Residential DPLCM breaking

From: Quinn Korbulic <korbulic_at_nmsu.edu>
Date: Mon, 16 Jul 2007 21:14:40 -0600

Hello,

I've recently been able to get UrbanSim running with data from the El Paso
MPO study area. I am incrementally adding more data as I go and have hit
another wall. I ran the model for 30 years w/o development constraints with
the results being less than compelling. So I added constraints and
re-estimated. Now when I run the model it breaks During around the 10th year
with the following error:

            Running Development Project residential Location Choice Model
(from urbansim.models.development_project_location_choice_model): started on
Mon Jul 1
6 20:51:15 2007
                project size: 1704
                Total number of individuals: 35
                res DPLCM chunk 1 out of 1.: started on Mon Jul 16 20:51:15
2007
                    Number of agents in this chunk: 35
                    urbansim.gridcell.is_developable_for_residential_units
 
urbansim.gridcell.developable_maximum_residential_units
                            urbansim.gridcell.total_maximum_development_SSS
 
urbansim.gridcell.is_in_development_type_group_SSS....0.0 sec, total 23192
KB, up 0 KB
                            urbansim.gridcell.total_maximum_development_SSS:
completed...0.1 sec, total 23196 KB, up 4 KB
 
urbansim.gridcell.developable_maximum_residential_units: completed...0.3
sec, total 23196 KB, up 28 KB
                    urbansim.gridcell.is_developable_for_residential_units:
completed...0.3 sec, total 23196 KB, up 32 KB
                    Choice set size: 30
                    Sampling locations ...
 
urbansim.gridcell.developable_minimum_residential_units....0.0 sec, total
22796 KB, up -400 KB
                    shape of weight_array: (35, 6226)
                    WARNING: Rows [0 1] of weight_array sum to 0 or less
                    WARNING: weight array dosen't have enough non-zero
counts, use sample with replacement
                    WARNING: prob_array doesn't sum up to 1, and is
normalized. Sum: 0.0
                    WARNING: There are less or equal non-zero weight (0) in
prob_array than the sample_size (30). Use probsample_replace.
                res DPLCM chunk 1 out of 1.: completed...0.5 sec, total
23432 KB, up 272 KB
            Running Development Project residential Location Choice Model
(from urbansim.models.development_project_location_choice_model):
completed...0.5 sec,
 total 23432 KB, up 272 KB
        Simulate year 2006: completed........6.9 sec, total 23432 KB, up
9352 KB
        Closing log file:
c:/urbansim_cache\run_319.2007_07_16_20_42\year_2006_log.txt
    Starting simulation for year 2006: completed...6.9 sec, total 23432 KB,
up 9352 KB
Start simulation run: completed..............7.0 sec, total 23432 KB, up
9356 KB
Closing log file:
c:/urbansim_cache\run_319.2007_07_16_20_42\run_multiprocess.log
Traceback (most recent call last):
  File "C:\workspace\urbansim\model_coordinators\model_system.py", line 572,
in ?
    s.run(resources)
  File "C:\workspace\urbansim\model_coordinators\model_system.py", line 137,
in run
 
write_datasets_to_cache_at_end_of_year=write_datasets_to_cache_at_end_of_yea
r)
  File "C:\workspace\urbansim\model_coordinators\model_system.py", line 363,
in _run_year
    self.vardict[outputvar] = self.do_process(locals())
  File "C:\workspace\urbansim\model_coordinators\model_system.py", line 441,
in do_process
    return eval(ev)
  File "<string>", line 0, in ?
  File "C:\workspace\opus_core\model.py", line 48, in logged_run_method
    results = run_method(*req_args, **opt_args)
  File "C:\workspace\opus_core\configurable.py", line 38, in
config_run_method
    results = run_method(*req_args, **opt_args)
  File
"C:\workspace\urbansim\models\development_project_location_choice_model.py",
line 61, in run
    LocationChoiceModel.run(self, *args, **kargs)
  File "C:\workspace\urbansim\models\location_choice_model.py", line 121, in
run
    debuglevel=debuglevel)
  File "C:\workspace\opus_core\choice_model.py", line 141, in run
    specification=specification, coefficients=coefficients)
  File "C:\workspace\opus_core\chunk_model.py", line 72, in run
    dataset, **kwargs).astype(result_array_type)
  File "C:\workspace\urbansim\models\location_choice_model.py", line 140, in
run_chunk
    choices = ChoiceModel.run_chunk(self,agents_index, agent_set,
specification, coefficients)
  File "C:\workspace\opus_core\choice_model.py", line 153, in run_chunk
    index = self.get_choice_index(nchoices, agent_set, agents_index,
agentsubset)
  File "C:\workspace\urbansim\models\location_choice_model.py", line 265, in
get_choice_index
    weight=self.weights, resources=self.run_config)
  File "C:\workspace\opus_core\configurable.py", line 38, in
config_run_method
    results = run_method(*req_args, **opt_args)
  File "C:\workspace\opus_core\samplers\weighted_sampler.py", line 119, in
run
    return_indices=True )
  File "C:\workspace\opus_core\sampling_toolbox.py", line 121, in
probsample_noreplace
    return probsample_replace(source_array, sample_size, prob_array=p_array,
return_indices=return_indices)
  File "C:\workspace\opus_core\sampling_toolbox.py", line 55, in
probsample_replace
    cum_prob = ncumsum(prob_array)
  File "C:\workspace\opus_core\misc.py", line 497, in ncumsum
    raise ValueError, "The probability array must sum up to 1. It is " + \
ValueError: The probability array must sum up to 1. It is -1.#IND
Running UrbanSim for year 2006 in new process: completed.................8.3
sec
Traceback (most recent call last):
  File "start_run.py", line 75, in ?
    run_manager.run_run(config)
  File "C:\workspace\opus_core\services\run_server\run_manager.py", line
160, in run_run
    model_system.run_multiprocess(run_resources)
  File "C:\workspace\urbansim\model_coordinators\model_system.py", line 479,
in run_multiprocess
    ForkProcess().fork_new_process(
  File "C:\workspace\opus_core\fork_process.py", line 53, in
fork_new_process
    raise StandardError("Problem with %s" % module_name)
StandardError: Problem with urbansim.model_coordinators.model_system

This is a similar error to one I was encountering with industrial ELCM where
the fix was to increase my incorrect and unreasonably low target vacancies.
In this case my target vacancies are where I want them to be and household
control totals are conservative. My assumption here is that since I
introduced development constraints, there is not enough space to develop
residential units.

I appreciate any suggestions. Thanks again.

Quinn Korbulic
Received on Mon Jul 16 2007 - 20:14:44 PDT

This archive was generated by hypermail 2.2.0 : Mon Jul 16 2007 - 20:14:46 PDT