Architecture Candidate Synthesis
About this pattern
This is a generated FPF pattern page projected from the published FPF source. It is canonical FPF content for this ID; it is not a FPF Reference product feature page.
How to use this pattern
Read the ID, status, type, and normativity first. Use the content for exact wording, the relations for adjacent concepts, and citations to keep active work grounded without pasting the whole specification.
Type: Architectural pattern Status: Stable Normativity: Normative unless explicitly marked informative
Use this pattern when a practitioner has a C.30-grounded architecture question for one exact described holon and needs to synthesize several candidate architecture configurations across selected structures before comparison, archive or front-policy work, selected-set result declaration, actual publication, or decision. Keep any obtaining C.30 ArchitectureRelation occurrences, the selected U.Structure values they relate to the holon, and any candidate, required, desired, or expected structures named only in an ArchitectureClaim distinct throughout the synthesis.
Keywords
- architecture candidate synthesis
- CandidateArchitecturePalette@Project
- selected structures
- architecture characteristics
- selected-structure contribution rows
- candidate configurations
- trade-off front
- retained alternatives.
Relations
C.30.TFSContent
Problem frame
Use this pattern when a practitioner has a C.30-grounded architecture question for one exact described holon and needs to synthesize several candidate architecture configurations across selected structures before comparison, archive or front-policy work, selected-set result declaration, actual publication, or decision. Keep any obtaining C.30 ArchitectureRelation occurrences, the selected U.Structure values they relate to the holon, and any candidate, required, desired, or expected structures named only in an ArchitectureClaim distinct throughout the synthesis.
Primary working reader: an architect or architecture-responsible practitioner preparing alternatives for one described holon before comparison, selection, selected-set result declaration, actual publication, local choice, or project decision.
Typical entry phrases:
First-minute use slice. A regulated product-family team has a C.30-grounded architecture question for one exact field-device-family holon. The question names its current obtaining ArchitectureRelation occurrences and their selected structures separately from candidate or expected structures stated only in the current ArchitectureClaim. The work question is synthesis: how should required functions, constructive modules, field placement, control responsibility, and certification evidence be coordinated so maintainability, substitutability, latency, and evidence reuse stay acceptable? Using C.32, the practitioner first records the selected structures and what each contributes to the synthesis, then records three candidate configurations: one shared module grammar with tighter evidence scope, one product-family split with lower interface burden, and one bounded exception that keeps the existing module split but changes evidence responsibility and reopen trigger. The team now has candidate architecture configurations under declared characteristics, not one attractive platform proposal and not new obtaining architecture relations by candidate wording.
The primary EntityOfConcern is the candidate architecture palette for one C.30-grounded synthesis question. Its inputs are the described holon, any obtaining ArchitectureRelation occurrences and their selected U.Structure participants, and any candidate, required, desired, or expected structures stated only in an ArchitectureClaim.
The described holon may be a system, product family, organization-as-system, discipline, AI-agent setup, built asset, episteme, Work occurrence, or another admitted holon kind. Do not admit a source label as a holon. For example, practice, culture, tradition, style, Method, or role may refer to a Method, Method relation structure, relation among local system-role kinds, classification, assignment, Work structure, episteme, source-local meaning, or C.36 cultural-evolution relation. Recover the actual object and claim through its subject pattern; route unresolved claim-bearing role wording through [E.10.ROLE](/generated/patterns/E.10.ROLE).
ClaimScope and a bounded model-use structure qualify the named use; neither becomes the holon. Architecture pressure may concern Method-family structures, relations among local kinds, classifications, or assignments. Keep each as a selected structure or separate input for the named architecture use, not as a holon kind or function bearer by label. C.32 is not software-system architecture by default; software-system sources are one source family and one domain example.
What goes wrong if C.32 is missed: the team optimizes one visible structure, such as modules, placement, team responsibility, control relation, or evidence package, and then treats that local improvement as architecture synthesis. The competing structures, architecture characteristics, losses, and alternatives disappear before they can be compared.
What C.32 buys in practice: a practitioner can build a small set of candidate architecture configurations, each grounded in selected structure changes, architecture characteristics, known losses, and patterns for the next questions.
Ordinary working move: name the selected structures that really change, name the few architecture characteristics that make the trade-off real, then write two to five candidate configurations with gain, loss, preserved structure, hidden loss, and next receiving use.
