No full-text available
To read the full-text of this research,
you can request a copy directly from the authors.
The Traveling Salesman Problem: A Computational Study
Abstract
This book presents the latest findings on one of the most intensely investigated subjects in computational mathematics--the traveling salesman problem. It sounds simple enough: given a set of cities and the cost of travel between each pair of them, the problem challenges you to find the cheapest route by which to visit all the cities and return home to where you began. Though seemingly modest, this exercise has inspired studies by mathematicians, chemists, and physicists. Teachers use it in the classroom. It has practical applications in genetics, telecommunications, and neuroscience. The authors of this book are the same pioneers who for nearly two decades have led the investigation into the traveling salesman problem. They have derived solutions to almost eighty-six thousand cities, yet a general solution to the problem has yet to be discovered. Here they describe the method and computer code they used to solve a broad range of large-scale problems, and along the way they demonstrate the interplay of applied mathematics with increasingly powerful computing platforms. They also give the fascinating history of the problem--how it developed, and why it continues to intrigue us.
... At present, centralized planning can be divided into precise optimization and approximate optimization. Precise optimization mainly includes branch definition [23], dynamic programming [24], and so on. Although precise optimization can obtain the global optimal solution, the computation time of the algorithm is long and increases dramatically with the complexity of the problem. ...
... The hashed printing areas of the same robot are merged into the last combination with only the same robot task, resulting in an integrated printing combination. Taking an optimized path combination as an example: ( [13,22,1,20], [0], [16,3], [2], [4], [10], [12], [9], [0], [6,21], [7], [5], [14], [15], [11,17], [0], [23,8,18], [24], [19]), 1-15 represents task A, 16-24 represents task B, 0 represents the rotation of the platform, and [13,22,1,20] represents a round of cooperative printing region. Due to the spatial constraint strategy of the global optimization algorithm, [2], [4], [10], [12], [9], [7], [5], [14], and [15] belonging to the same task are distributed in different cooperative combinations. ...
... The path combinations of the odd-numbered subregion are moved to the front of the first rotation of the platform, and the path combinations of the even-numbered subregions are moved to the back of the first rotation of the platform. Taking an optimized path combination as an example: ( [13,22,1,20], [0], [16,3], [0], [6,21], [2,4,10,12,9,7,5,14,15], [11,17], [0], [23,8,18], [24,19]). The optimized combination sequence is as follows: ( [13,22,1,20], [6,21], [2,4,10,12,9,7,5,14,15], [11,17], [0], [16,3], [23,8,18], [24,19]). ...
To improve the efficiency of multi-material additive manufacturing and enhance the safety of multi-robot cooperative printing with uncertain execution delays, a dual-robot cooperative path planning method is proposed for layer-by-layer printing. In the proposed algorithm, the description of the printing region is reconstructed by simply using the rectangular envelope region and a two-dimensional directed line segment. The adjacency list of the printing region is established to guide the optimization direction. Therefore, redundant information about the printing region is effectively removed, which is conducive to the optimization of the problem. A multi-round cooperation strategy with multiple synchronous starting points is proposed to accommodate uncertain execution delays by separating the space of the dual-robot printing area, so as to avoid the potential collision risk of dual-robot. To further optimize the printing efficiency, local strategies are used to reduce the makespan. Hence, a better printing order can be obtained, and states of cooperative and non-cooperative printing can be unified. In addition, the corresponding NC control strategy is designed for the industrial application of the cooperative strategy. The simulation result shows that the method proposed in this paper can effectively reduce the makespan of dual-robot cooperative additive manufacturing, and accommodate the uncertain execution delays of the dual-robot.
... The TSP, a well-known NP-hard problem in computer science and operations research, involves finding the shortest possible route that visits a set of cities exactly once and returns to the starting city [7]. As obtaining the optimal solution for large-scale instances of the TSP is computationally intractable, the MST algorithm serves as an effective heuristic method to find near-optimal solutions within a reasonable computational time [1], [22]. ...
... Notably, in practice, the MST tour often performs better than the theoretical worst-case bound, making it a valuable heuristic for large-scale TSP instances [1], [9], [19]. While the MST algorithm is not guaranteed to find the optimal solution for the TSP, it serves as a heuristic approach to obtain a reasonably good approximation within a reasonable computational time [1], [9]. ...
... Notably, in practice, the MST tour often performs better than the theoretical worst-case bound, making it a valuable heuristic for large-scale TSP instances [1], [9], [19]. While the MST algorithm is not guaranteed to find the optimal solution for the TSP, it serves as a heuristic approach to obtain a reasonably good approximation within a reasonable computational time [1], [9]. The MST algorithm is often used as a starting point or a component in more advanced algorithms or metaheuristics designed to tackle the TSP more effectively [7], [13]. ...
