newmark 1.14.0

• The BM19 pulse-motion D50 equation now uses the intercept -0.148 from the official erratum when Ts < 0.1 s and PGV > 150 cm/s, replacing 2.480. Other regimes and the ordinary-motion equations used by the project tables and hazard integration are unchanged.
• Licence changed from MIT to AGPL (>= 3). Files in R/ carry SPDX licence headers. LICENSE.note has been removed.
• Column site spectra are reused at VS30 rounded to the nearest 1 m/s. ShearDraws retains its stored VS30 values. Mean-row site response uses the row’s stored VS30 for PGATable, the common ky grid and reference metadata.
• buildDnTables() requires draws, the paired Ts and VS30 of ShearDraws. The NS movements of each pair cycle through its columns in realization order. Each column supplies its site response, spectral sampling periods and displacement period. The movement count, model weights, table schemas and empirical exceedance calculation are unchanged.
• PGA, Sa13 and Sa15 are centered once over the complete cell sample on the column means weighted by their assigned movements. Column subsets are not centered separately. Identical columns reproduce the intensities and model deviates of a single-column sample with the same seed and NS. Public sampleIntensities() and sampleSpectra() calls are unchanged.
• The DnTable project step reads ShearDraws.Rds. Generate it explicitly with the ShearDraws step if absent. ShearTable mean rows continue to define the common ky grid, PGATable and reference metadata; PGA.s holds each movement’s sampled PGA. The hazard-integration functions are unchanged.

newmark 1.13.0

• integrateDnHazard() and integrateDnHazardProfiles() accept BM19 among the weights, with the spectral acceleration at its degraded period 1.3 Ts and its ordinary-motion equation, as the DnTable products evaluate it. The project route keeps BM19 for shallow crustal regimes (subduction false), so the hazard integration can now take the model set of such a project. Every other model integrates as before.

newmark 1.12.0

• New function buildShearDraws(): the realizations of the soil columns that buildShearTable() summarizes, 100 per geometry and material, each with the period of the truncated shear beam evaluated on its own mo and VSo with the rounding of the summary rows, together with the shear table summarized from the same draws. Realizations beyond 100 (NS) are drawn in a second pass after every pair has its first 100, so the first 100 and the shear table do not depend on NS. As a rule of the product, a realization whose fitted mo is negative is returned with mo forced to 0 and with an and Ts evaluated at that 0, and a message counts them, while the shear table summarizes the fitted mo; a realization with mo of 2 or more, where the first-mode root is not defined, stops the function naming it. buildShearTable() keeps its arguments, its random numbers and its values.
• The new ShearDraws step of process follows ShearTable and runs only when selected. It requires seed, draws the realizations again from the contract and the seed, and publishes ShearDraws.Rds at the root of path.tables only when their summary equals the published ShearTable.Rds value for value; it never writes ShearTable.Rds. Its message reports how many realizations it publishes with a negative mo forced to 0. The contract key NS.shear (a whole number of at least 100, 100 when absent) sets the realizations per pair and is read by this step alone. Listing and query know the table.
• New function integrateDnHazardProfiles(): the annual displacement exceedance rate of integrateDnHazard() mixed over site profiles, each a pair of Ts and VS30 with an optional weight, and over yield coefficients with weights, under a declared independence of profile and yield coefficient. The Poisson probabilities, the band and the rare-tail bound are formed on the mixed rates. It reports each component, the weighted fraction of components with ky outside a calibration range, and the Monte Carlo standard error of the mixture over the profiles (mixtureError, bandMixtureError), which is not an error bound. It is an operation of the displacement family, where the command takes ky as a table of ky and w. integrateDnHazard() does not change.
• The exports that the installer verifies and newmark doctor checks now include integrateDnHazard() and listDisplacementFamilies(), which the command already called but which that list had left out.
• The products ShearTable, DnTable, kmaxTable and DnPOETable and their columns do not change.

