Use homological grading , with zero terms below degree zero. In the projective model structure on nonnegative chain complexes, the classes are:
Thus a cofibrant complex has a projective module in each degree, and every object is fibrant. An projective acyclic fibration is equivalently a degreewise surjective quasi-isomorphism, including degree zero. To see the degree-zero assertion, lift a degree-zero homology class using the homology isomorphism, then lift the boundary discrepancy using surjectivity in degree one. Conversely a degreewise surjective quasi-isomorphism has the requisite positive-degree surjections.
Let be the sphere chain complex with in degree and zero elsewhere. Let , for , be the disk chain complex with in degrees and identity differential. The generating cofibrations for nonnegative chain complexes are
Each is a model cofibration: its degreewise cokernel is a copy of in one degree. The extra degree-zero generator must not be omitted.
Here is the lifting characterization of an acyclic chain-complex fibration. A right lifting property for says that whenever and satisfy , there is with and . The generator says that is surjective. For a degreewise surjective quasi-isomorphism, its kernel is acyclic by the long exact sequence in homology. Choose a preliminary lift of . Then is a cycle in the kernel; subtract an element of the kernel whose boundary is this cycle. This gives the required .
Conversely the lifting conditions make the kernel acyclic: take and any kernel cycle . They also imply degreewise surjectivity inductively. Having surjectivity below degree , lift to an element of , correct its boundary within the acyclic kernel to make it a cycle, and apply the lifting condition to lift . For the chosen degree-zero element is already a cycle. A degreewise surjective map with acyclic kernel is a quasi-isomorphism by the long exact sequence in homology. Hence -injectives are exactly acyclic fibrations.
An object is sequentially small if for every sequential diagram , the natural map
is bijective. Surjectivity says that a map out of factors through a finite stage; injectivity says that two such maps agreeing in the colimit agree at a later finite stage. The relative version restricts to the stated class of sequential diagrams. The domains in have this property: maps from are cycles in one degree, and filtered colimits commute with these finite equations. Their sequential smallness is also permitted by the hint.
Apply the small object argument to the map . Put , . Given , take the set of all commutative squares with a generator on the left and on the right. Form the pushout of the coproduct of all these along their maps into , obtaining . Each bottom map induces the compatible map . Repeat for all , and set . This gives
Concretely, attaching adds a free generator in degree whose boundary is the chosen existing cycle; attaching adds a free degree-zero generator. Thus is an injection, and its cokernel in each degree is free on the newly attached generators. It is therefore a model cofibration; this also illustrates the relative cell complex description of the left factor.
For a lifting square into , sequential smallness of its domain factors the top map through some . The commutativity is already an equality in , so this is one of the squares attached at stage . Its new cell supplies a lift . Hence has the right lifting property with respect to every element of , and is an projective acyclic fibration. Every map has the required cofibration–acyclic-fibration factorization. Only the domains need sequential smallness; there is no requirement that , or the coproduct of all cells be small.
For a cohomological spectral sequence, and the next page is its cohomology. In the bounded setting, convergence of a spectral sequence to means that each has a finite decreasing, exhaustive and separated filtration with
Here is the eventual stable value. This determines the associated graded module of , rather than automatically a canonical direct-sum decomposition of . For unbounded filtrations additional completeness/convergence conditions are necessary; boundedness removes those issues here.
The bounded filtered-complex convergence theorem states the following. If is a decreasing filtration by cochain subcomplexes, preserved by the differential, finite in each cochain degree (uniform bounds , are sufficient), then there is a spectral sequence
The abutment filtration is . Its finite length ensures stabilization and the displayed limiting-page identification. The filtered cochain complex need not itself be bounded in cochain degree. In the degreewise finite version, the bounds in degrees suffice to stabilize the terms of total degree .
For a double cochain complex bounded in both indices, form the total cochain complex with anticommuting differentials and total differential . If the original differentials commute, inserting the usual sign in one of them gives this convention. Filtering by the first index gives and then . Filtering by the second index gives the spectral sequence taking horizontal cohomology first and vertical cohomology next. Both filtrations are finite, so both converge to . This is the two spectral sequences of a bounded double complex construction.