Search and rescue missions are critical endeavours aimed at locating survivors trapped in the aftermath of various calamities such as earthquakes, landslides, tsunamis etc. The time efficiency required for these missions is pivotal in saving lives, prompting the adoption of robotics with advanced technology to expedite operations. Effective planning and strategizing play key roles in enhancing a robot's efficiency, particularly in prioritizing areas of need. Typically, search and rescue missions often present a high level of complexity due to a multitude number of factors to be considered when planning and optimizing operations. Factors such as determining hot-spot regions, harsh weather conditions, terrain complexity, environmental hazards, and time sensitivity, just to a mention few, introduce intricacies to the task of effective planning. For instance, a factor such as environmental conditions is considered dynamic as it may change from time to time. Additionally, the criticality of different regions may change dynamically as soon as the information is made available, further complicating the task for both rescue robots and even fast responders. This research addresses the challenge of optimizing path planning in search and rescue missions, treating it as a variant of the travelling salesman problem and proposes a hybrid technique that incorporates both algorithmic and non-algorithmic techniques, to tackle the problem. The hybrid technique leverages a tweaked version of the U-net neural network that is trained on two pieces of information: static data, data that encircles topological map data such as slope and dynamic data deduced from map findings data that would be vital in singling out hotspots and priority regions on the map. Blending these two pieces of information, an amalgamated cost, a value that incorporates both priority and cost of traversal, is determined to aid robot path planning decisions. Training a neural network on this data enables it to predict the associated amalgamated cost map, permitting real-time path planning through a real-time amalgamated cost map that is generated during the robot's mission.
... However, the origin of the name is not clear, as there is no documentation concerning a specific author. What the scientific community seems to agree on is that the name was minted in the 1930s at the Princeton University (Applegate et al., 2007). ...
... TSP has several applications such as planning, logistics, manufacture of microchips. With slight modifications, is found as a sub-problem in many areas, such as DNA sequencing (Applegate et al., 2007). In these applications, the cities represent, customers, soldering points, or DNA fragments, and the distances represent traveling times, cost, or a similarity measures between DNA fragments. ...
... In many applications, additional constraints, such as limited resources or time windows, may be imposed. Furthermore, several different approaches have emerged for solving the TSP, including exact algorithms, heuristics, metaheuristics and even the Hybridization of them (Applegate et al., 2007;Anbuudayasankar et al., 2014). ...
In this thesis, we propose mathematical optimization models and algorithms for variants of routing problems. The first contribution consists of models and algorithms for the Traveling Salesman Problem with Time-dependent Service times (TSP-TS). We propose a new Mixed Integer Programming model and develop a multi-operator genetic algorithm and two Branch-and-Cut methods, based on the proposed model. The algorithms are tested on benchmark symmetric and asymmetric instances from the literature, and compared with an existing approach, showing the effectiveness of the proposed algorithms. The second work concerns the Pollution Traveling Salesman Problem (PTSP). We present a Mixed Integer Programming model for the PTSP and two mataheuristic algorithms: an Iterated Local Search algorithm and a Multi-operator Genetic algorithm. We performed extensive computational experiments on benchmark instances. The last contribution considers a rich version of the Waste Collection Problem (WCP) with multiple depots and stochastic demands using Horizontal Cooperation strategies. We developed a hybrid algorithm combining metaheuristics with simulation. We tested the proposed algorithm on a set of large-sized WCP instances in non-cooperative scenarios and cooperative scenarios.
... HE Traveling Salesman Problem (TSP) is one of the most prominent members of the rich set of well-known combinatorial optimization problems with real life application potential. It is a Nondeterministic Polynomial hard (NP-hard) problem [1]. Given a set of cities (graph nodes) along with the costs of travel between each pair of them (the costs or lengths assigned to the edges), the TSP goal is to find the cheapest (shortest) way of visiting all the cities exactly once, and then returning to the starting point [1] [2]. ...
... It is a Nondeterministic Polynomial hard (NP-hard) problem [1]. Given a set of cities (graph nodes) along with the costs of travel between each pair of them (the costs or lengths assigned to the edges), the TSP goal is to find the cheapest (shortest) way of visiting all the cities exactly once, and then returning to the starting point [1] [2]. ...
... The task of the TSP is to find a route through a given set of cities with the shortest possible length (cost). Mathematically, it means to find the shortest Hamiltonian tour in a graph [1]. ...
There are many factors that affect the performance of the evolutionary and memetic algorithms. One of these factors is the proper selection of the initial population, as it represents a very important criterion contributing to the convergence speed. Selecting a conveniently preprocessed initial population definitely increases the convergence speed and thus accelerates the probability of steering the search towards better regions in the search space, hence, avoiding premature convergence towards a local optimum. In this paper, we propose a new method for generating the initial individual candidate solution called Circle Group Heuristic (CGH) for Discrete Bacterial Memetic Evolutionary Algorithm (DBMEA), which is built with aid of a simple Genetic Algorithm (GA). CGH has been tested for several benchmark reference data of the Travelling Salesman Problem (TSP). The practical results show that CGH gives better tours compared with other well-known heuristic tour construction methods.
... In the setting of graphs for metric TSP, we can find the initial tour using, e.g., the Christofides-Serdyukov algorithm [12,34,36] and use a k-OPT heuristic [3] to further improve the solution. The k-OPT heuristic solves a decision problem where we ask if a given tour can be improved by replacing k edges in the tour with k new edges. ...
... Generally, We obtain a set of cycles and an s-t path covering all the vertices inG i . (b) Create a new undirected graph G c i where each vertex represents a cycle obtained in Step (1a) above, and there is an edge between vertices if the corresponding cycles can be joined into a bigger cycle by performing a 2-opt exchange [3] at a unit square between them. If there are multiple such unit squares, we pick one that adds the minimum weight to the joined cycle. ...