newmark 1.11.0

• buildDnPOETable() counts the weighted empirical fraction of stored pointwise displacement realizations exceeding DaH * Hs, by site, demand, geometry, material, yield acceleration and threshold. assembleDnPOETables() combines the site tables. The fraction is conditional on the demand (return period or MCE) and inherits the positive lognormal branch represented by the flexible-model draws; stored zeros, including SR08 underflow, count as non-exceedance.
• The new DnPOETable step follows kmaxTable in process and can run alone from the accepted ShearTable and DnDraws, without PGATable or a new simulation. It publishes per-site and global DnPOETable products, available through listing and query. Both functions are CLI operations. This product does not integrate displacement hazard or add an extrapolated column.

newmark 1.10.0

• New function integrateDnHazard(): the annual rate at which the Newmark displacement of a sliding mass with a fixed yield coefficient exceeds a threshold, integrated over one site’s rock hazard curves with the site factor, the displacement regressions and their weights; it reports the Poisson probability over the exposure, the contribution of a band of return periods, a bound of the rare tail beyond the curves, the calibration range of the yield coefficient of each model and the numerical diagnostics of its quadratures. The command exposes it as the other operations of the displacement family.
• A tables argument given as a JSON array whose members are all file references is decoded like the object form; it was refused as if it were neither an array nor an object of tables.
• A file that an argument of the command refers to and that does not exist is a usage error, exit 2 with the newmark: line alone, as an unreadable referenced .json file already was; it exited 1 as a failed operation.
• sampleSoilLayers() counts the layers of a column as the ceiling of its height over the layer thickness with a stated rounding tolerance. The quotient of two decimal lengths can land a few units of double precision above an integer, as with 2.1 m divided by 0.3 m, and the column got one layer more than its height holds. Only a quotient that exceeds an integer by at most four times the double-precision epsilon, relative to its value, changes its count.
• The tests of the numerical periods and roots that compare two equal evaluations accept a difference in the last digit: under the Intel MKL library the two evaluations differ there, as CRAN’s additional check of 1.9.0 showed.
• The tests that reproduce the products of one hundred realizations byte for byte, the dense quadratures of integrateDnHazard() and the steep upper tail of integrateSiteSpectrum() run outside CRAN, in continuous integration and locally; on CRAN’s Windows server the whole check took more than ten minutes with them.

newmark 1.9.1

• The library signals its notes, progress and warnings as classed conditions of the AOM family (moduleNote, moduleProgress and moduleWarning) with the fields operation, phase, unit, path, count and total. Their texts lose the prefixes dry run:, <step>:, evaluateSiteFactor: and >>, and the kmax warnings name their table in parentheses. A failure after a publication of processProject() has the classes newmarkPartialPublication, modulePartialPublication and moduleFailure. The command reads its arguments from the launcher, whose renderer writes these conditions on stderr and takes --verbose and --quiet.
• The private predicates .isName() and .isKmaxIdentifier() become one predicate, .isIdentifier(), and .dnDemandsOK() and .hasKmaxDemands() become .hasDemands(). They are internal; no exported function changes its behavior.
• The test files are renamed to the test-<function>.R form; for example, test-kmax-table.R is now test-buildKmaxTable.R.
• The package help page, ?newmark, describes all the families of exported functions.
• New dataset VancouverSpectra: the uniform hazard spectra of the 2020 National Building Code of Canada for Vancouver on rock, from the seismic hazard tool of Natural Resources Canada. The examples of convertAEP(), evaluateUniformHazard() and evaluateSiteSa() read it.
• New function listDisplacementFamilies(): the families of displacement models (ensemble, rigid and flexible) that the displacement and kmax tables summarize as trees, each with the identifiers of its models. The command line runs it as the operation of the same name.
• ?newmark loses the package name from its title. The help pages of estimatePGV() and estimateAI() cite Verri (2023), the source of the rock records their relations were fitted on. On CRAN, the tests that repeat a step of processProject() skip; one run of each step and the dry run still execute there.
• Mw.gmdp takes one magnitude for every demand, as before, or an object that names each return period of TR.gmdp and MCE with its own magnitude; buildDnTables() takes the same as a named vector in Mw. The DnTable step applies the magnitude of each demand. A contract with one magnitude gives the same tables as before.
• newmark doctor prints one JSON report of the installation, as every module of the family does: schema, package, version, summary and the checks library, exports, packages, tools and receipt, each with name, required, status and detail; it exits 1 when a required check fails, and doctor --help lists the checks. describeModule(resources) returns that report. newmark --version prints two lines, the package with its version and the build commit.
• A .json file that an argument of the command refers to and that cannot be read as JSON is a usage error, exit 2 with the newmark: line alone, as an invalid INPUT.json already was; it exited 1 as a failed operation.
• The accumulations that feed a discrete choice, a rounding or a cut in the site mixture, the site spectrum, the soil profile and the PGA and displacement tables sum in plain double on every platform, as the kmax, displacement and mixture summaries have since 1.9.0; the tables are unchanged on macOS, where the reference products were recorded.