Adoption test: after using C.32, another practitioner can see at least two structurally different candidate configurations, the selected-structure changes, the architecture characteristics under pressure, each gain and loss, the source-return condition, and the next receiving use.
Use C.32 only for candidate palette construction. Do not use it to ground the architecture claim, recover one structure, build characteristic criteria rows, design eval programs, handle architecture-influence correspondence, run archive or front-policy work, declare a selected-set result, publish it to an audience, choose locally, or decide the project architecture.
Use [C.32.MWA](/generated/patterns/C.32.MWA) instead when several structures of Methods, Work, subjects and their descriptions, capabilities and providers, and cultural change do not line up one-for-one and the needed result is one usable practice architecture. Keep C.32 for a palette of candidate configurations for one grounded architecture question.
Common exits by claim kind:
[C.30](/generated/patterns/C.30)grounds the described holon, any obtainingArchitectureRelation, its selectedU.Structure, and any separateArchitectureClaim;[C.30.ASV](/generated/patterns/C.30.ASV),[A.6.F](/generated/patterns/A.6.F), and[A.6.M](/generated/patterns/A.6.M)recover structural views, function wording, and module-interface relations.[C.32.HCS](/generated/patterns/C.32.HCS),[C.32.ACS](/generated/patterns/C.32.ACS),[C.32.ACE](/generated/patterns/C.32.ACE),[C.25](/generated/patterns/C.25),[C.31](/generated/patterns/C.31),[C.31.ASAP](/generated/patterns/C.31.ASAP), and[C.16](/generated/patterns/C.16)govern starter heads, project criteria rows, eval programs, Q-Bundles, modularity or scale-preference claims, and measurement.[C.32.MLAO](/generated/patterns/C.32.MLAO),[C.32.CONWAY](/generated/patterns/C.32.CONWAY),[C.32.FAIL](/generated/patterns/C.32.FAIL), and[C.29](/generated/patterns/C.29)govern residual-reducing frames, architecture-influence and transformed-architecture correspondence, candidate repair, and mathematical-lens use.- Use
[A.19.CPM](/generated/patterns/A.19.CPM)for comparison,[A.19.SelectorMechanism](/generated/patterns/A.19.SelectorMechanism)for set-returning selection,[C.18](/generated/patterns/C.18)and[C.19](/generated/patterns/C.19)for archive, front, and current-pool treatment,[G.5](/generated/patterns/G.5)for selected-set result declaration,[C.11](/generated/patterns/C.11)for local choice, and[C.32.PAD](/generated/patterns/C.32.PAD)for a project architecture decision. When audience availability is current, use[E.17](/generated/patterns/E.17)for a source-backed publication face and return to source and[E.24.PUB](/generated/patterns/E.24.PUB)for the publication occurrence, form, carrier, audience, bounded use, and availability. [C.30.AD](/generated/patterns/C.30.AD),[E.17](/generated/patterns/E.17),[E.24.PUB](/generated/patterns/E.24.PUB),[A.10](/generated/patterns/A.10), and[B.3](/generated/patterns/B.3)govern architecture-description, publication-face, publication-occurrence, evidence, and assurance claims, respectively.
The first useful output is CandidateArchitecturePalette@Project. It is the project working record for candidate-palette construction. The name does not introduce a new U.* kind, and the record does not carry selection, publication, evidence, assurance, or decision authority.
For a first pass, fill only the described holon, synthesis question, intended palette use, current architecture relations and selected structures that change the question, selected-structure contribution rows, live architecture-characteristic rows, candidate configurations, and palette stop condition. Add ClaimScope or a bounded model-use structure only when it changes synthesis; add other optional refs only when they change the next use of the palette:
Across C.32, @Project is a compatibility and retrieval cue, not a project kind or relation assertion. CandidateArchitecturePalette@Project, ArchitectureSynthesisFrame@Project, and ArchitectureCharacteristicImprovementLoop@Project establish no composite project work, context, authority, viewpoint, or parthood by name. When one of these records is genuinely local to one actual project, identify the exact composite U.Work and the direct relation by which synthesis framing, palette construction, or improvement feedback concerns that work. Otherwise no project-work relation is implied. A cited ArchitectureQuestionCard@Project transfers neither project locality nor architecture truth: each affirmative currentArchitectureRelationRef must already resolve to one obtaining C.30 occurrence, while candidate, required, desired, or expected structure remains claim content until the C.30 predicate is independently satisfied.