... These MIP instances could be solved independently, and hence in an embarrassingly parallel fashion. Alternatively, our framework allows the use of approximation algorithms or heuristics to solve the subproblems instead of solving the MIP model to optimality [3]. We could also reuse optimal solutions for cells that reoccur across multiple layers. ...
We explore efficient optimization of toolpaths based on multiple criteria for large instances of 3D printing problems. We first show that the minimum turn cost 3D printing problem is NP-hard, even when the region is a simple polygon. We develop SFCDecomp, a space filling curve based decomposition framework to solve large instances of 3D printing problems efficiently by solving these optimization subproblems independently. For the Buddha model, our framework builds toolpaths over a total of 799,716 nodes across 169 layers, and for the Bunny model it builds toolpaths over 812,733 nodes across 360 layers. Building on SFCDecomp, we develop a multicriteria optimization approach for toolpath planning. We demonstrate the utility of our framework by maximizing or minimizing tool path edge overlap between adjacent layers, while jointly minimizing turn costs. Strength testing of a tensile test specimen printed with tool paths that maximize or minimize adjacent layer edge overlaps reveal significant differences in tensile strength between the two classes of prints.
... The Travelling salesman problem (TSP) is prevalent in combinatorial optimisation (CO) as it involves searching for a route that traverses each node exactly once and has minimal route/tour length. It has numerous applications in telecommunications, circuit board design, DNA sequencing, transportation and theoretical computer science [1]. It is an NP-Hard problem, i.e., there are instances for which finding an optimal solution is a time-consuming process. ...
... It is an NP-Hard problem, i.e., there are instances for which finding an optimal solution is a time-consuming process. Traditional approaches to tackle such complex optimisation problems employ two main strategies: exact algorithms [1], and heuristics [2]. Exact algorithms are based around enumeration and can solve TSP optimally, e.g., branch-andbound or integer programming formulations, but they generally cannot scale to larger instances [1]. ...
... Traditional approaches to tackle such complex optimisation problems employ two main strategies: exact algorithms [1], and heuristics [2]. Exact algorithms are based around enumeration and can solve TSP optimally, e.g., branch-andbound or integer programming formulations, but they generally cannot scale to larger instances [1]. On the other hand, heuristics employ problem-specific knowledge and carefully engineered approaches and parameters to find near-optimal solutions. ...
The Travelling Salesman Problem is a classical combinatorial optimisation problem (COP). In recent years, learning to optimise approaches have shown success in solving TSP problems. However, they focus on one type of TSP instance, where the points are uniformly distributed in Euclidean spaces (easy instances). Such approaches cannot generalise to other embedding spaces that represent various levels of difficult instances, e.g., TSP instances where the points are distributed in a non-uniform manner and spherical spaces. Obtain optimal solutions for easy instances is achievable and can be used as training data to solve various TSP instances. However, acquire optimal solutions for complex TSP instances is difficult and time-consuming. Hence, this paper introduces a new learning-based approach based on a convolutional neural network combined with a Long Short-Term Memory, referred to as the non-Euclidean TSP network (NETSP), that utilises randomly generated instances (easy instances) to solve various common TSP instances (complex TSP instances). We have demonstrated its superiority over state-of-the-art methods for various TSP instances. We performed extensive experiments that indicate our approach generalises across many instances and scales to larger instances.
... To address these challenges, in this work, we present Threshold Filtering Packing (TFP), a new packing approach that packs sequences of related yet diverse samples, encouraging context richness and reasoning across sample boundaries. Specifically, we employ a greedy algorithm inspired by the Traveling Salesman Problem (TSP) (Applegate et al., 2006) to efficiently map out a path for segmentation into multiple packs. TFP is implemented to further refine these packs by ensuring that overly similar samples are not grouped together. ...
... A basic approach involves using a greedy algorithm to select samples with the smallest Euclidean distance to their corresponding embeddings, ensuring each sample is included only once. This is essentially a greedy algorithm for the TSP (Applegate et al., 2006). Intuitively, k-NN can select the same data point multiple times when retrieving the nearest neighbors, whereas TSP selects each data point only once. ...
Packing for Supervised Fine-Tuning (SFT) in autoregressive models involves concatenating data points of varying lengths until reaching the designed maximum length to facilitate GPU processing. However, randomly concatenating data points and feeding them into an autoregressive transformer can lead to cross-contamination of sequences due to the significant difference in their subject matter. The mainstream approaches in SFT ensure that each token in the attention calculation phase only focuses on tokens within its own short sequence, without providing additional learning signals for the preceding context. To address these challenges, we introduce Threshold Filtering Packing (TFP), a method that selects samples with related context while maintaining sufficient diversity within the same pack. Our experiments show that TFP offers a simple-to-implement and scalable approach that significantly enhances SFT performance, with observed improvements of up to 7\% on GSM8K, 4\% on HumanEval, and 15\% on the adult-census-income dataset.
... The Traveling Salesman Problem (TSP) consists in finding the shortest Hamiltonian cycle in a complete graph G = (V , E) of n nodes, weighted on the edges (Applegate et al. 2007). In this paper we consider its symmetric version, i.e., the graph is undirected and the distance between two nodes is the same, irrespective of the direction in which an edge is traversed. ...
