0% found this document useful (0 votes)
6 views10 pages

Computational Algorithms

The document discusses the Time-Dependent Multimodal Transport Problem (TDMTP) and introduces the Parallel Multimodal Time-Dependent Shortest Path (PMTDSP) Algorithm, which aims to optimize travel paths in dynamic transport networks. It highlights the algorithm's ability to efficiently manage large transport systems through parallel computing, addressing challenges such as traffic congestion and varying travel conditions. The document also evaluates the algorithm's strengths and weaknesses, emphasizing its scalability and performance while noting potential hardware and synchronization issues.

Uploaded by

Nellius
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as DOCX, PDF, TXT or read online on Scribd
0% found this document useful (0 votes)
6 views10 pages

Computational Algorithms

The document discusses the Time-Dependent Multimodal Transport Problem (TDMTP) and introduces the Parallel Multimodal Time-Dependent Shortest Path (PMTDSP) Algorithm, which aims to optimize travel paths in dynamic transport networks. It highlights the algorithm's ability to efficiently manage large transport systems through parallel computing, addressing challenges such as traffic congestion and varying travel conditions. The document also evaluates the algorithm's strengths and weaknesses, emphasizing its scalability and performance while noting potential hardware and synchronization issues.

Uploaded by

Nellius
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as DOCX, PDF, TXT or read online on Scribd
You are on page 1/ 10

Coursework (Assignment weighting 30%)

Module: Computational Algorithms and Paradigms 7COM1078


Student Name: Student ID:
Paper Title:
1. Describe the computational problem and research questions addressed in the paper

(approx. 250 words) [Marks 20]

Computational Problem:

The Time-Dependent Multimodal Transport Problem (TDMTP) requires solution of

discovering optimized travel paths through systems where trip expenditure including duration

and fares and journey comfort levels change throughout the hours. Scalability becomes a

challenge during application of large transport networks because Dijkstra's algorithm

alongside various other conventional algorithms cannot properly manage multimodal

transport constraints that evolve dynamically (Ayed et al., 2011).

Urbans areas experience continuous rise in traffic congestion that causes mobility

performance to decline. The solution complexity arises because the system requires

integrating various transportation methods including trains, buses and private cars while

taking into consideration time-based variations of travel duration and waiting periods and

mode-transition costs (Galvez-Fernandez et al., 2009). The traditional pathfinding methods

experience performance degradation under these circumstances because they cannot manage

real-time data effectively and parallel computing synchronization efficiently. The paper

introduces Parallel Multimodal Time-Dependent Shortest Path (PMTDSP) Algorithm based

on hypergraph decomposition to overcome existing limitations. The parallel strategy divides

computational networks into multimodal subgraph segments by region and transport type

which enables efficient shortest path computations (Zhao, 2010). The algorithm performs

task distribution between multiple processors in a dynamic manner to enhance efficiency

through shared memory synchronization. An improved performance framework results from


decomposing the network for parallel computation processing which allows its use in large-

scale real-world applications.

The research addresses three main objectives: efficient transport management through

parallel computing and shortened computational durations as well as significant scalability

for real-time operation of transport systems.

Research Questions:

i. What is the most efficient method to split a transport network into smaller

computational units?

ii. Which methods enable maximum efficiency when parallelizing the TDMTP solution

process?

iii. The computational performance comparison between the proposed PMTDSP

algorithm and serial algorithms for the PMTDSP process.

iv. What impact does the number of different transportation systems have on algorithm

execution speed together with how networks get divided between parallel processors?

2. Extract the proposed computational algorithm’s pseudo-code (word count depends

on the algorithm) [Marks 10]

Algorithm 1: PMTDSP Algorithm


# PMTDSP Algorithm

def PMTDSP_Algorithm(Hg, s, d, t0):


"""
PMTDSP Algorithm
:param Hg: Graph (R, B)
:param s: Start node
:param d: Destination node
:param t0: Start time
"""

# Initialization
cost_best_solution = float('inf')
g = get_graph_containing_node(s)

# Define destination vertices


destination_vertices = transfer_nodes(g) | bridge_nodes(g) | {d}

# Generate initial task


task = IMT(g, s, destination_vertices, t0)

# Initialize stack with task


stack = set()
stack.add(task)

# Start execution with all executors


executors = get_all_executors()
for executor in executors:
executor.start(stack, d)

# Wait until the stack is empty


while stack:
wait()

Algorithm 2: Executor for Processing Tools


# Executor for Processing Tools
def executor_processing_tools(stack, final_node):
"""
Executor function that processes tasks from the stack.
:param stack: Task stack
:param final_node: Destination node
"""
while True:
if stack: # Check if stack is not empty
task = stack.get_task()

# Compute shortest paths from task source to multiple destinations


paths = SP_OneToMany(task.source, task.destinations, task.time)

for path in paths:


last_node = path.last_node()
arrival_time = path.get_arrival_time()

# If final node is reached, update global solution


if last_node == final_node:
update_global_solution(path, cost_best_solution)

# If arrival time is valid, update node cost and add new task to stack
elif arrival_time <= cost_of_node(last_node) and arrival_time <=
cost_best_solution:
update_node_cost(last_node, arrival_time)
graph = get_graph_containing_node(last_node)

# Get new destination vertices


dest_vertices = transfer_nodes(graph) | bridge_nodes(graph) | {final_node}

# Create new task and add to stack


new_task = IMT(graph, last_node, dest_vertices, arrival_time)
stack.add(new_task)
3. Identify the inputs the algorithm takes and outputs it produces (word count depends

on the algorithm) [Marks 10]

Inputs:

Hg – A hypergraph representation of the transport network

s – Source node (starting point in the network)

d – Destination node

t0 – Initial departure time

M – Set of available transport modes (e.g., bus, train, car, etc.)