Problem
Architecture synthesis is the constructive middle of architecture work. A practitioner may already know the described holon, architecture question and intended use, some obtaining architecture relations and selected structures, and some concerns, but still need to configure those structures together before later comparison or decision can be honest.
The typical synthesis problem is multi-structure. State each required function or functioning claim through the predicate and bearer recovered with A.6.F. Candidate module, placement, control, transformation-flow, information, evidence, Method, Work, local-kind, classification, or assignment structures may constrain or help explain a candidate, but a kind or assignment establishes no functioning, participation, capability, function bearing, or Work. A control relation can improve supervision while increasing timing or responsibility burden; an information structure can improve maintenance access when exposed through a digital-twin view while still hiding source-return loss; a team structure can improve flow while failing to match module or deployment structure. Every positive responsibility claim still needs its direct domain predicate, actual participants, applicability, and occurrence identity, or the exact A.6.RCD missing governor.
A functional architecture is not enough by itself. A function graph, use case decomposition, workflow, neural cell graph, Method step, or source function from cultural or practice material can enter architecture synthesis only after A.6.F identifies the function or functioning claim and its possible bearer, and after any selected structure or source label is recovered. If no bearer can satisfy the predicate recovered with A.6.F under the relevant module, Method, resource, placement, control, Work, evidence, local-kind, classification, or assignment constraints, the candidate must be repaired before it enters comparison, selection, local choice, or decision work. Those neighboring structures constrain the candidate; they do not bear the function by label.
The typical synthesis problem is also multi-characteristic. Architecture characteristics such as cohesion, coupling, substitutability, evidence reuse, work repeatability, latency, locality, control separation, source-return cost, and composite quality families often compete. Functional demands describe what the holon is to do; architecture characteristics describe whether the selected structures make those demands maintainable, controllable, evolvable, replaceable, inspectable, and otherwise acceptable in the current context.
One recurring candidate-generation heuristic is idealization: ask whether an existing bearer or resource can carry an additional required function under the selected-structure constraints, whether a support bearer can disappear, or whether a more general scale-amenable bearer can replace several special bearers. Admit that heuristic only as a candidate. The candidate must name the functions transferred to a bearer, the bearer removed or generalized, the architecture characteristics improved and worsened, and any BLP scale window or waiver when scale advantage is claimed.
Use C.32 to make the constructive translation explicit. Build a small palette whose candidates answer: which selected structures are configured together, which architecture characteristics improve or worsen, which constraints remain admissible, what source detail must remain recoverable, and which pattern supplies the next claim or test.
Forces
Solution
Create an ArchitectureSynthesisFrame@Project when the selected structures and characteristics are not yet visible enough. The frame is a temporary visibility aid for C.32 use; the palette remains the first useful output. Then create a CandidateArchitecturePalette@Project. Treat the palette as a small constructive object over selected structures of a described holon, not as a checklist, not as a decision, not as a selected-set result declared under G.5, and not as a publication occurrence.
Work in seven steps:
- Anchor the palette to one described holon or holon family, synthesis question, and intended next use. Name any current C.30 architecture relations and selected structures that can change that question.
- Write the smallest useful set of selected-structure contribution rows. Start with the functional demand and candidate bearer recovered with
A.6.F, constructive module or manufacture structure, and placement or deployment structure when they shape the question; add control, transformation-flow, Method, Work, local-kind relation or classification, assignment, information, evidence, scale, or other selected structures only when they change the synthesis question. Send unresolved claim-bearing “role” wording throughE.10.ROLE. For each required function, name at least one admissible bearer under the declared constraints. - Reference the architecture-characteristic criteria rows and any Q-Bundle slots that make the trade-off real. Separate functional demand, architecture characteristics, criteria rows, eval results, and decisions.