... Notice that our paper is focused on making the optimization of the 4-OPT neighborhood practical, but it does not try to assess the performance of such neighborhood in finding good-quality tours. Indeed, the TSP problem is today very effectively solved, even to optimality, by using sophisticated mathematical programming based approaches, such as Concorde (Applegate et al. 2007). No matter how ingenious, heuristics can hardly be competitive with these approaches when the latter are given enough running time. ...
We describe an effective algorithm for exploring the 44-OPT neighborhood for the Traveling Salesman Problem. 44-OPT moves change a tour into another by replacing four of its edges. The best move can be found by a Θ(n4) algorithm by complete enumeration, but a Θ(n3) dynamic programming algorithm exists in the literature. Furthermore a Θ(n2) algorithm also exists for a particular subset of symmetric 44-OPT moves. In this work we describe a new procedure which behaves, on average, slightly worse than a quadratic algorithm over all moves (estimated at O(n2.5)) and like a quadratic algorithm on the symmetric moves. Computational results are reported which show the effectiveness of our strategy compared to other algorithms for finding the best 44-OPT move, and discuss the strength of the 44-OPT neighborhood compared to 2- and 33-OPT.
... The traveling salesman problem (TSP) is one of the most well-known optimization problems in the literature [41]. Several methods, ranging from exact techniques (branch and bound/cut/price) [41] to fast heuristics and approximation algorithms [42,43], have been developed to solve the TSP. ...
... The traveling salesman problem (TSP) is one of the most well-known optimization problems in the literature [41]. Several methods, ranging from exact techniques (branch and bound/cut/price) [41] to fast heuristics and approximation algorithms [42,43], have been developed to solve the TSP. Heuristics generally aim to trade off optimality for computational efficiency. ...
This paper addresses a MinMax variant of the Dubins multiple traveling salesman problem (mTSP). This routing problem arises naturally in mission planning applications involving fixed-wing unmanned vehicles and ground robots. We first formulate the routing problem, referred to as the one-in-a-set Dubins mTSP problem (MD-GmTSP), as a mixed-integer linear program (MILP). We then develop heuristic-based search methods for the MD-GmTSP using tour construction algorithms to generate initial feasible solutions relatively fast and then improve on these solutions using variants of the variable neighborhood search (VNS) metaheuristic. Finally, we also explore a graph neural network to implicitly learn policies for the MD-GmTSP using a learning-based approach; specifically, we employ an S-sample batch reinforcement learning method on a shared graph neural network architecture and distributed policy networks to solve the MD-GMTSP. All the proposed algorithms are implemented on modified TSPLIB instances, and the performance of all the proposed algorithms is corroborated. The results show that learning based approaches work well for smaller sized instances, while the VNS based heuristics find the best solutions for larger instances.
... Un ciclo Hamiltoniano en un grafo corresponde a un camino que pasa por todos los nodos sin repetición y volviendo al inicio. Estos ciclos son de gran interés para resolver problemas del tipo vendedor viajero o TSP (Travelling Salesman Problem) [1]. Este problema tiene grandes aplicaciones en logística bajo el nombre de Vehicle-Routing-Problem [2], en secuenciación de cadenas de DNA [3], diseño de circuitos para microchip (VLSI) [4] y cristalografía [5], entre otros. ...
... Prob(P |Data, N ) ∝ Prob(Data|P, N ) * Prob(P ) (1) El parámetro N también depende de la discretización de P k usada para generar grafos. Esto origina una distribución posterior que puede muestrearse según algún algoritmo mcmc. ...
There are many questions about the statistical properties of random graphs, particularly those related to cyclic
structures. However, theoretical advances have been made in
the sparse connection regime. Recent results on the Kahn-Kalai
conjecture show that there is a limiting connection probability
beyond which it’s very likely to find Hamiltonian cycles. It is
shown that this probability is P ∼ log(n)/n where n is the
number of nodes. We explore experimentally around this limit
by showing its empirical statistical behavior. These results are
useful in configuring various engineering problems based on
sparse graphs.
... The total amount of routes to be explored is determined with the following equation: (n-1)! [14]. It means that in a network of 5 nodes, the total amount of routes to evaluate will be (5-1)! ...
... When the number of nodes increases, the total number of routes increases rapidly. This type of problem can be symmetric, when the distance between 2 nodes are the same, in other words, when the distance from node A to node B is the same from node B to A [6][7] [14]. By making this assumption, the total number of routes is effectively halved ((n-1)!)/2 [6] [7]. ...
There are many branches within Computer Science, one of
them is the Computational Complexity Theory, which studies
problems that require the implementation of an algorithm
used to solve one or many calculations. Even if an algorithm
can solve a problem, it might not be feasible to use due to the
high computational cost that the algorithm requires [3]. This
is how the Computation Complexity Theory begins, it
analyzes the capacity and limitations on Computer Systems
when solving a problem, it validates elements such as
memory, storage and processing speed, to mention some of
them. Scientists like Richard Karp have developed different
studies about problems that could be solved in polynomial
time [4]. From these studies the classification of problems
was originated as N, NP and NP-Hard [3]-[5].
