. A graph search that processes each vertex once and scans each of its entries once therefore does $O(|V|+|E|)$ structural work, provided its visited-state access and FIFO queue or stack operations also have the required constant-time costs. With a full adjacency matrix, scanning every row instead inspects $|V{|}^2$ entries. This is how the representation enters the analysis of breadth-first search and depth-first search.

For a directed graph, store $v$ in $\mathrm{Adj}[u]$ for the edge $u\to v$. The reverse edge $v\to u$ is separate and is stored only if it exists. Consequently, list length is outdegree, not indegree, and each directed edge contributes one entry in total. An incoming-list index is additional storage.^[ref]

Orient the seven edges of our example from earlier to later labels. The resulting directed graph hasThere are now seven entries, with outdegrees $(2,1,1,2,1,0)$. In particular, $F$ needs an empty list even though it is reached from $D,E$: incoming edges do not make $F$'s outgoing list nonempty. Replacing each directed edge by an undirected graph connection would lose this distinction.

Our running simple graph excludes self-loops and repeated edges. If those restrictions are relaxed, specify how the adjacency list encodes them. Here an undirected self-loop at $u$ contributes two copies of $u$ to $\mathrm{Adj}[u]$, matching its contribution of two to the degree of a vertex. A directed self-loop contributes one copy, and one each to outdegree and indegree. Distinct parallel edges retain distinct entries; a multigraph's list length can therefore exceed its number of distinct neighbours.^[ref][ref]

For example, add one undirected self-loop at $A$ to the seven-edge example. Its list becomes $(B,C,A,A)$, the vertex degree of $A$ becomes four, and the total entry count becomes sixteen for eight edges. If an application instead stores a self-loop once, it must account for its doubled undirected vertex degree separately. In a weighted graph, an entry can be an ordered pair $(v,w)$ meaning destination $v$ and weight $w$; the weight belongs to the edge, not to the destination vertex.

An adipocyte is a cell specialized for lipid storage and regulated mobilization, with secretory and signaling functions. Different adipose depots and cell states have different properties. Storage and release are dynamic processes rather than a permanently fixed reservoir.

An execution model whose instructions obtain operands from a stack and put results back on that stack.

Stereochemical sequence rules rank substituents for R and S configuration. Compare the directly attached chemical atoms by atomic number first, then resolve ties by comparing successive shells at the first difference. Isotopes and multiple chemical bonds require additional rules. This ranking convention does not involve optical rotation.^[ref]

For a tetrahedral stereocenter, rank its four substituents using stereochemical sequence rules. With lowest-priority group pointing away, a clockwise $1\to 2\to 3$ sequence is $R$ and a counterclockwise sequence is $S$. Looking from the other side reverses the apparent sequence, so the viewing condition is essential. The geometric letters A/B/C/D in the tetrahedral mirror-image test carry no assigned priority. Neither $R$ nor $S$ determines positive or negative optical rotation.^[ref]

Southern Illinois University's ET 304A guide, Lab 4: Voltage Measurements and Meter Loading Effects, documents an instructive test: build equal-arm voltage dividers on a $20\mathrm{V}$ supply, then measure across the lower arm with a digital multimeter, analog multimeter, and oscilloscope. Starting with two $10\mathrm{k\Omega }$ resistors, repeat with equal pairs from $220\mathrm{k\Omega }$ to $4.7\mathrm{M\Omega }$. It requires separate recorded readings, unloaded predictions, and predictions including each instrument's input electrical resistance. The guide obtains meter resistances from specification sheets and scope resistance from the input-panel marking. Equal ratios control the unloaded target while changing susceptibility to voltmeter loading. The provided tables are blank: the guide specifies a procedure, not measured outcomes.^[ref]

Preparation and accuracy checks are made concrete by Berkeley's Lab 1: Introductory Experiments and Linear Circuits I. Its different divider uses nominal $470\mathrm{k\Omega }$ and $10\mathrm{k\Omega }$ resistors and a $24\mathrm{V}$ supply on a breadboard. Measure isolated resistors first, record actual supply voltage, recalculate with these measured values, then record the junction voltage with the digital multimeter. It explicitly compares actual values with resistor tolerance and the final voltage with a range-dependent accuracy bound from the instrument manual. This is an uncertainty check, not an independent measurement calibration.^[ref]