- Generate candidate architecture configurations. A candidate claim may propose, for example, changed decomposition, allocation, A.6.F function bearing, bearer count, placement, interface grammar, a control or transformation-flow relation, Method use, future assignment conditions, an independently established responsibility relation, evidence scope, information structure, or a bounded exception. Modal candidate wording creates no assignment occurrence and proves no Work occurred. Use a WorkPlan, policy, commitment, permission, decision, or other truthful prospective object when one applies. For actual precise Work, recover each exact actual performer System through A.13 and let A.15.1 independently admit the dated Work and enacted Method; add an assignment occurrence, its declared species, and F.6 only when the candidate account or its receiving use expressly consumes precise assignment-bound attribution through the same obtaining A.13 assignment. F.6 identifies neither assignment nor performer, missing or failed F.6 leaves the Work intact, and an assignment never carries responsibility by itself.
- For each candidate, state selected structure changes, expected architecture gain, known architecture loss, constraint fit, preserved structure, lost or hidden structure, and source-return condition.
- When a front, archive, search result, or pool-treatment policy is being used, cite
C.18,C.19, or NQD and OEE support as generation or retention support only. Keep the C.32 candidate content separate from archive work, front membership, pool treatment, selected-set result declaration, actual publication, and local choice. - Stop when the palette contains the fields required by the pattern for the next question, such as comparison, C.18 or C.19 front-policy use, selected-set result declaration, actual publication, local choice, decision, or repair.
These contribution rows are not an audit checklist. Together they name only the structures that actually change the candidate configuration.
Architecture-characteristic improvement loop. C.32 is one turn in a continuing improvement cycle over architecture characteristics, not a one-shot search for final form. The practitioner starts with characteristic pressure or criteria rows from C.32.ACS, C.31, C.25, C.16, C.16.P, C.31.ASAP, or a local Q-Bundle; synthesizes candidate selected-structure changes; and records which criteria rows are expected to improve and which protected rows may worsen.
ArchitectureCharacteristicImprovementLoop@Project is a local feedback record for reopening C.32 synthesis when characteristic pressure changes. It is not an E.23 method, an ACE eval program, a comparison rule, a selection result, or a decision.
Keep each receiving claim with its subject pattern.
Criteria rows stay with C.32.ACS; Q-Bundles with C.25; scale preference with C.31.ASAP; measurement with C.16; eval programs and eval results with C.32.ACE.
Improvement-question framing and repeated-improvement method stay with E.22 or E.23.
Use A.19.CPM for comparison, A.19.SelectorMechanism for set-returning selection, G.5 for selected-set result declaration, C.11 for local choice, and C.32.PAD for a project architecture decision. For publication, use E.17 for a source-backed face and source return and E.24.PUB for the publication occurrence and audience availability.
For this loop, bring only the changed characteristic pressure into C.32 and return the next candidate palette.
Open the next synthesis question from the resulting eval result, front relation, retained alternative, rejected candidate, or source-return trigger.
An eval result that cohesion improved, evidence reuse decayed, coupling changed, latency worsened, or exception growth changed does not choose an architecture. A practitioner may use it as feedback only after the bearer, criteria row, scale or qualitative reading frame, selected structures, parity frame, and pattern for the next question are recoverable.
Candidate architecture changes are local C.32 entries for candidate configurations. They are not FPF work occurrences, method steps, or receiving-pattern claims. A change is admissible only when the selected structure being changed is named.
Didactic mini-slices. Use these as examples of the kind of work C.32 expects, not as domain-specific templates.
When one independently typed architecture-side or other source constrains transformed-side architecture content for a changed referent, use [C.32.CONWAY](/generated/patterns/C.32.CONWAY) before using Conway, mirroring, or inverse-Conway language in candidate synthesis. The practitioner names the changed referent and any actual A.3.4 transformation separately, each influence source by exact kind and its direct relation only when that occurrence is asserted, and, for each actual architecture side, the exact C.30 described holon, obtaining ArchitectureRelation, and selected U.Structure; modal architecture content stays in an exact ArchitectureClaim. Without an admitted and satisfied direct influence predicate, the pressure stays synthesis-local in the C.32.CONWAY frame with its missing-governor, unresolved-grounding, or false-predicate disposition and no exact pair row. Candidate work then names influence-source-side, transformed-side, joint, or bounded-mismatch changes, architecture characteristics under pressure, expected gains, known losses, and source-return conditions.
