OpenQuake Engine 3.15.0

[Michele Simionato (@micheles)]

• Added a check on sum(srcs_weights) == 1 for mutex sources

[Kendra Johnson (@kejohnso)]

• Fixed disaggregation by lon, lat in presence of multiFaultSources

[Michele Simionato (@micheles)]

• Replaced command oq download_shakemap with oq shakemap2gmfs
• Raised an error for missing required IMTs in ShakeMap grid files
• Extended the custom_site_id to 8 characters
• Restricted the accepted characters in risk IDs
• Extended disagg_by_src to mutually exclusive sources (i.e. Japan)
  and managed "colon" sources specially

[Anne Hulsey (@annehulsey)]

• Contributed Mag_Dist_TRT and Mag_Dist_TRT_Eps disaggregations

[Michele Simionato (@micheles)]

• Internal: added a way to disable the DbServer from openquake.cfg or
  by setting OQ_DATABASE=local
• Implemented total_losses, even for insurance calculations
• Optimized "computing risk" in the event_based_risk calculator (~30% faster)
• Changed the magnitude binning formula, thus fixing some disaggregation
  calculations (for instance when there is a single magnitude for a TRT)
• Changed the aggrisk/aggcurves exporters in presence of insurance losses
• Internal: changed how avg_losses, src_loss_table and agg_curves-stats are
  stored to simplify the management of secondary losses
• Internal: we have now repeatable rupture IDs in classical PSHA

[Pablo Iturrieta (@pabloitu)]

• Added support for negative binomial temporal occurrence models

[Marco Pagani (@mmpagani), Michele Simionato (@micheles)]

• Added support for disaggregation in case of mutually exclusive sources

[Michele Simionato (@micheles)]

• Fixed error message when trying to compute disagg_by_src with too many
  sources: in some cases, it contained a misleading reference to point sources
• Reorganized the Advanced Manual; changed the theme to be consistent with
  the OpenQuake manual
• Internal: added command oq db engine_version
• Added a check for required site parameters not passed correctly
• Fixed ps_grid_spacing approximation when the grid is degenerate
• Logging a warning when starting from an old hazard calculation
• The extra fields of the site collection were lost when using --hc

[Julián Santiago Montejo Espitia]

• Implemented the Arteta et al. (2021) subduction model for Colombia

[Michele Simionato (@micheles)]

• Added host field to the job table (dbserver restart required)
• --exports=csv was not honored for the realizations output; now it is

[Paul Henshaw (@pslh), Sandra Giacomini]

• Ported the OpenQuake manual from latex to reStructuredText format

[Michele Simionato (@micheles)]

• Entered automatically in sequential mode if there is not enough memory
• Raised an early error for missing risk IDs in the vulnerability files
• Changed the definition of aggrisk again to ensure consistency with the
  average losses

[Tom Son (@SnowNooDLe)]

• Added width and hypo_depth estimation to Campbell and Bozorgnia (2014)

[Michele Simionato (@micheles)]

• Improved the precision of the ps_grid_spacing approximation
• Added a check for missing mags when using GMPETables
• Fixed a bug in upgrade_nrml -m for point sources with different usd/lsd
• Automatically discard irrelevant TRTs in disaggregation calculations

[Ashta Poudel, Anirudh Rao (@raoanirudh), Michele Simionato (@micheles)]

• Added the ability to run connectivity analysis in event_based_damage
  and scenario_damage calculation with an appropriate exposure

[Tom Son (@SnowNooDLe), Michele Simionato (@micheles)]

• Added ztor estimation to Campbell and Bozorgnia (2014)

[Michele Simionato (@micheles)]

• Internal: removed REQUIRES_COMPUTED_PARAMETERS
• Using PointMSR when the reqv approximation is enabled
• Internal: changed the rupture storage for classical calculations
• Optimized rupture instantiation for point sources
• Optimized distance calculations for point sources

[Tom Son (@SnowNooDLe), Claudio Schill]

• Simple performance improvement of Kuehn et al. 2020 model

[Michele Simionato (@micheles)]

• Changed again the string representation of logic tree paths and added
  an utility hazardlib.lt.build to build trees from literal lists
• Fixed the field source_info.trti in the datastore to point to the
  correct tectonic region type index and not to zero
• Added a check for inconsistent IDs across different kinds of risk functions
  (i.e. fragility and consequence)
• Fixed a logging statement that could run out of memory in large calculations
• Optimized iter_ruptures for point sources by vectorizing the generation of
  planar surfaces by magnitude, nodal plane and hypocenter

[Tom Son (@SnowNooDLe)]

• Implemented a missing piece in Chiou & Youngs (2014) model
  Predicted PSA value at T ≤ 0.3s should be set equal to the value of PGA
  when it falls below the predicted PGA

[Marco Pagani (@mmpagani)]

• Added the possibility of disaggregating in terms of epsilon*
• Added a method to compute the cross-correlation matrix
• Added Hassani & Atkinson (2018)
• Added Hassani & Atkinson (2020)

[Michele Simionato (@micheles)]

• Fixed disaggregation returning NaNs in some situations with
  nonParametric/multiFaultSources
• Bug fix: not storing far away ruptures coming from multiFaultSources
• Implemented CScalingMSR
• Optimized context collapsing in classical calculations
• Setting ps_grid_spacing now sets the pointsource_distance too
• Saving memory in preclassical calculations on machines with 8 cores or less
• Changed the magnitude-dependent maximum_distance feature to discard
  ruptures below minmag and above maxmag
• Added the ability to estimate the runtime of a calculation by using
  the --sample-sources option
• Fixed a wrong formula in modifiable_gmpe.add_between_within_stds
• Reduced the stress on the memory in classical calculations, thus
  improving the performance
• Setting the truncation_level to the empty string is now forbidden;
  some GMFs calculations not setting truncation_level can now give
  different results since truncation_level=None is now replaced with
  truncation_level=99