... A constrained combinatorial optimization problem (CCOP) is formally defined as a triple (n, c, S), where n ∈ N is the problem size, c : n → R is the objective function, and S ⊆ n is the feasible set. Given a CCOP, the task is to minimize c over all b ∈ S. Prominent examples of CCOPs are the Job-Shop Scheduling problem [36][37][38], the Knapsack problem [39,40], the Maximum Independent Set problem [41][42][43], the Minimum Spanning Tree problem [44], and the Traveling Salesperson Problem [45], for many of which the corresponding decision problem is NP-complete [46]. 1 Throughout this article, we will only consider CCOPs that lie in NP as this implies that, for any given b ∈ n , calculating c(b) and verifying whether b ∈ S can be done in time polynomial in n. ...
We present a unified quantum-classical framework for addressing NP-complete constrained combinatorial optimization problems, generalizing the recently proposed Quantum Conic Programming (QCP) approach. Accordingly, it inherits many favorable properties of the original proposal such as mitigation of the effects of barren plateaus and avoidance of NP-hard parameter optimization. By collecting the entire classical feasibility structure in a single constraint, we enlarge QCP's scope to arbitrary hard-constrained problems. Yet, we prove that the additional restriction is mild enough to still allow for an efficient parameter optimization via the formulation of a generalized eigenvalue problem (GEP) of adaptable dimension. Our rigorous proof further fills some apparent gaps in prior derivations of GEPs from parameter optimization problems. We further detail a measurement protocol for formulating the classical parameter optimization that does not require us to implement any (time evolution with a) problem-specific objective Hamiltonian or a quantum feasibility oracle. Lastly, we prove that, even under the influence of noise, QCP's parameterized ansatz class always captures the optimum attainable within its generated subcone. All of our results hold true for arbitrarily-constrained combinatorial optimization problems.
... Travelling salesman projection separability (TSPS). The TSPS is a measure we propose that is based on the solution of the travelling salesman problem, or TSP for short, which is one of the most intensively studied problems in computational mathematics [64]. TSP could be described as the problem of a salesman seeking the shortest tour through a N number of cities, passing by each city only once, and returning to the original starting point [65]. ...
Complexity science studies physical phenomena that cannot be explained by the mere analysis of the single units of a system but requires to account for their interactions. A feature of complexity in connected systems is the emergence of mesoscale patterns in a geometric space, such as groupings in bird flocks. These patterns are formed by groups of points that tend to separate from each other, creating mesoscale structures. When multidimensional data or complex networks are embedded in a geometric space, some mesoscale patterns can appear respectively as clusters or communities, and their geometric separability is a feature according to which the performance of an algorithm for network embedding can be evaluated. Here, we introduce a framework for the definition and measure of the geometric separability (linear and nonlinear) of mesoscale patterns by solving the travelling salesman problem (TSP), and we offer experimental evidence on embedding and visualization of multidimensional data or complex networks, which are generated artificially or are derived from real complex systems. For the first time in literature the TSP’s solution is used to define a criterion of nonlinear separability of points in a geometric space, hence redefining the separability problem in terms of the travelling salesman problem is an innovation which impacts both computer science and complexity theory.
... However, no practical quantum advantage for optimization problems [6] could be proven and the application of quantum-assisted methods for industrial use cases is still in its infancy. In classical computing, mathematicians and computer scientists have worked decades to find extremely good approximate algorithms even for NP-hard problems, culminating in complex algorithms like the Concorde Solver for the Traveling Salesperson Problem (TSP) [7]. Clearly, the bar for quantum-enhanced algorithms lies high. ...
When trying to use quantum-enhanced methods for optimization problems, the sheer number of options inhibits its adoption by industrial end users. Expert knowledge is required for the formulation and encoding of the use case, the selection and adaptation of the algorithm, and the identification of a suitable quantum computing backend. Navigating the decision tree spanned by these options is a difficult task and supporting integrated tools are still missing. We propose the QuaST decision tree, a framework that allows to explore, automate and systematically evaluate solution paths. It helps end users to transfer research to their application area, and researchers to gather experience with real-world use cases. Our setup is modular, highly structured and flexible enough to include any kind of preparation, pre-processing and post-processing steps. We develop the guiding principles for the design of the ambitious framework and discuss its implementation. The QuaST decision tree includes multiple complete top-down paths from an application to its fully hybrid quantum solution.
... Various ways of solving the Chinese postman problem can be found, among others, in the literature [32], [41]. In turn, various solutions to the travelling salesman problem are presented, among others, in [4], [5], [7], [37], [41], [55], [56], [57], [59], [64]. This problem has also been solved in terms of dynamic graphs, which take into account dynamic changes in the networks they represent, such as adding/removing edges or vertices ( [2], [71], [76]). ...
... The concept of NP-hardness is viewed as the threshold that distinguishes problems that can be solved efficiently with available computational resources from those that are computationally intractable. Strategies and heuristics to tackle optimization problems belonging into the NP-hard [1] class have been intensively studies for decades [2,3,4]. During this quest, concepts like Integer Linear Programming (ILP) [5,6] and Boolean Satisfiability Problem (SAT) [1,7] have emerged as fundamental mathematical constructions in formulating real-life problems including number partitioning [8], Job-Shop Scheduling [9], optimal trading trajectory [10], prime factorization [11,12], fault detection [13] and Exact Cover [14]. ...