Keep the candidate palette as the C.32 result. [C.32.CONWAY](/generated/patterns/C.32.CONWAY) carries the architecture-influence correspondence frame or one exact reusable pair-row episteme. Influence alone supplies no acting System, local system-role kind, System-classification judgment, assignment, Work, changed-referent identity, or transformation participation. Transformation, acting and Work attribution, exact influence, transformation-flow, and module-interface claims belong to [A.3.4](/generated/patterns/A.3.4), [A.12](/generated/patterns/A.12), [A.2.1](/generated/patterns/A.2.1), [A.15.1](/generated/patterns/A.15.1), [F.6](/generated/patterns/F.6), the direct influence pattern, [E.18](/generated/patterns/E.18), C.30.TFS-REL, or [A.6.M](/generated/patterns/A.6.M) when current. Structural-similarity or preservation claims belong to [C.29](/generated/patterns/C.29) when they are current.
A richer dossier is optional. Open it only when one candidate must carry source views, relation notes, measurements, C.29 lens outputs, evidence notes, or failure repairs that affect the next architecture use. Ordinary C.32 use should remain one row per candidate configuration.
Downstream use. The C.32 result is architecture-specific candidate content. Use [G.5](/generated/patterns/G.5) to declare a selected-set result, [C.11](/generated/patterns/C.11) for a fixed local choice, and [C.32.PAD](/generated/patterns/C.32.PAD) for a project architecture decision. Use [C.18](/generated/patterns/C.18) or [C.19](/generated/patterns/C.19) for archive, front, pool-treatment, or generation policy when that claim is current. Use [C.30.AD](/generated/patterns/C.30.AD) for architecture-description work. For publication, use [E.17](/generated/patterns/E.17) for a source-backed face and source return and [E.24.PUB](/generated/patterns/E.24.PUB) for the publication occurrence and audience availability.
Stop condition. Stop C.32 when the palette can support the next use without hiding the selected structures, architecture-change kind, architecture gain, architecture loss, constraint fit, source-return condition, or pattern for the next question.
Lowering condition. Leave C.32 candidate construction when the needed architecture claim is not grounded, the item is only a source artifact, only one configuration is visible, the candidate lacks selected-structure change, the functional demand has no feasible bearer, the architecture gain or loss is unnamed, or the next use is already comparison, selection, selected-set result declaration, actual publication, local choice, decision, evidence, or assurance. Use [C.30](/generated/patterns/C.30) for grounding, the source or description pattern for source artifacts, [C.32.FAIL](/generated/patterns/C.32.FAIL) for candidate repair, and the named pattern for the next question when the downstream claim is current. Reopen C.32 when a criteria row, eval result, retained alternative, front relation, source-return trigger, or source-currentness change alters the selected structures under pressure or the acceptable loss profile.
Archetypal Grounding
Tell. The regulated product-family first-minute slice in the Problem frame shows the minimum complete move: one grounded question, a small set of selected-structure contribution rows, three genuinely different candidate configurations, explicit gains and losses, and no premature decision. Show. The cases below vary the described holon and the selected structures while keeping that move recognizable. Show again. The didactic mini-slices in Solution 4 show what to repair when a required function has no feasible bearer. These three views are one grounding set, not three additional procedures.
Bias-Annotation
Scope: Limited to constructing a small candidate architecture palette for one C.30-grounded architecture question about one described holon or holon family. C.32 is not a universal architecting Method, a software-architecture default, a comparison or selection rule, a publication route, or a decision procedure.
Conformance Checklist
Common Anti-Patterns and How to Avoid Them
Architecture trade-off failures
More repair cues
Consequences
Rationale
Architecture practice needs a method between a grounded architecture question and an architecture decision. Use C.30 to ground the question over selected structures of a described holon. Use C.30.ASV, A.6.F, A.6.M, C.30.LCA, C.30.TFS-REL, C.25, and C.31 to recover the particular structures and characteristics. Later, use C.18 or C.19 for front, archive, or pool treatment, G.5 for selected-set result declaration, E.17 and E.24.PUB for their distinct publication jobs, C.11 for local choice, and the applicable decision pattern for a project decision.
Use C.32 for the constructive middle: building a small set of candidate architecture configurations whose selected structures, allocations, characteristic trade-offs, known losses, source-return conditions, and patterns for the next questions are explicit.