newmark 1.9.0

This release replaces the interface of newmark 1.3.0, the previous CRAN release. The package was rebuilt around functions named by what they compute, one family at a time: soil profiles, site response, hazard inputs, displacements and the pseudostatic coefficient.

Breaking changes

• The 34 exports of 1.3.0 are gone. The two that had survived as compatibility adapters, buildGMDP() and fitSaFMixture(), are retired in this release: fitSaFMixture() is replaced by integrateSiteMixture(), and the hazard inputs that buildGMDP() staged are read with readHazardWorkbook(), convertAEP() and evaluateUniformHazard().
• Removed exports of 1.3.0 with no function of the same name: approx.spline, buildQSpline, checkUHS, designUHS, Dn_AM88, Dn_BM17, Dn_BM19, Dn_BT07, Dn_JB07, Dn_SR08, F_ST17, F_ST20, fitDnCurve, fitDnModel, fitDnScenario, fitDnScenarios, fitKmaxScenario, fitKmaxScenarios, fitModel.Ts, fitSaF, geSiteTable, getCylinderRoots, getDnKy, getKyLimits, getSiteProperties, interpolateSaTable, invertDnDraws, rhoBJ, SaF_ST17, sampleSaCorr, SIDtoVs30, Vs30toSID.
• The site amplification models ST17 and ST20 are selected through the model argument of evaluateSiteFactor() and the site response functions. The displacement models AM88, BM17, BM19, BT07, JB07 and SR08 are selected through selectDisplacementModels() and evaluated by evaluateRigidDisplacement(), evaluateFlexibleDisplacement() and solveKmax().
• The dataset SiteTable is no longer shipped. CylinderRoots, ShearModelParameters and SiteClass remain.
• Depends is R (>= 4.2). digest, Hmisc, randomForest, rpart, stringr and triangle are no longer imported; readxl moves to Suggests; jsonlite and parallel are imported.

New interface

• Soil profiles: simulateSoilProfiles(), sampleSoilLayers(), evaluateSoilLayers(), characterizeProfile(), summarizeProfiles(), estimateVS30(), classifySite(), estimateRayleighPeriod(), estimateWeightedRayleighPeriod(), evaluateBeamPeriod(), propagateBeamPeriod(), propagateSitePeriod(), solveColumnPeriod(), solveProfilePeriods(), solveWedgePeriod(), buildRootModel(), evaluateRoot(), buildGeometry(), buildShearTable().
• Site response: evaluateSiteFactor(), evaluateSiteSa(), integrateSiteMixture(), integrateSiteSpectrum(), refineSiteMixture(), refineSiteSpectrum(), simulateSiteSpectrum(), summarizeMixture(), interpolateSpectrum(), buildQuantileSpline(), correlateResiduals(), sampleSpectra().
• Hazard inputs: readHazardWorkbook(), convertAEP(), evaluateUniformHazard(), buildPGATable(), sampleIntensities(), estimateAI(), estimatePGV().
• Displacements and pseudostatic coefficient: selectDisplacementModels(), evaluateRigidDisplacement(), evaluateFlexibleDisplacement(), evaluateDisplacements(), summarizeDisplacements(), summarizeEnsemble(), buildDnTables(), assembleDnTables(), buildKyGrid(), buildKyLimits(), solveKmax(), summarizeKmax(), buildKmaxTable(), assembleKmaxTables().
• Projects: processProject(), listProjectTables(), queryProjectTable(), describeModule().