For a controlled comparison, additionally record the selected instrument range, probe factor, and input coupling; connect each instrument separately to the same electrical reference potential, and monitor the input rather than treating the supply label as an exact voltage. These are proposed controls for interpreting the documented procedures. A voltage indication and a resistance indication from the same digital multimeter do not independently validate its measurement calibration; a traceable reference comparison would be needed for that. Resistor tolerance describes component variation, while measurement uncertainty describes what is known about measured values. Using nominal arms alone can confuse mismatch with loading: equal nominal arms each within one-percent tolerance can give unloaded ratios from $0.495$ to $0.505$.

The inference to test is specific: once actual arm values, input voltage, and meter input electrical resistance are accounted for, the readout should approach the loaded voltage divider prediction within the remaining measurement uncertainty. A discrepancy can instead reflect contacts, changing temperature, source droop, or an input that is not a fixed passive resistor. A source with positive series electrical resistance $R_s$ can be modeled by replacing $R_1$ with $R_1+R_s$ only when $V_s$ denotes its ideal internal voltage; when the actual voltage across the divider is measured, that terminal value is already $V_{\mathrm{in}}$. During settling, or with appreciable input capacitance and changing signals, the steady direct current calculation is insufficient. Neither institutional procedure supplies measured values for the original $10\mathrm{V}$ example above.

For $R_1,R_2,R_L>0$ and $V_s\ne 0$, define the fractional shortfall in output magnitude from the unloaded value byThis is a loading bias relative to the unloaded target, not the uncertainty of an instrument's measurement of the connected circuit. The equality with the signed ratio also holds for negative $V_s$, because the output retains its sign. The Thévenin resistance $R_{\mathrm{th}}=R_1\parallel R_2$ summarizes the divider's susceptibility: the relevant comparison is $R_L/R_{\mathrm{th}}$, not merely a large absolute $R_L$.

Set $a=R_1/R_2>0$ and $x=R_L/R_2>0$. ThenThe derivative proves that increasing the load electrical resistance always reduces this error. For equal divider arms, $e=1/(1+2x)$: $x=1$ gives $33\frac{1}{3}\%$, $x=10$ gives $100/21\%=4.7619\dots \%$, and $x=100$ gives $100/201\%=0.497512\dots \%$. Thus a “ten times the lower resistor” rule does not guarantee one-percent agreement with the unloaded target.

A quantitative design requirement follows without a rule of thumb. For a specified $0<ϵ<1$,For equal $10\mathrm{k\Omega }$ arms, $R_{\mathrm{th}}=5\mathrm{k\Omega }$, so one-percent loading error requires $R_L\ge 495\mathrm{k\Omega }$. Reducing both divider electrical resistances reduces loading but increases unloaded supply electric current and Joule heating: $P=V_s^2/(R_1+R_2)$. The stipulated equal-arm example dissipates $5\mathrm{mW}$ in total when unloaded. This tradeoff explains why the ratio alone does not determine a useful divider.

Connect a passive electrical load of electrical resistance $R_L>0$ from the junction to the reference. In the steady direct current model, $R_L$ and $R_2$ share the same electric potential difference and are connected in parallel. Their equivalent electrical resistance isThis is the replacement required when a voltmeter is connected across $R_2$. The inspected teaching guide explicitly models the meter this way.^[ref]

The Kirchhoff's current law proof makes the changed assumption visible:The upper resistor carries the sum of the lower and load electric currents, so the two original resistors no longer have equal electric currents. In the stipulated $R_L=10\mathrm{k\Omega }$ example, $R_p=5\mathrm{k\Omega }$, the upper current is $2/3\mathrm{mA}$, and each lower branch carries $1/3\mathrm{mA}$.

