Protection against unauthorised disclosure of information. Access control can prevent a subject from reading a file, while cryptographic confidentiality protects message contents against specified adversaries using cryptography.

Protection of data against unauthorised or accidental alteration. It is distinct from information confidentiality and service availability.

A participant whose correct behaviour and protection of secret state are assumptions of a security argument. It can mediate authentication and session key transport, but the protocol still needs to authenticate its responses.

Reuse of previously observed valid messages to obtain an acceptance or action in a different context. Cryptographic nonces, fresh session keys and explicit context binding can prevent particular replay patterns.

An attack on service availability by exhausting resources or preventing useful work. Requiring inexpensive unauthenticated messages to trigger expensive private-key operations creates an amplification opportunity.

A moderately expensive computation whose result can be checked cheaply. Request-specific puzzles can mitigate a computational denial-of-service attack if verification precedes expensive service and solved puzzles cannot be reused indefinitely.

Original article in Journal of the ACM 32(3), pp. 652–686. Author-hosted publication record and abstract: www.cs.cmu.edu/~sleator/papers/Self-Adjusting.htm

The article introduces splay trees and proves a logarithmic amortized time complexity for their standard operations.

Official historical API documentation for the mutable stack: the empty constructor, push, pop and nonEmpty support an explicit inorder traversal stack. The API's deprecation notices concern later collection choices; they do not remove these documented operations from the cited version.

Original Czech article in Práce Moravské přírodovědecké společnosti 3, pp. 37–58. The opening formulation assigns pairwise distinct positive costs to connections and asks for minimum total cost while retaining connection between all points.

Digital archive: dml.cz/handle/10338.dmlcz/500114

The official archived API documents lazy lists, memoization of their evaluated values, and retention of cached tails while earlier cells remain reachable. The companion object supplies Stream.iterate for repeated function application. This source documents the Scala 2.12 API family; it does not establish the exact compiler patch release used for a historical examination.

Defines Scala trait linearization, inherited member lookup, and evaluation of trait initialization statements.

Specifies method overloading, numeric widening, and instance method selection in Scala.

The compiler's methodSig, typesFromOverridden, and assignTypeToTree use an inherited return type as an expected type when checking an overriding method body. This permits numeric widening of a literal in that body while still inferring more specific reference result types where appropriate.

Examination paper of the University of Oxford, Preliminary Examination in Computer Science, Trinity Term 2015, dated 19 June 2015. The original has seven pages; the last contains the prescribed natural deduction rules. Released archive: Lawrence Attachment.zip, member 2015-cs4.pdf. The paper's printed section order is linear algebra, digital electronics, and introduction to formal proof, despite the different ordering in its cover title.

Primary source: released examination archive, member a10100w1-4.pdf. The original eleven-page cover identifies the Preliminary Examination in Computer Science, Long Vacation 2015, Wednesday 9 September. The archive folder name is not the examination year. Page 8, Question 6(b)(iii), explicitly identifies “saddle point” with “point of inflexion”; that usage is specific to this paper. The definition in Question 6(a)(ii) leaves endpoint values unspecified, which does not affect its Fourier series.

The University of Oxford preliminary computer science examination whose original cover gives Trinity Term 2015, Thursday 25 June, and formal code A10100W1. The released archive contains it as 2015-cs3.pdf: publicly released examination archive. Printed pages 2–4 contain Questions 1–3; pages 5–7 contain Questions 4–6; pages 8–10 contain Questions 7–9.

The printed codomain in Question 2(b) is inconsistent with its formula, and an unrestricted zero extension of the binomial coefficients makes the premise in (ii) vacuous beyond the finite row. Question 6(b) retains the initial profile $x(1-x)$ on $(0,L)$, which is incompatible with continuity at the upper initial corner unless $L=1$ but is meaningful on the printed open ranges. Question 9(c)(iii) omits the integration variable after both differential signs. These are source qualifications, not missing source material.

The “Variance Annotations” section specifies variance positions and lower-bound widening, used when an immutable class's output type parameter is covariant while a method must accept a possibly broader output value.

Examination dated 5 June 2015, Trinity term, University of Oxford. The public archive contains the paper as the member 2015-adsa.pdf of Lawrence Attachment.zip. The three original pages contain the cover and thirteen numbered subparts on amortized analysis, flow networks, and linear programming.

The four-page University of Oxford Finals compiler examination dated Thursday 4 June 2015, Trinity Term 2015. The cover permits answers to two of three questions. Public primary-source archive: Lawrence Attachment.zip, member 2015-compilers.pdf. The paper concerns ordered case statements with expression-valued labels, nested procedure environments, and common subexpression elimination for array swaps.

Hilary Term 2023 lecture slides. Slides 47–48, on PDF pp. 62–63, state the linear speedup theorem for $k>1$ tapes and explain the block simulation. The source is used for that general theorem, not as an answer key for an examination.