oasislmf.pytools.gul.manager

This file is the entry point for the gul command for the package.

Attributes

logger

item_map_key_dtype

Functions

adjust_byte_mv_size(byte_mv, max_bytes_per_coverage)

Adjust buff size so that the buffer fits the longest coverage

gul_get_items(input_path[, ignore_file_type])

Load the items from the items file.

generate_item_map(items, coverages)

Generate item_map as a hashmap + jagged array; requires items to be sorted.

run(run_dir, ignore_file_type, sample_size, ...[, ...])

Execute the main gulpy worklow.

compute_event_losses(event_id, coverages, ...)

Compute losses for an event.

write_losses(event_id, sample_size, loss_threshold, ...)

Write the computed losses.

Module Contents

oasislmf.pytools.gul.manager.logger[source]
oasislmf.pytools.gul.manager.adjust_byte_mv_size(byte_mv, max_bytes_per_coverage)[source]

Adjust buff size so that the buffer fits the longest coverage

Parameters:

  • byte_mv – numpy byte array

  • max_bytes_per_coverage – max size possible to accommodate all the coverage in byte_mv

Returns:

numpy byte array

Return type:

byte_mv

oasislmf.pytools.gul.manager.gul_get_items(input_path, ignore_file_type=set())[source]

Load the items from the items file.

Parameters:

  • input_path (str) – the path pointing to the file

  • ignore_file_type (Set[str]) – file extension to ignore when loading.

Returns:

Tuple[Dict[int, int], List[int], Dict[int, int], List[Tuple[int, int]], List[int]]

vulnerability dictionary, vulnerability IDs, areaperil to vulnerability index dictionary, areaperil ID to vulnerability index array, areaperil ID to vulnerability array

oasislmf.pytools.gul.manager.item_map_key_dtype[source]
oasislmf.pytools.gul.manager.generate_item_map(items, coverages)[source]

Generate item_map as a hashmap + jagged array; requires items to be sorted.

Items must be sorted by (areaperil_id, vulnerability_id). Builds:

  • hashmap: (areaperil_id, vulnerability_id) → pair_index

  • jagged: pair_index → item indices via ja_offsets / ja_item_idxs

Parameters:

  • items (numpy.ndarray) – 1-d structured array sorted by (areaperil_id, vulnerability_id).

  • coverages (numpy.ndarray) – coverage id to information on items.

Returns:

packed hashmap table. item_map_hm_keys (np.array[item_map_key_dtype]): key storage for the hashmap. item_map_ja_offsets (np.array[oasis_int]): CSR offsets (N_pairs + 1).

Return type:

item_map_hm (np.array[uint8])

oasislmf.pytools.gul.manager.run(run_dir, ignore_file_type, sample_size, loss_threshold, alloc_rule, debug, random_generator, peril_filter=[], file_in=None, file_out=None, ignore_correlation=False, **kwargs)[source]

Execute the main gulpy worklow.

Parameters:

  • run_dir – (str) the directory of where the process is running

  • ignore_file_type (set(str)) – file extension to ignore when loading

  • sample_size (int) – number of random samples to draw.

  • loss_threshold (float) – threshold above which losses are printed to the output stream.

  • alloc_rule (int) – back-allocation rule.

  • debug (bool) – if True, for each random sample, print to the output stream the random value instead of the loss.

  • random_generator (int) – random generator function id.

  • peril_filter (list[int], optional) – list of perils to include in the computation (all included if empty). Defaults to [].

  • file_in (str, optional) – filename of input stream. Defaults to None.

  • file_out (str, optional) – filename of output stream. Defaults to None.

  • ignore_correlation (bool) – if True, do not compute correlated random samples.

  • **kwargs – additional keyword arguments, accepted and ignored so that callers can forward a wider parameter dict.

Raises:

ValueError – if alloc_rule is not 0, 1, or 2.

Returns:

0 if no errors occurred.

Return type:

int

oasislmf.pytools.gul.manager.compute_event_losses(event_id, coverages, coverage_ids, items_data, last_processed_coverage_ids_idx, sample_size, recs, rec_idx_ptr, damage_bins, loss_threshold, losses, alloc_rule, do_correlation, rndms_base, eps_ij, corr_data_by_item_id, arr_min, arr_inv_factor, norm_inv_cdf, arr_min_cdf, arr_norm_factor, norm_cdf, z_unif, debug, max_bytes_per_item, byte_mv, cursor)[source]

Compute losses for an event.

Parameters:

  • event_id (int32) – event id.

  • coverages (numpy.array[oasis_float]) – array with the coverage values for each coverage_id.

  • coverage_ids (numpy.array[int]) – array of unique coverage ids used in this event.

  • items_data (numpy.array[items_data_type]) – items-related data.

  • last_processed_coverage_ids_idx (int) – index of the last coverage_id stored in coverage_ids that was fully processed and printed to the output stream.

  • sample_size (int) – number of random samples to draw.

  • recs (numpy.array[ProbMean]) – all the cdfs used in event_id.

  • rec_idx_ptr (numpy.array[int]) – array with the indices of rec where each cdf record starts.

  • damage_bins (List[Union[damagebindictionaryCsv, damagebindictionary]]) – loaded data from the damage_bin_dict file.

  • loss_threshold (float) – threshold above which losses are printed to the output stream.

  • losses (numpy.array[oasis_float]) – array (to be re-used) to store losses for all item_ids.

  • alloc_rule (int) – back-allocation rule.

  • do_correlation (bool) – if True, compute correlated random samples.

  • rndms_base (numpy.array[float64]) – 2d array of shape (number of seeds, sample_size) storing the random values drawn for each seed.

  • eps_ij (np.array[float]) – correlated random values for damage sampling.

  • corr_data_by_item_id (np.array[correlations_dtype]) – correlation values by item id.

  • arr_min (float) – minimum value of the inverse Gaussian cdf lookup table.

  • arr_inv_factor (float) – scaling factor to index the inverse Gaussian cdf lookup table.

  • norm_inv_cdf (np.array[float]) – inverse Gaussian cdf lookup table.

  • arr_min_cdf (float) – minimum value of the Gaussian cdf lookup table.

  • arr_norm_factor (float) – scaling factor to index the Gaussian cdf lookup table.

  • norm_cdf (np.array[float]) – Gaussian cdf lookup table.

  • z_unif (np.array[float]) – reusable buffer for correlated random values.

  • debug (bool) – if True, for each random sample, print to the output stream the random value instead of the loss.

  • max_bytes_per_item (int) – maximum bytes to be written in the output stream for an item.

  • byte_mv (numpy.array) – byte view of where the output is buffered.

  • cursor (int) – index of int32_mv where to start writing.

Returns:

updated value of cursor, last last_processed_coverage_ids_idx

Return type:

int, int

oasislmf.pytools.gul.manager.write_losses(event_id, sample_size, loss_threshold, losses, item_ids, alloc_rule, tiv, byte_mv, cursor)[source]

Write the computed losses.

Parameters:

  • event_id (int32) – event id.

  • sample_size (int) – number of random samples to draw.

  • loss_threshold (float) – threshold above which losses are printed to the output stream.

  • losses (numpy.array[oasis_float]) – losses for all item_ids

  • item_ids (numpy.array[ITEM_ID_TYPE]) – ids of items whose losses are in losses.

  • alloc_rule (int) – back-allocation rule.

  • tiv (oasis_float) – total insured value.

  • byte_mv (numpy.ndarray) – byte view of where the output is buffered.

  • cursor (int) – index of int32_mv where to start writing.

Returns:

updated values of cursor

Return type:

int