The printed left/left tensor expression needs a handedness repair. Over an arbitrary ring, a tensor product of modules over pairs a right module with a left module. To retain the order of the printed formula, take to be a bounded cochain complex of projective right -modules and a left -module. Alternatively, keep left, take right, and write and . For a commutative ring no repair is needed. We prove the first, correctly typed formulation.
Finite projective dimension gives a finite projective resolution by left modules. Form for . On take and . These anticommute, and the double cochain complex is bounded in both directions. Projective modules are flat modules, so taking vertical cohomology first gives
The first identification uses flatness of ; the second is the Tor functor computed by resolving its left-module argument . Balancedness of the Tor functor allows either correctly sided projective resolution to compute it, by the double-resolution argument.
Taking horizontal cohomology first instead uses flatness of . The resolution then leaves only in column . Its remaining differential is , so the augmentation is a quasi-isomorphism. Apply the convergence theorem to this underlying abelian-group double complex. The two spectral sequences therefore prove the Künneth spectral sequence:
These are abelian groups in general; an extra module structure requires appropriate bimodule hypotheses. The index is nonpositive, so is the nonnegative homological index of the Tor functor.
Finally write the cycle modules as and the boundary modules as . If every boundary module is projective, the short exact sequence splits. Thus is projective. The short exact sequence is now a length-one projective resolution, so for . The page occupies only columns . Every for goes columns to the right, hence has zero source or zero target. Therefore
This two-column degeneration of a Künneth spectral sequence gives the edge short exact sequences
Degeneration itself does not supply a canonical splitting. Projective boundaries also do not force to be projective: a complex with differential multiplication by over has projective boundary but a cohomology group.
For central , define the Koszul complex on central ring elements using formal exterior symbols:
with zero terms outside and differential
The terms are free bimodules with the formal symbols commuting with coefficients. Centrality makes a bimodule map, and the terms in cancel in pairs because the commute. This defines the Koszul complex over a possibly noncommutative ring without using an undefined exterior algebra of arbitrary one-sided modules. In particular is a well-defined chain complex of left modules.
A regular sequence on a module means that multiplication by is injective on for each , usually with the additional convention that the final quotient is nonzero. The homology vanishing below uses only the injectivity conditions. For one element, the chain complex is , with zero first homology and zeroth homology .
For the induction let . Adjoining the last generator identifies the new Koszul complex with the mapping cone of multiplication by on . In the convention its differential is ; the identification is . The long exact sequence in homology of this mapping cone gives
The induction hypothesis is for and . Regularity makes injective. Hence the new positive homology is zero and its zeroth homology is . The augmentation to this quotient induces these homology isomorphisms, giving the quasi-isomorphism
where the right side is placed in degree zero. The results used are the long exact sequence in homology of a degreewise short exact sequence of complexes and its mapping cone form, together with the stated induction; no unmentioned acyclicity criterion is needed.
For , is a regular sequence: is a non-zero-divisor in , and multiplication by is injective in , even when is composite. The Koszul resolution of is
With and acting as zero, applying the Hom functor gives the cochain complex
Thus, more generally, the Ext groups between polynomial-ring residue modules are
and vanish in all degrees above two. For the equal-modulus case,
These are -modules through and reduction modulo . If the multiplicative self-Ext functor is desired, the Koszul self-Ext algebra is the exterior algebra on two degree-one generators over . The two contractions on the Koszul resolution lift these classes, square to zero and anticommute, so this description also holds for composite and characteristic two. For coprime , multiplication by is invertible on , so both its kernel and cokernel vanish:
Choose a projective presentation , with a projective module. Continuing it to a projective resolution shows
Indeed a degree-one cocycle in the Hom functor applied to the projective resolution descends to , while the degree-one coboundaries are exactly restrictions of maps from . This establishes the description of the Ext functor without first assuming the extension classification.
For a module extension with inclusion and quotient map , projectivity lifts to . Since , there is a unique map with . A different lift changes by for some . An equivalence of module extensions, which is the identity on both end modules, also preserves this class. We have therefore defined a map from extension classes to .