E – Set of edges representing transport links

Outputs:

Shortest Path(s) – The computed optimal route(s) considering time-dependent constraints

Cost of Path – The time or cost associated with the computed shortest path

4. Explain the proposed algorithm using simple language (approx. 250 words) [Marks

20]

The PMTDSP algorithm executes its intended function to discover efficient shortest routes

across intricate time-dependent transport network environments. The algorithm splits the

complete network into separate smaller graphs using both transportation type and

geographical zones as its organizational structure (Ayed et al., 2011). A given subgraph

corresponds to a specific transport method including buses or trains and private vehicles and

various subgraphs link through exchange points called bridges.

From its starting point at the source node with initial departure time the algorithm locates

appropriate transport subgraphs. The algorithm utilizes a parallel version of the Dijkstra

algorithm (Galvez-Fernandez et al., 2009) to identify the minimum paths for all destination
nodes which includes transfer nodes. The system distributes each computational

responsibility to processors which enables parallel execution of multiple shortest path

computations simultaneously. As the algorithm reaches transfer nodes or bridges it creates a

new computation task to find the minimum path within the next subgraph. The system

extends itself dynamically which leads to the complete investigation of available pathways. A

shared memory model functions with computation synchronization that maintains the best

solution outputs (Zhao, 2010). Short path computations in extensive networks become more

efficient through the use of parallel computing operations in the PMTDSP algorithm. The

algorithm proves high suitability for real-time transportation route optimization because it

provides swift and precise results essential for urban mobility and congestion mitigation.

5. Analyze the time complexity of the proposed algorithm (word count depends on the

algorithm) [Marks 20]

The modified Dijkstra’s approach serves as the foundation of the algorithm to run shortest

path calculations in O((V + E) log V) worst-case complexity. The parallelized operation

boosts efficiency levels substantially. The algorithm achieves its optimal speedup of O((V +

E) log V / p) with p processors during execution. The efficiency gain depends on network

partitioning intensity together with the amount of transport methods since additional

partitioning levels enable more parallelization opportunities.

6. Critically evaluate the algorithm's strengths and weaknesses (approx.300 words)

[Marks 20]

Strengths:
i. A parallel computing framework enables this algorithm to handle big transport

systems fast and a real-world setting will utilize its high scalability through efficient

large-scale processing (Ayed et al., 2011).

ii. The simultaneous execution of multiple shortest path calculations brought through

parallel processing achieves better performance speed compared to traditional serial

approaches (Galvez-Fernandez et al., 2009).

iii. Shared memory optimization cuts down computational redundancy which allows

efficient resource management across processors according to Zhao (2010).

iv. The flexible design allows the algorithm to operate over all types of real-life

transportation systems and enables its use in various urban mobility situations.

Weaknesses:

i. Hardware and Expert Technology Create installation barriers because specialized

parallel computing hardware needs technical expertise for setup and routine

maintenance.

ii. Parallel execution includes synchronizing operations through specialized mechanisms

that eventually create both slowness in performance alongside performance-related

obstacles.

iii. Real-time flexibility suffers because frequent transport schedule changes force the

algorithm to generate new solutions which impacts its operational speed.

Discussion:

The PMTDSP algorithm demonstrates a fundamental enhancement for TDMTP resolution

since it optimizes the processing speed especially when dealing with extensive problem sizes.

The algorithm benefits from parallel computing because it improves both processing speed

and efficiency when handling extensive datasets which proves useful for urban traffic
management and planning. The algorithm confronts two main limitations from both hardware

limitations and synchronization expenses in the execution process. The system faces two

main issues which include excessive computational costs because of real-time updates as well

as suboptimal parallel execution optimization. The system's widespread deployment requires

evaluation of the available infrastructure limitations. The researchers should add flexibility to

the algorithm by implementing dynamic task scheduling methods in conjunction with AI-

based heuristic optimization techniques for performance optimization.

Note: Please report the word count for sections 1, 4, and 6 [word count: 832] after
completing your assignment.
References

Ayed, H., Habbas, Z., Khadraoui, D., & Galvez-Fernandez, C. (2011). A parallel algorithm

for solving time dependent multimodal transport problem. 2011 14th International

IEEE Conference on Intelligent Transportation Systems (ITSC), 722–727.

https://doi.org/10.1109/itsc.2011.6082973

Galvez-Fernandez, C., Khadraoui, D., Ayed, H., Habbas, Z., & Alba, E. (2009). Distributed

approach for solving time-dependent problems in Multimodal Transport Networks.

Advances in Operations Research, 2009, 1–15. https://doi.org/10.1155/2009/512613

Zhao, P. (2010). Sustainable urban expansion and transportation in a growing megacity:

Consequences of urban sprawl for mobility on the urban fringe of Beijing. Habitat

International, 34(2), 236–243. https://doi.org/10.1016/j.habitatint.2009.09.008

You might also like