update architecture diagram for release

- add test_loading_columns_fw() in test_ue.py;
- update the release date in README.md;
- replace TransOMS with OpenDTA systematically;
- fix legacy comments on including path geometry info for output_columns()
  in tutorial.ipynb and usecases.md;
- polish where is needed.

Author: jdlph

README.md

@@ -5,7 +5,7 @@
 
 Path4GMNS is an open-source, cross-platform, lightweight, and fast Python path engine for networks encoded in [GMNS](https://github.com/zephyr-data-specs/GMNS). Besides finding static shortest paths for simple analyses, its main functionality is to provide an efficient and flexible framework for column-based (path-based) modeling and applications in transportation (e.g., activity-based demand modeling). Path4GMNS supports, in short,
 
-1. finding (static) shortest path between two nodes,
+1. finding (static) shortest paths,
 2. performing either path-based or link-based User-Equilibrium (UE) traffic assignment,
 3. conducting dynamic traffic assignment (DTA) after UE.
 4. evaluating multimodal accessibility and equity,
@@ -24,7 +24,7 @@ Path4GMNS also serves as an API to the C++-based [DTALite](https://github.com/jd
 
 1. **[Tutorial](https://github.com/jdlph/Path4GMNS/blob/master/tutorial/tutorial.ipynb)** written in Jupyter notebook with step-by-step demonstration.
 2. **[Documentation](https://path4gmns.readthedocs.io/en/stable/)** on Installation, Use Cases, Public API, and more.
-3. **[TransOMS](https://github.com/jdlph/TransOMS)** on the C++ equivalent.
+3. **[OpenDTA](https://github.com/jdlph/OpenDTA)** on the C++ equivalent.
 
 We highly recommend that you go through the above [Tutorial](https://github.com/jdlph/Path4GMNS/blob/master/tutorial/tutorial.ipynb), no matter you are one of the existing users or new to Path4GMNS.
 
@@ -35,7 +35,7 @@ $ pip install path4gmns
 ```
 
 > [!IMPORTANT]
-> v0.10.0 introduces the link-based UE that adopts the Frank-Wolfe algorithm with line search. As it does not preserve any paths, each iteration runs faster than the path-based procedure with less memory footprint. Besides, it fixes potential issues on invoking DTALite. Please **discard all old versions**.
+> v0.10.0 introduces the link-based UE that adopts the Frank-Wolfe algorithm with line search. As it does not preserve any paths, each iteration runs faster than the path-based procedure with less memory footprint. Besides, it fixes potential issues of invoking DTALite. Please **discard all old versions**.
 
 > [!WARNING]
 > find_shortest_path() computes the shortest path per travel time rather than distance for v0.9.9.post1 and any earlier versions. See [Issue #58](https://github.com/jdlph/Path4GMNS/issues/58) for details. v0.9.10 offers the correct implementation with the flexibility to switch between time and distance.
@@ -60,7 +60,7 @@ The Python modules are written in **Python 3.x**, which is the minimum requireme
 
 ## How to Cite
 
-Li, P. and Zhou, X. (2025, December 12). *Path4GMNS*. Retrieved from https://github.com/jdlph/Path4GMNS
+Li, P. and Zhou, X. (2025, December 15). *Path4GMNS*. Retrieved from https://github.com/jdlph/Path4GMNS
 
 ## Please Contribute
 

docs/source/imgs/architecture.png

@@ -14,4 +14,4 @@ An easy and smooth installation process by **low dependency** is one of our majo
 
 In order to resolve this issue, we have deprecated node sum and introduced a side-by-side column comparison in Path4GMNS only. As columns between an OD pair are largely different in number of nodes, this comparison can be very efficiently. Slight improvements are actually observed in both running time and convergence gap over the original implementation.
 
-DTALite uses arrays rather than STL containers to store columns. These arrays are fixed in size (1,000), which prevents a fast filtering using the number of nodes as described above. For two (long) columns only different in the last few nodes, this side-by-side comparison has to be continued until the very end and ruins the performance. Thus, we decide **NOT TO ADOPT** this updated implementation to DTALite and leave it to **[TransOMS](https://github.com/jdlph/TransOMS)**.
+DTALite uses arrays rather than STL containers to store columns. These arrays are fixed in size (1,000), which prevents a fast filtering using the number of nodes as described above. For two (long) columns only different in the last few nodes, this side-by-side comparison has to be continued until the very end and ruins the performance. Thus, we decide **NOT TO ADOPT** this updated implementation to DTALite and leave it to **[OpenDTA](https://github.com/jdlph/OpenDTA)**.

docs/source/implnotes.md

@@ -14,7 +14,7 @@ efficient and flexible framework for column-based (path-based) modeling and
 applications in transportation (e.g., activity-based demand modeling).
 Path4GMNS supports, in short,
 
-1. finding (static) shortest path between two nodes,
+1. finding (static) shortest paths,
 2. performing either path-based or link-based User-Equilibrium (UE) traffic assignment,
 3. conducting dynamic traffic assignment (DTA) after UE.
 4. evaluating multimodal accessibility and equity,

docs/source/index.rst

@@ -173,8 +173,8 @@ column_upd_num = 20
 
 pg.find_ue(network, column_gen_num, column_upd_num)
 
-# if you do not want to include geometry info in the output file,
-# use pg.output_columns(network, False)
+# if you want to include path geometry info in the output file,
+# use pg.output_columns(network, True)
 pg.output_columns(network)
 pg.output_link_performance(network)
 ```
@@ -598,7 +598,7 @@ print('writing agent trajectories')
 pg.output_agent_trajectory(network)
 ```
 
-If you have route_assignment.csv (i.e.columns) from a previous run or DTALite, you can bypass find_ue() and directly load it to conduct simulation.
+If you have route_assignment.csv (i.e., columns) from a previous run or DTALite, you can bypass find_ue() and directly load it to conduct simulation.
 
