Online text: fncbook.com/adaptive/

§5.7 of the authors’ online textbook. Its difference-based error estimation is distinguished from a universal accuracy guarantee; no online-edition date is inferred.

Cited passages: §5.7.1, equation 5.7.4: a difference-based error estimate; §5.7.2: interval bisection; Example 5.7.2 and concluding discussion: requested accuracy is not guaranteed.

Online text: fncbook.com/rk/

Cited passages: Fundamentals of Numerical Computation, §6.4.3, equations 6.4.10–6.4.11: multistage recipe and tableau.

The hertz, symbol $\mathrm{Hz}$, is the SI unit of frequency: $1\mathrm{Hz}=1{\mathrm{s}}^{-1}$ for a periodic phenomenon. A $50\mathrm{Hz}$ oscillation completes fifty cycles per second and has period of an oscillation $20\mathrm{ms}$. The unit name is written in lower case in ordinary prose, while the symbol retains its capital H.

Angular frequency is instead expressed in radians per second. A frequency of $1\mathrm{Hz}$ corresponds to angular frequency $2\pi \mathrm{rad}{\mathrm{s}}^{-1}$, not $1\mathrm{rad}{\mathrm{s}}^{-1}$. The SI uses separate unit names to keep these physically different quantities distinguishable even though both involve inverse time.^[ref]

Online text: mathcs.clarku.edu/~djoyce/elements/bookIX/propIX36.html

Cited passages: Book IX, Proposition 36: statement and concluding proof; Euclid, Elements IX.36, translated text and commentary by David Joyce.

Substrate-level phosphorylation forms ATP by enzyme-catalyzed transfer of a phosphate group from a phosphorylated substrate to ADP. In canonical glycolysis, two such transfers occur for each three-carbon route, giving four gross ATP per glucose. Unlike oxidative phosphorylation, these transfers do not require an electron transport chain.^[ref]

The pyruvate kinase example under Coupled-enzyme readout of glycolytic ATP formation separates the material transfer from its detection. A favourable coupled chemical reaction supplies the driving change in Gibbs free energy; the enzyme supplies a catalytic route. Counting ATP-forming transfers gives a gross yield, while subtracting upstream ATP use gives the net yield.

Oxidative phosphorylation couples respiratory redox conversion to ATP formation through chemiosmosis. ATP synthase uses inward proton flow to join ADP and inorganic phosphate.^[ref]

The intermediate electrochemical gradient explains why respiratory uncoupling can preserve oxygen consumption while reducing ATP formation. An electron ledger alone cannot determine the recovery of useful free energy.

An electron transport chain transfers electrons from an electron donor through electron carriers to a terminal electron acceptor. In the mitochondrial respiratory chain, some transfers drive protons across the inner mitochondrial membrane; their return through ATP synthase can drive ATP formation.^[ref]

The useful separation is between an electron route and a proton route. The former ends in a redox chemical reaction; the latter changes an electrochemical gradient across a membrane. Energy coupling connects them without making electrons and protons interchangeable. An oxygen readout consequently need not be an ATP readout.

This article focuses on the mitochondrial respiratory chain. Bacteria can use other terminal electron acceptors in anaerobic respiration. In chloroplast light-dependent reactions, light supplies energy and linear electron transport connects water oxidation to NADP+ reduction, releasing oxygen. Thus neither oxygen consumption nor the mitochondrial arrangement defines every electron transport chain.^[ref]

Online text: discrete.openmathbooks.org/dmoi3/sec_propositional.html

Cited passages: Discrete Mathematics: An Open Introduction, 3rd edition, §3.1 Truth Tables: conjunction, disjunction and negation tables; §3.1 Logical Equivalence: definition and comparison of truth-table columns; §3.1 Example 3.1.3 and De Morgan’s Laws box; Front Matter: title, whole-book author, edition, displayed date August 12, 2025; Institutional Oscar Levin profile, Publications entry: third edition (2018).

The manufacturer-issued Protein Preparation Handbook reports a paired dialysis comparison: $2\mathrm{mL}$ of bovine serum albumin, $0.25mg/mL$ in $1\mathrm{M}$ sodium chloride, was held behind a $10\mathrm{kDa}$ MWCO membrane against $45\mathrm{mL}$ of water in a conical tube, shaken at room temperature. The bath was replaced after two hours. A conventional stirred-beaker treatment against $2\mathrm{L}$ of water supplied the comparison control. Residual salt was followed by sample electrical conductivity; results averaged two samples. This readout tests small-ion removal, not protein retention.^[ref]

Its separate retention experiment used individual proteins at $1mg/mL$ in saline or carbonate-bicarbonate buffer solution, pH $9.4$, with membranes rated $2$, $3.5$, $7$, $10$ and $20\mathrm{kDa}$. After seventeen hours at $4{}^{\circ }\mathrm{C}$, retained protein was estimated by a bicinchoninic acid assay. The size series and membrane series compare retention; they do not isolate shape independently of mass. The figure caption supplies neither a no-membrane recovery control nor a complete uncertainty analysis. These are manufacturer comparisons, not proof of universal recovery or unchanged enzyme function.^[ref]

The two experiments answer complementary questions: did the small solute leave, and did the desired macromolecule remain? In a new experiment, a bath blank and a starting salt standard would help interpret electrical conductivity, while a matched undialyzed protein aliquot and recovered volume would help distinguish handling loss from dilution. Those are proposed additional controls, not controls claimed for the published figures. Other ions also contribute to electrical conductivity, so the readout cannot generally identify sodium chloride alone in an unfamiliar buffer solution.