For a digital multimeter used as a voltmeter, let its input electrical resistance on the selected voltage range be $R_m$. With no other load, set $R_L=R_m$; if an intended load is also present, use $R_L\parallel R_m$ in place of $R_L$. Two instruments connected simultaneously add two branches. Their ideal indicated value equals the voltage of this altered circuit; voltmeter loading can exist even with perfect measurement calibration. Real readout additionally has measurement uncertainty, offsets, and finite resolution. Consult the actual range and input specifications rather than assigning every digital multimeter the same $R_m$.

For an oscilloscope, the resistance-only model applies to settled direct current with a compatible input coupling setting. The oscilloscope input and cable also have capacitance; changing signals can therefore have loading that depends on frequency. A probe changes the impedance seen at its tip and the relation between tip voltage and displayed voltage; use the probe-and-input combination's specification and attenuation setting. NJIT's guide models the parallel electrical resistance and capacitance and adjusts probe compensation with a square-wave readout.^[ref]

The negative slope of the Thévenin terminal response with delivered electric current $I$. Thévenin source suppression determines it from the ratio of test voltage to entering test electric current. It need not represent a physical positive resistor inside the original network.

An authored teaching experiment uses an adjustable steady-voltage supply, a digital multimeter, fixed resistors, and a $10\mathrm{k\Omega }$ potentiometer or resistance decade box. Its original circuit uses a $10\mathrm{V}$ supply and $3.3$, $6.8$, and $4.7\mathrm{k\Omega }$ resistors; the student compares $8.2$ and $2.2\mathrm{k\Omega }$ electrical loads. The readout is load voltage, recorded against prediction and deviation. The student then removes the electrical load to measure the open-circuit voltage, disconnects the supply and substitutes a wire to measure the Thévenin resistance with the multimeter's resistance mode, and restores the supply for an independent method: adjust the variable electrical load until the voltmeter reads half the unloaded voltage. Finally the student builds the equivalent and repeats both load measurements. Its tables are blank recording forms, not reported observations. ^[ref]

The half-voltage inference assumes negligible voltmeter loading and positive $R_{\mathrm{th}}$. A finite voltmeter input electrical resistance $R_m$ actually gives an apparent unloaded readingWhen the same voltmeter remains connected across a variable electrical load $R_L$, the network sees $R_p=R_L{R}_m/(R_L+R_m)$. Setting its reading to $V_m/2$, rather than to the true $V_{\mathrm{oc}}/2$, givesThus the uncorrected half-reading method underestimates positive $R_{\mathrm{th}}$. In the original $12\mathrm{V}$ example, a stipulated $R_m=10\mathrm{k\Omega }$ gives $V_m=120/17\mathrm{V}\approx 7.059\mathrm{V}$ and half-reading load $R_L=20/17\mathrm{k\Omega }\approx 1.176\mathrm{k\Omega }$, instead of $\frac{4}{3}\mathrm{k\Omega }$. These are model predictions for a deliberately loading meter, not the specification or observed readout of the source experiment's instrument.

Finite readings can also estimate the terminal line without a physical short. For two distinct load settings with corrected delivered electric currents $I_1,I_2$ and terminal voltages $V_1,V_2$,If the voltmeter remains across each electrical load, the delivered electric current is $I_k=V_k(1/R_{L,k}+1/R_m)$, including the meter branch. Omitting it biases the inferred slope.

Useful additional controls are to measure each resistor separately before assembly, verify terminal polarity, repeat an earlier load setting to detect supply drift, and compare a third setting with the inferred Thévenin terminal response. These are proposed controls, not reported outcomes of the source experiment. Closely spaced electric currents make the slope estimate sensitive to voltage resolution and measurement error; widely separated settings may trigger source current limiting or Joule heating that changes electrical resistance. Lead and contact electrical resistance, meter input uncertainty and rounding limit the inference. An ohmmeter injects its own test electric current; its reading on a powered network is not the suppressed-network ratio. Measurements with independent sources physically removed do not characterize an active dependent electrical source whose control or supply has also been disconnected.