Conversely, for , form the pushout of a module extension
The map sends to , and sends to . The first is injective because is injective. If , write ; then , proving exactness in the middle. The last map is surjective. Thus is a module extension. If , the map is an equivalence of module extensions from to . Finally identifies the pushout of a module extension built from an existing extension with its middle module. These two constructions are inverse, proving the classification:
The zero class corresponds to a split short exact sequence. Fixed end modules matter: equivalence does not permit arbitrary automorphisms of or .
In the second calculation the acting group is the infinite cyclic group. Its modules are modules over the group ring , not merely over the underlying ring . The trivial representation has acting as the identity. Its projective resolution is
Multiplication by is injective, and the augmentation quotient is . Applying the Hom functor into the trivial module makes the differential zero, so . Taking two copies gives .
An explicit representative of this extension of trivial modules for an infinite cyclic group is the abelian group , with injection , quotient , , and action
This is an invertible action: the inverse subtracts the same multiples of . Every underlying abelian-group extension splits because is free, so any group-module extension has this form after choosing lifts of the quotient basis. Replacing those lifts by multiples of does not alter . An equivalence fixing the ends has exactly such changes of lifts, so two representatives are equivalent precisely when their ordered pairs agree. Only splits as a group-module extension; overlooking the group action would incorrectly give a single class.
A feasible per-flow allocation satisfies for each active route and . It has proportional fairness if every feasible competitor obeys
By the first-order inequality for the strictly concave logarithm, this is equivalent to maximizing weighted logarithmic utility over the feasible allocations. Inactive-route rates are immaterial; set them to zero as a convention.
Write , , , and . The capacity constraints are and . For , put . Each active local route uses its remaining capacity . Omitting constants, the utility is
Its maximum has ; when , the maximum is the endpoint . Hence for active routes
These formulas also give and when . When there is no allocated service. Equivalently, the aggregate route rates for are
This is the proportionally fair allocation on a two-resource linear network for a two-resource linear flow network.
For independent Poisson processes of documents and independent document sizes with exponential distribution of rate , the memorylessness of the exponential distribution makes the population a continuous-time Markov chain. Its transition intensities are
The departure intensity sums the independent completion hazards . There is no departure on an inactive route. Set , with .
To prove the proposed stationary distribution, define the balance function of a flow-level network
For an active local route, , while for the through route this ratio is . Thus . The weights
satisfy detailed balance on every arrival-departure pair:
Each total jump rate is bounded by , so the chain is a nonexplosive Markov chain; normalized detailed-balance weights therefore give its stationary distribution, the stationary law of a two-resource linear flow network.
Use the negative binomial series to sum first over :
The remaining two geometric series give
All summands are nonnegative. Consequently the sum is finite exactly under the stability conditions for a two-resource linear flow network
These say that the offered load on each unit-capacity resource is strictly below one. Equality is insufficient: at least one geometric series then diverges. If all arrival rates are positive, the chain is irreducible and the normalizable law is its unique equilibrium law.
Finally put , . Summing over yields
Thus the two local-route counts are independent, each with geometric distribution on the nonnegative integers. The conditional through-route count has negative binomial distribution
For positive traffic rates this depends on the local counts, so all three route counts are not independent. More quantitatively, , and likewise for . If zero arrival rates are allowed, the exceptional independent cases are , or , when the relevant counts are deterministic zero. This distinction is local-count independence with through-flow dependence.
Write for a route congestion price and for resource demand. A route with weighted logarithmic utility chooses , since its marginal utility equals its route congestion price. Each resource raises its price when and lowers it when ; controls the speed. This is a multiplicative resource-price dynamics model of supply-demand feedback.
For a well-defined finite network, assume positive capacities, nonempty routes, and strictly positive initial resource congestion prices. The relevant rank condition is full row rank, . Zero initial prices are invariant under the printed differential equation, so the unrestricted phrase “all trajectories” needs a qualification; a concrete counterexample is given below.
Define . On the domain with , direct differentiation gives
Thus is a concave function, and it is a strictly concave function when has full row rank. It has a maximum: as , its negative linear capacity term dominates its logarithmic terms, while on bounded sets if any route congestion price approaches zero. A maximizing price vector therefore exists with every , though some individual resource congestion prices may be zero. The Karush-Kuhn-Tucker conditions are
They express feasibility and complementary slackness: positively priced resources are fully used, and an underused resource has zero price. Full row rank gives uniqueness of this maximizing price vector.