Consider an original ideal model with sample volume $V_s$, bath volume $V_b$, and initial sample solute concentration $c_0$. Assume each volume is fixed, a freely permeant, nonbinding solute, no chemical reactions, complete mixing on each side, and equal concentrations at equilibrium. Each fresh bath initially contains none of this solute. The protein is perfectly retained in this model.

The first equilibration redistributes the initial amount of substance $c_0{V}_s$ through both compartments. Conservation givesThis denominator includes the sample itself. For the chosen $V_s=1\mathrm{mL}$ and $V_b=9\mathrm{mL}$, $c_1/c_0=1/10$, rather than $1/9$. Nine-tenths of the initial solute is in the bath and one-tenth remains in the sample; nothing has vanished during equilibration.

Discarding the equilibrated bath removes its solute. A second fresh bath starts with only $c_1{V}_s$ of solute in the entire system, so the same argument applies again. Induction gives the geometric progressionHere $n$ counts completed equilibrations with fresh baths: $n=1$ is the first bath, $n=2$ follows one replacement, and $n=3$ follows two replacements. The normalized sample concentrations are respectively $0.1$, $0.01$ and $0.001$. Leaving the first bath in place for longer cannot lower the ideal equilibrium value below $0.1$.

The same conservation argument handles a fresh bath already containing a permeant solute at fixed concentration $b$:Thus the sample approaches $b$, not necessarily zero. For example, if a hypothetical new buffer component starts at zero in the sample and at $b$ in every bath, its sample concentration reaches $0.9b$, $0.99b$ and $0.999b$ in the three equilibrations. Outward removal and inward replacement follow the same mathematics.

Equilibrium fractions do not specify elapsed time. An illustrative well-mixed linear ordinary differential equation model within one bath isHere $K$ is an effective transfer coefficient with units of volume per time. Adding the two equations conserves $V_s{c}_s+V_b{c}_b$; subtracting their normalized forms givesThe difference undergoes exponential decay. A bath replacement resets $c_b$, renewing the driving difference without making $c_s$ jump instantly. The plotted points are equilibrium endpoints, not a measured time course or a timetable for an arbitrary protein.

Real sample volume can change through osmosis; proteins can be lost through leakage or adsorption to the semipermeable membrane. Consequently, a measured decrease in protein concentration alone cannot distinguish dilution from loss of protein. Measuring recovered volume as well as protein concentration permits an estimate of recovered protein mass.^[ref]

Buffer exchange changes the small components surrounding a retained macromolecule toward those of a desired buffer solution. Components absent from fresh dialysate tend to leave; supplied components can enter. Each permeant solute has its own exchange problem.^[ref]

For a dilute, nonbinding permeant solute, an ideal equilibrium calculation approaches the bath's specified concentration. That approximation need not describe a solute bound to a protein or a charged mixture with additional electrical constraints. Retaining a protein also does not by itself establish that its protein structure or catalytic function survives the change of buffer solution.

The molecular weight cut-off rates retention of reference molecules. One manufacturer's definition uses the mass at which approximately $90\%$ remains after seventeen-hour dialysis, tested with globular proteins. A $10\mathrm{kDa}$ rating uses dalton units: a mass, not a length or an absolute passage threshold.^[ref]

A semipermeable membrane has a distribution of passage sizes, and a molecule's shape affects passage as well as its mass. Near the nominal MWCO, slow passage can make substantial retention compatible with eventual leakage. The protein-retention experiment below makes retention an observable quantity rather than a universal yes-or-no property inferred from the label.

The external chemical solution on the other side of a dialysis semipermeable membrane. Its composition supplies the boundary condition for exchange: it can receive unwanted small solutes or supply desired small solutes. A bath replacement discards the old dialysate and introduces a new reservoir with a specified composition.

Online text: chalmersgu-data-structure-courses.github.io/dsabook/html/section-9.1.html

The inspected online edition is undated. The displayed 2016–2026 range has no stated date role and is not taken as a publication year.

Cited passages: Data Structures and Algorithms, section 9.1.1: complete shape and binary heap example; Data Structures and Algorithms, section 9.1.1: full shape and Huffman coding tree example; Data Structures and Algorithms, section 9.1.1: perfect shape, equal leaf levels, and Figure 9.4; Data Structures and Algorithms, section 9.1.2, height pseudocode: null returns -1.