We aim to advance the state-of-the-art in Quadratic Unconstrained Binary Optimization formulation with a focus on cryptography algorithms. As the minimal QUBO encoding of the linear constraints of optimization problems emerges as the solution of integer linear programming (ILP) problems, by solving special boolean logic formulas (like ANF and DNF) for their integer coefficients it is straightforward to handle any normal form, or any substitution for multi-input AND, OR or XOR operations in a QUBO form. To showcase the efficiency of the proposed approach we considered the most widespread cryptography algorithms including AES-128/192/256, MD5, SHA1 and SHA256. For each of these, we achieved QUBO instances reduced by thousands of logical variables compared to previously published results, while keeping the QUBO matrix sparse and the magnitude of the coefficients low. In the particular case of AES-256 cryptography function we obtained more than 8x reduction in variable count compared to previous results. The demonstrated reduction in QUBO sizes notably increases the vulnerability of cryptography algorithms against future quantum annealers, capable of embedding around 30 thousands of logical variables.
... Given the importance of the branching step, most commercial and open-source solvers for MICO problems implement branching techniques like strong branching [2], pseudocosts [3], and reliability branching [4], or combinations thereof. Some of these have been ported to the MINLO world [5], while other branching techniques such as violation transfer [6] have been developed for MINLO specifically. ...
In the branch-and-bound algorithm, branching is the key step to deal with the nonconvexity of the problem. For Mixed Integer Linear Optimization (MILO) problems and, in general, Mixed Integer Nonlinear Optimization (MINLO) problems whose continuous relaxation is convex, branching on integer and binary variables suffices, because fixing all integer variables yields a convex relaxation. General, nonconvex MINLO problems, on the other hand, require spatial branching, i.e., branching on continuous variables. While spatial branching could be seen as necessary for the general MINLO class only, we show that when the branching point is carefully chosen, spatial branching can be more effective than its integer counterpart for a special class of problems arising from Support Vector Machines with Ramp Loss (SVMRL), which can be modeled as mixed integer problems with linear constraints and a convex quadratic objective. We present a strong spatial branching approach for SVMRL coupled with a procedure to strengthen the continuous relaxation, then report on computational tests on known instances from the literature where our approach yields a significant improvement in solve time.
... Travelling Salesman Problem (TSP) merupakan permasalahan penentuan rute terpendek yang diawali dari titik start untuk mengunjungi sekumpulan titik tepat sekali dan diakhiri dengan kembali ke titik start (Bisma & Sanggala, 2023). Pada dunia nyata banyak permasalahan yang dapat didefinisikan sebagai TSP, diantaranya adalah rute perjalanan turis, rute bus sekolah, rute pengiriman barang dan lain-lain (Applegate et al., 2011). Jika sebuah TSP mempunyai banyak titik, maka akan menjadi sebuah NP-Hard Problem dimana untuk memperoleh solusi terbaiknya diperlukan waktu perhitungan yang tidak wajar untuk ditunggu (Wu, 2020). ...
Travelling Salesman Problem (TSP) merupakan permasalahan penentuan rute terpendek yang diawali dari titik start untuk mengunjungi sekumpulan titik tepat sekali dan diakhiri dengan kembali ke titik start. Evolutionary Algorithm (EA) merupakan sebuah metaheuristic yang dapat diaplikasikan pada berbagai permasalahan optimasi, termasuk TSP. Solver merupakan Excel Add-In untuk menyelesaikan permasalahan optimasi. Solver menggunakan tiga algoritma, yaitu LP Simplex, GRG Nonlinear, dan EA. Dengan adanya kemampuan EA untuk menyelesaikan TSP dan Solver yang mampu menjalankan EA, maka dapat disimpulkan bahwa penyelesaian TSP dapat dilakukan dengan memanfaatkan Solver. Untuk membuktikan kemampuan tersebut maka diperlukan sebuah TSP Instance yang akan diselesaikan oleh EA dan Solver. AK-47-TSP Instance merupakan salah satu TSP Instance yang terdapat pada Russian TSP Instances. Dengan menggunakan EA & Solver, panjang rute terpendek dari AK-47-TSP Instance adalah 20.998 Km.
... The asymptotic of this algorithm is O(n 2 *2 n ), where is the number of vertices of the graph. This algorithm finds the optimal solution but is not applicable for n>17, due to too long execution time [12]. ...
The share of small shipments in the total volume of cargo transportation has increased significantly. However, irrational routes are formed, customers’ requirements for timely delivery of goods are ignored, and vehicles of irrational carrying capacity are used on distribution routes, which significantly increases logistics costs and the cost of goods. The limitations and problems of existing solutions were analysed. Theoretic methods are not applicable to the Vehicle Routing Problem with more than 17 customers due to computational complexity. The heuristic methods that were examined focus on creating the shortest routes but do not consider customer priority, route restrictions, or potential service strategies. Therefore, a metaheuristic Genetic algorithm was chosen for route formation tasks. The research is aimed at developing a simulation model, which is based on the use of a genetic algorithm, to create optimal delivery routes for small shipments to customers within a city. The study includes modelling the route formation using a genetic algorithm for various types of cargo bicycles and trucks in a dynamic urban environment. As a result of simulation modelling, distribution routes were formed and operational parameters were optimized on cargo delivery routes in the city for various types of vehicles with different carrying capacities.