 ```python
 import path4gmns as pg

docs/source/usecases.md

@@ -57,6 +57,23 @@ def test_loading_columns(sample_data_dir, tmp_output_dir):
     output_link_performance(network, output_dir=tmp_output_dir)
 
 
+def test_loading_columns_fw(sample_data_dir, tmp_output_dir):
+    """ test loading columns to set up demand info for find_ue_fw()
+
+    WARNING
+    -------
+    This is just for testing purpose. It is NOT recommended to use!
+    """
+    network = read_network(input_dir=sample_data_dir)
+    load_columns(network, input_dir=tmp_output_dir)
+
+    rel_gap_tolerance = 1e-2
+    rel_gap = find_ue_fw(network, max_iter=20, rel_gap_tolerance=rel_gap_tolerance)
+    assert rel_gap <= rel_gap_tolerance
+
+    output_link_performance(network, output_dir=tmp_output_dir)
+
+
 def test_mixed_invoking1(sample_data_dir):
     """ test resolution on issue #51 (https://github.com/jdlph/Path4GMNS/issues/51)
     """

tests/test_ue.py

@@ -79,7 +79,7 @@
    "id": "annoying-berry",
    "metadata": {},
    "source": [
-    "Navigate to the target data set directory. In this tutorial, we will use the [data set](https://github.com/jdlph/Path4GMNS/tree/dev/tests) where this Jupyter notebook is located for simplicity. You can check your present working directory (PWD) by the following commnad."
+    "Navigate to the target data set directory. In this tutorial, we will use the [directory](https://github.com/jdlph/Path4GMNS/blob/dev/tutorial) where this Jupyter notebook is located for simplicity. You can check your present working directory (PWD) by the following commnad."
    ]
   },
   {
@@ -296,7 +296,7 @@
     "\n",
     "#### 3.3.2 Path-Based UE\n",
     "\n",
-    "With the required demand file, then you can conduct path-based UE traffic assignment via the following lines."
+    "With the required demand file, you can conduct path-based UE traffic assignment via the following lines."
    ]
   },
   {
@@ -316,8 +316,8 @@
     "# path-based UE only\n",
     "pg.find_ue(network, column_gen_num, column_upd_num)\n",
     "\n",
-    "# if you do not want to include geometry info in the output file,\n",
-    "# use pg.output_columns(network, False)\n",
+    "# if you want to include path geometry info in the output file,\n",
+    "# use pg.output_columns(network, True)\n",
     "# output column information to route_assignment.csv\n",
     "pg.output_columns(network)\n",
     "# output link performance to link_performance.csv\n",