The same middle repeats during improvement. A later criteria-row change, scale-row change, C.16 reading, C.25 or C.31 pressure change, C.31.ASAP scale-preference change, or C.18 or C.19 front, archive, or retained-alternative relation can reopen C.32 when it changes the architecture-characteristic pressure, the selected structures under stress, or the acceptable loss profile. The practitioner then synthesizes another candidate palette; the trigger does not decide the architecture.
The work is to make synthesis real enough that architecture content is available for a later front, comparison, selected-set result declaration, actual publication, or decision.
SoTA-Echoing
These rows show how source practice contributes to C.32. The opening of each second-column entry classifies the source use; the opening of each transfer states its disposition. The blocked-overread column gives the use limit, and the source-currentness boundary below gives the reopen rule. Software-system sources are comparison inputs, examples, or lineage only; they do not narrow C.32 to IT architecture.
Source-currentness boundary. Use each source row only for the C.32 candidate-generation move that the row transfers. If a named standard, guide, book edition, survey, or research line changes that move, recheck the row before using it again. If a receiving FPF pattern named in the row changes how it handles the source family, recheck the row before using it again. If the project needs comparison, selection, selected-set result declaration, actual publication, local choice, decision, evidence, or assurance, leave C.32 and open the pattern for the next question. Rows named as lineage, such as TRIZ ideality, information hiding, or mature DSM lineage, stay lineage until a current source relation is recovered.
Relations
- Builds on:
C.30for the exact described holon, obtainingArchitectureRelationoccurrences, their selectedU.Structureparticipants, and separately identifiedArchitectureClaimcontent;C.30.P,C.30.ASV,A.22,A.6.F,A.6.M,C.32.HCS,C.32.ACS,C.32.ACE,C.25,C.31,C.31.ASAP,C.16,C.16.P,E.22,E.23,C.19.1,C.30.LCA,C.30.TFS-REL,E.18,A.3.4,A.15, and local patterns for recovering source-side architecture referents. - Uses:
C.30.ILCwhen a residual starts the candidate work;C.32.MLAOwhen residual-reducing multilevel framing is being used;C.32.CONWAYwhen exact influence-source and transformed-side architecture content must be co-synthesized without inferring acting, Work, or transformation facts;C.32.FAILwhen a candidate needs repair before explicit comparison, selection, local choice, or decision;C.32.ACEwhen candidate eval results are needed before later comparison or selection;C.33when a source, description, view, decision record, eval report, handoff, or realized observation captures only part of selected structure;C.34when candidate or source structures need preservation adequacy or correspondence adequacy;C.35to recover the kind and assess the adequacy of a generated or discovered result before candidate palette use;C.29when mathematical-lens use is being claimed. - Patterns for the next questions:
A.19.CPMfor explicit comparison claims,A.19.SelectorMechanismfor set-returning selection claims,G.5for selected-set result declaration,C.18andC.19for archive, front, or pool-treatment policy,C.11for fixed local choice,C.30.ADfor architecture-description work,E.17for a source-backed publication face and source return,E.24.PUBfor the publication occurrence and audience availability, andC.32.PADfor project architecture decisions. - P2S docking:
C.32.P2Suses C.32 for the candidate-synthesis stages after problem pressure, selected structures, architecture characteristics, and structural uncertainty have been recovered; C.32 continues to define the candidate palette. - Routes to:
C.32.MWAwhen one usable practice-architecture answer must be synthesized from several structures that do not line up one-for-one; C.32 retains general candidate-palette construction. - Boundary: Use C.32 to construct a candidate architecture palette for one grounded architecture question over selected structures of a described holon. C.35 may support C.32 by recovering a generated or discovered result's kind and assessing its adequacy for candidate-palette use, but C.35 does not select candidates, publish sets, or decide the project architecture. Evidence, assurance, gate, release, work authorization, Method rules, ethical mediation, and causal claims use their own patterns when those claims are being made.
Footer marker
Use C.32 to construct a first useful palette of architecture candidate configurations for one grounded architecture question. Later front-policy, selected-set result declaration, actual publication, local choice, architecture-description, decision, gate, release, and authority-relation claims require their own definitions and tests.
C.32:End
Last Updated: 2026-09-10 — upstream FPF commit a87d0ef4 (github.com/ailev/FPF)