... The TSPS is a measure we propose that is based on the solution of the travelling salesman problem, or TSP for short, which is one of the most intensively studied problems in computational mathematics 53 . TSP could be described as the problem of a salesman seeking the shortest tour through an number of cities, passing by each city only once, and returning to the original starting point 54 . ...
Complexity science studies physical phenomena that cannot be explained by the mere analysis of the single units of a system but requires to account for their interactions. A feature of complexity in connected systems is the emergence of mesoscale patterns in a geometric space, such as groupings in bird flocks. These patterns are formed by groups of points that tend to separate from each other, creating mesoscale structures. When multidimensional data or complex networks are embedded in a geometric space, some mesoscale patterns can appear respectively as clusters or communities, and their geometric separability is a feature according to which the performance of an algorithm for network embedding can be evaluated. Here, we introduce a framework for the definition and measure of the geometric separability (linear and nonlinear) of mesoscale patterns by solving the travelling salesman problem (TSP), and we offer experimental evidence on embedding and visualization of multidimensional data or complex networks, which are generated artificially or are derived from real complex systems. For the first time in literature the TSP’s solution is used to define a criterion of nonlinear separability of points in a geometric space, hence redefining the separability problem in terms of the travelling salesman problem is an innovation which impacts both computer science and complexity theory.
... At present, centralized planning can be divided into precise optimization and approximate optimization. Precise optimization mainly includes branch definition [23], dynamic programming [24], and so on. Although precise optimization can obtain the global optimal solution, the computation time of the algorithm is long and increases dramatically with the complexity of the problem. ...
To improve the efficiency of multi-material additive manufacturing and enhance the safety of multi-robot cooperative printing with uncertain execution delay, a dual-robot cooperative path planning method is proposed for the layer-by-layer printing mode. In the proposed algorithm, the description of the printing region is reconstructed by simply using the rectangular envelope region and two-dimensional directed line segment, and the adjacency list of the printing region is established to guide the optimization direction. Therefore, redundant information about the printing region is effectively removed, which is conducive to the optimization of the problem. A multi-round cooperation strategy with multiple synchronous starting points is proposed to accommodate uncertain execution delays by separating the space of the dual-robot printing area, so as to avoid the potential collision risk of dual-robot. To further optimize the printing efficiency, local strategies are used to reduce the makespan. Hence, a better printing order can be obtained, and states of cooperative and non-cooperative printing processes can be unified. In addition, the corresponding NC control strategy is designed for the industrial application of the cooperative strategy. The simulation result shows that this method can effectively reduce the makespan of dual-robot cooperative additive manufacturing, and accommodate the uncertain execution delay of the dual-robot.
... We briefly summarize the state-of-the-art exact and heuristic methods for the symmetric TSP (STSP) and the ATSP. For the STSP, the solver Concorde implements the currently best exact approach using the branch-cut-and-price procedure of Applegate et al. (2007). A comparison of the best exact approaches can be found in Laporte (2010). ...
... Random Algorithm can be a solution for solving this problem. Using Random on SA, makes SA become Random Savings Algorithm (RSA TSP dapat diaplikasikan pada berbagai bidang, termasuk logistik (rute bus sekolah, pengiriman paket), pengurutan genome, masalah pengeboran, data clustering dan lain-lain (Applegate et al., 2011). TSP merupakan sebuah dasar untuk berbagai masalah yang lebih besar. ...
Travelling Salesman Problem (TSP) is the problem for finding the shortest route starting from start node then visiting number of node exactly once and finally go back to start node. Savings Algorithm (SA) is a heuristic for solving TSP. In the Savings Algorithm, the first step that must be taken is to calculate the Savings for each pair of nodes. Then the Savings values that have been obtained are sorted from the largest Savings to the smallest Savings. Route is made by first inserting into the route the pair of nodes who has the highest Savings value. Sometimes there are many pairs of node who have the same Savings value, so it will become problem for SA to choose one of them. Random Algorithm can be a solution for solving this problem. Using Random on SA, makes SA become Random Savings Algorithm (RSA). Performance RSA on solving TSP must be tested on TSP Instance. Two important criterias on the test are solution route and CPU Time. Russian TSP Instances contain ten TSP Instances, on which RSA can be tested. The test result shows that RSA can improve the length of existing route rapidly.
... In the next section, we will see how this bound determines the accuracy of a private measure in T . A natural framework for this step is related to Traveling Salesman Problem (TSP), which is a central problem in optimization and computer science, and whose history goes back to at least 1832 [6]. ...
Differential privacy is a mathematical concept that provides an information-theoretic security guarantee. While differential privacy has emerged as a de facto standard for guaranteeing privacy in data sharing, the known mechanisms to achieve it come with some serious limitations. Utility guarantees are usually provided only for a fixed, a priori specified set of queries. Moreover, there are no utility guarantees for more complex—but very common—machine learning tasks such as clustering or classification. In this paper we overcome some of these limitations. Working with metric privacy, a powerful generalization of differential privacy, we develop a polynomial-time algorithm that creates a private measure from a data set. This private measure allows us to efficiently construct private synthetic data that are accurate for a wide range of statistical analysis tools. Moreover, we prove an asymptotically sharp min-max result for private measures and synthetic data in general compact metric spaces, for any fixed privacy budget ε bounded away from zero. A key ingredient in our construction is a new superregular random walk, whose joint distribution of steps is as regular as that of independent random variables, yet which deviates from the origin logarithmically slowly.