To prove convergence, rather than merely monotonicity of , use the relative-entropy Lyapunov function for resource prices
where a summand with means . Every summand is nonnegative and vanishes only at its reference price. Along a positive-price trajectory,
The last equality follows by substituting , and complementary slackness removes .
A sublevel set of bounds every price from above and bounds every coordinate with away from zero. Each route contains at least one such coordinate, because . All route congestion prices consequently remain bounded away from zero, so the vector field is regular on the resulting compact set. Positive coordinates cannot hit zero in finite time, since their equations have the form times a bounded function. These bounds also give global existence.
The LaSalle invariance principle now applies. Equality in requires and . Then and , so every such point is an equilibrium satisfying the same Karush-Kuhn-Tucker conditions. If has full row rank, is injective, making this set the singleton . Thus
for every strictly positive initial price vector. This proves full-row-rank convergence of multiplicative resource prices.
If is row-rank deficient, all maximizing price vectors still have the same route-price vector: otherwise strict concavity of makes their midpoint improve . Consequently the limiting rates are unique, while individual prices may not be. Positive-price trajectories actually converge to a single point in that possibly non-singleton equilibrium set. To see this, choose any subsequential limit in it, and form the same with reference . It is nonincreasing, and along the chosen subsequence it tends to zero. Hence it tends to zero on the whole trajectory, forcing .
For example, two identical unit-capacity resources serving one route of weight one have . With , obeys , while is constant. The equilibrium prices can be any nonnegative pair with sum one, and positive initial values select the limiting split; the route rate always tends to one. Rank deficiency can, but need not, cause price nonuniqueness. This example also shows why full column rank alone is insufficient.
Finally, the zero-price qualification is essential even for an invertible matrix. Take
The vector has positive route congestion prices and rates . Resource one is balanced, but resource two is overloaded; its zero price keeps its derivative zero. This is a boundary equilibrium of multiplicative price dynamics, distinct from the optimizing equilibrium
Therefore full row rank gives the unique optimizing equilibrium and convergence from strictly positive prices; it does not give uniqueness of every equilibrium on the nonnegative orthant.
Let be the finite nonempty set of routes for source-sink pair , with fixed nonnegative demand . A Wardrop equilibrium is a feasible route-flow vector such that every route carrying positive flow has minimum total delay among the routes for its pair. If
this means that, for each , some satisfies for , with equality whenever .
The link-route incidence matrix has when route uses link and zero otherwise, or the traversal multiplicity if repeated traversals are permitted. The source-sink route incidence matrix has when route serves pair and zero otherwise. Thus gives link throughputs, and gives the fixed demands.
The feasible route-flow set is a product of finite simplexes, hence nonempty and both a compact set and a convex set. The Beckmann potential
is continuous and convex, so it attains a minimum. Its route derivative is . At a minimum, moving a small amount of flow from any used route to any other route for the same pair cannot decrease the potential. Hence every used route has minimum route delay, exactly the Wardrop equilibrium condition. Conversely, at a Wardrop equilibrium , for any feasible ,
because the used flow has cost and competing routes cost at least . The first-order inequality for a convex function makes a global minimum. This proves both existence and the specified optimization characterization.
Since each is strictly increasing, its primitive is a strictly convex function. Therefore the objective is strictly convex as a function of link loads , and the equilibrium throughputs are unique. The mapping need not be injective on , so route flows need not be unique.
For a concrete route-flow nonuniqueness at a Wardrop equilibrium example, use one unit-demand pair with two successive stages, each containing two parallel links. There are four routes, indexed by their two link choices. Give every link delay . For every ,
produces load on all four links. Every route has delay one, so each vector is a Wardrop equilibrium with the same unique link loads.
To implement minimum average delay, minimize the total delay
The average differs only by the fixed total demand , provided it is positive. Choose any global minimizer , which exists by compactness, and write . Its first-order route-exchange conditions are the Wardrop equilibrium conditions for marginal link costs
A genuinely traffic-dependent choice guaranteed under exactly the printed hypotheses is the calibrated toll function
Here . The calibrated base charges are nonnegative because differentiable increasing delays have nonnegative derivatives, and the added penalties are nonnegative and nondecreasing in . The perceived link costs remain strictly increasing, and at they equal . Hence is a tolled Wardrop equilibrium. The same strictly convex Beckmann potential argument makes all its equilibrium link loads equal to ; every tolled equilibrium therefore has the globally minimum average delay. This is optimal-flow calibrated congestion tolling.
The familiar marginal external cost toll is , whose perceived-cost primitive is exactly . It implements the optimum directly when the total-delay objective is convex, for example when each is convex. That extra hypothesis does not follow from the printed strict increase of : for two identical parallel links with and total demand six, equal loads of three form a tolled Wardrop equilibrium, but the second derivative of at three is , so an unequal nearby split lowers total delay. The calibrated traffic-dependent toll above establishes the requested existence without silently assuming this additional convexity.
For the Erlang loss formula, assume a Poisson process of arrivals with rate , independent holding times with exponential distribution of mean , unit capacity usage per accepted call, and rejection without waiting when all units are occupied. Here is a positive integer and the offered load is . The call population on is a birth-death process; detailed balance gives
By Poisson arrivals see time averages, the fraction of arrivals rejected is the time probability of full occupancy. Therefore
The paper's is the offered load, rather than necessarily an arrival rate. Exponential holding times suffice for this derivation; no unproved insensitivity of loss networks assumption is needed.
A loss network with fixed routing has capacities , call types , and a link-route incidence matrix with exactly when a type- call requires link . An accepted call holds one unit on every link of its route for its holding time; if any required link is full, the call is rejected everywhere. Take independent Poisson processes of calls, independent holding times, and route offered loads . For the finite-variable formulation below, take ; a zero-capacity link and all routes using it may be removed first.
The Erlang fixed point approximation treats blocking at different links as independent and thins the traffic offered to link by acceptance at the other links. Its reduced offered load is
The equality uses : only routes with contribute. Treating this reduced stream as Poisson leads to . Independence and Poisson thinning here are the approximation, not exact properties of a general loss network.
To construct the convex potential for the Erlang fixed point, write and , the carried load of an Erlang loss resource, equivalently the mean occupancy in the single-link stationary distribution. Differentiating its finite sums gives, for ,
The variance is positive because the finite occupancy law is nondegenerate for . Furthermore maps increasingly onto , while increases from zero to . Set
In particular , is continuous and strictly increasing, and as .
For
the first sum is convex, and each integrated strictly increasing function is strictly convex. Thus is a strictly convex function on . Since tends to a positive limit on every coordinate, is a coercive function there; it has a unique minimizer. Its coordinate derivative is
The Erlang fixed point equation is exactly . For an unused link, and the optimum is ; for a used link with positive offered load, the derivative at zero is negative and the optimum is positive. Hence every fixed point gives the unique minimizer, and the minimizer gives a fixed point. The Erlang fixed point is unique under these fixed-routing assumptions.
A convergent cyclic substitution for the Erlang fixed point updates one coordinate at a time, repeatedly cycling through all links:
Always use the most recently available values of the other coordinates. Start, for example, with . In coordinates this is exact coordinate descent for : with other coordinates fixed, the displayed update is the unique coordinate minimizer.
Here is a convergence proof. All iterates stay in the initial compact sublevel set of the coercive function , and their potential values decrease to a limit. Each single-coordinate update is continuous, since its reduced load is continuous and finite and . Thus a whole-sweep map is continuous. If a full-sweep subsequence converges to , continuity gives , because both consecutive potential values converge to the same limit. Strict coordinate convexity means equality can occur only when no coordinate changes during that sweep. Thus minimizes every coordinate, satisfies the nonnegative-orthant first-order conditions, and is the unique global minimizer of . Every subsequential limit is therefore the same point, proving convergence of the full-sweep iterates and, by continuity of the updates, all intermediate iterates. This argument applies to sequential substitution; it does not assume simultaneous updates converge.
First consider an M-M-K queue with servers, arrival rate , and service rate at each server. Assume so that a stationary distribution exists. Its population is a birth-death process with rates and . The equilibrium probabilities satisfy detailed balance
Thus the population process is a reversible Markov chain. In its time reversal of a continuous-time Markov chain, forward departures become upward jumps whose rate is the constant , independent of the state. They can be generated by a rate- Poisson process independent of the initial reversed state. Reversing time preserves the Poisson process law, proving the equilibrium departure process is Poisson with rate , the Burke theorem for a multi-server queue.
The same argument gives the departure component of quasireversibility needed for the tandem calculation: the present upstream population is independent of its entire past departure process. Indeed, these departures become future arrivals in the reversed chain, independent of its initial population. First check this on finite time windows, then extend to the full past. A Poisson output alone, without this independence, would not justify multiplying queue-empty probabilities.
Now let , where denotes the mean service time at passport control, so the service rate at each of its two servers is . The stationary distribution of the M-M-s queue with has normalizing sum
Hence its empty probability is . Its output feeds a stable M/D/1 queue with rate and deterministic service time . In a stationary work-conserving single-server queue, the busy fraction is the rate of completed work, , so the empty probability is .
The downstream stationary workload is a causal function of past upstream departures. For example, it is the limit of workloads constructed from arrivals in increasingly long finite past intervals; stability makes this stationary construction well defined. The backward independence just proved therefore makes the two present queue populations independent. Consequently
With walking time , the downstream arrival stream is the upstream departure stream shifted by . Its present workload depends only on departures up to time . The number walking at time zero is the number of departures in , which has Poisson distribution of mean and is independent of the earlier departures. The upstream present population is independent of both parts of its past departure stream. Thus the three empty events factor, giving
The factor is the empty probability of the deterministic-travel infinite-server stage. Both printed stability inequalities ensure the stationary calculations apply.

Pinned article: Introduction to the OurBigBook Project

Welcome to the OurBigBook Project! Our goal is to create the perfect publishing platform for STEM subjects, and get university-level students to write the best free STEM tutorials ever.
Everyone is welcome to create an account and play with the site: ourbigbook.com/go/register. We belive that students themselves can write amazing tutorials, but teachers are welcome too. You can write about anything you want, it doesn't have to be STEM or even educational. Silly test content is very welcome and you won't be penalized in any way. Just keep it legal!
We have two killer features:
  1. topics: topics group articles by different users with the same title, e.g. here is the topic for the "Fundamental Theorem of Calculus" ourbigbook.com/go/topic/fundamental-theorem-of-calculus
    Articles of different users are sorted by upvote within each article page. This feature is a bit like:
    • a Wikipedia where each user can have their own version of each article
    • a Q&A website like Stack Overflow, where multiple people can give their views on a given topic, and the best ones are sorted by upvote. Except you don't need to wait for someone to ask first, and any topic goes, no matter how narrow or broad
    This feature makes it possible for readers to find better explanations of any topic created by other writers. And it allows writers to create an explanation in a place that readers might actually find it.
    Figure 1.
    Screenshot of the "Derivative" topic page
    . View it live at: ourbigbook.com/go/topic/derivative
  2. local editing: you can store all your personal knowledge base content locally in a plaintext markup format that can be edited locally and published either:
    This way you can be sure that even if OurBigBook.com were to go down one day (which we have no plans to do as it is quite cheap to host!), your content will still be perfectly readable as a static site.
    Figure 5. . You can also edit articles on the Web editor without installing anything locally.
    Video 3.
    Edit locally and publish demo
    . Source. This shows editing OurBigBook Markup and publishing it using the Visual Studio Code extension.
  3. https://raw.githubusercontent.com/ourbigbook/ourbigbook-media/master/feature/x/hilbert-space-arrow.png
  4. Infinitely deep tables of contents:
    Figure 6.
    Dynamic article tree with infinitely deep table of contents
    .
    Descendant pages can also show up as toplevel e.g.: ourbigbook.com/cirosantilli/chordate-subclade
All our software is open source and hosted at: github.com/ourbigbook/ourbigbook
Further documentation can be found at: docs.ourbigbook.com
Feel free to reach our to us for any help or suggestions: docs.ourbigbook.com/#contact