... The Selection function typically selects the most competitive individuals' chromosomes from the mating pool based on the survival strategy, The survival strategy is the fitness score of the individual. Crossover and mutation functions primarily provide better searching diversification in the solution space [5]. The Genetic algorithm is divided into the following steps, details are provided in the following paragraphs. ...
Genetic Algorithm is an optimization technique inspired by nature. The technique has been used by scientists and engineers for real-life search and optimization problems. This work makes use of the genetic algorithm for the solution of the traveling salesman problem. This work focuses on real-time problems; the algorithm is used to find the optimum path for sales travelers inside the Khyber Pakhtunkhwa Province of Pakistan. The solution provides the shortest distance between the cities to be traveled and gives the optimal route. The coding is done in Python-3 and software is developed for the traveling salesmen, where the salesmen can select the cities, they want to travel, and the software will provide the optimal path and the distance.
... The Quadratic Assignment Problem (QAP) is currently among the most extensively investigated NP-hard combinatorial optimization problem, in which even for small instances, solving them has been proven to be very difficult. The largest QAP instance ever solved and proven to be optimal to date has a size of 128 [1], as opposed to the Travelling Salesman Problem (TSP), where the largest TSP ever solved to optimality is of size 85900 cities [2]. ...
... A literatura relacionada ao TSP é extensa devido à sua relevância prática [Applegate, 2007;Laporte, 2007;Derigs, 2009;Matai et al., 2010], bem como por ele ocorrer como subproblema de diversos problemas de rotas. ...
... Traditional TSP algorithms can be classified into three categories, i.e., exact algorithms, approximate algorithms and heuristic algorithms. Concorde (Applegate et al. 2007) is one of the fastest exact solvers. It models TSP as a mixedinteger programming problem, and then adopts a branch and cut algorithm (Padberg and Rinaldi 1991) to search the solution. ...
Traveling Salesman Problem (TSP), as a classic routing optimization problem originally arising in the domain of transportation and logistics, has become a critical task in broader domains, such as manufacturing and biology. Recently, Deep Reinforcement Learning (DRL) has been increasingly employed to solve TSP due to its high inference efficiency. Nevertheless , most of existing end-to-end DRL algorithms only perform well on small TSP instances and can hardly generalize to large scale because of the drastically soaring memory consumption and computation time along with the enlarging problem scale. In this paper, we propose a novel end-to-end DRL approach, referred to as Pointerformer, based on multi-pointer Transformer. Particularly, Pointerformer adopts both reversible residual network in the encoder and multi-pointer network in the decoder to effectively contain memory consumption of the encoder-decoder architecture. To further improve the performance of TSP solutions, Pointerformer employs both a feature augmentation method to explore the symmetries of TSP at both training and inference stages as well as an enhanced context embedding approach to include more comprehensive context information in the query. Extensive experiments on a randomly generated benchmark and a public benchmark have shown that, while achieving comparative results on most small-scale TSP instances as SOTA DRL approaches do, Pointerformer can also well generalize to large-scale TSPs.
... The Traveling Salesman Problem (TSP) is a classical routing problem where many approaches, applications and solution methods can be found in surveys such as (Applegate et al. 2006;Gutin and Punnen 2007), and Matai et al. (2010). Our study addresses Customers in G are divided into g priority groups, each group V α includes nodes with the same priority α ∈ P. Here, P = {1, . . . ...
The Traveling Salesman Problem (TSP) is a well known problem in operations research with various studies and applications. In this paper, we address a variant of the TSP in which the customers are divided into several priority groups and the order of servicing groups can be flexibly changed with a rule called the d-relaxed priority rule. The problem is called the Clustered Traveling Salesman Problem with Relaxed Priority Rule (CTSP-d). We propose two new Mixed Integer Linear Programming (MILP) models for the CTSP-d and a metaheuristic based on Iterated Local Search (ILS) with operators designed for or adapted to the problem. The experimental results obtained on the benchmark instances show that two new models performs better than previous ones, and ILS also proves its performance with 13 new best known solutions found and significant stability compared to existing metaheuristics.
... Traditional TSP algorithms can be classified into three categories, i.e., exact algorithms, approximate algorithms and heuristic algorithms. Concorde (Applegate et al. 2007) is one of the fastest exact solvers. It models TSP as a mixed-integer programming problem, and then adopts a branch and cut algorithm (Padberg and Rinaldi 1991) to search the solution. ...
Traveling Salesman Problem (TSP), as a classic routing optimization problem originally arising in the domain of transportation and logistics, has become a critical task in broader domains, such as manufacturing and biology. Recently, Deep Reinforcement Learning (DRL) has been increasingly employed to solve TSP due to its high inference efficiency. Nevertheless, most of existing end-to-end DRL algorithms only perform well on small TSP instances and can hardly generalize to large scale because of the drastically soaring memory consumption and computation time along with the enlarging problem scale. In this paper, we propose a novel end-to-end DRL approach, referred to as Pointerformer, based on multi-pointer Transformer. Particularly, Pointerformer adopts both reversible residual network in the encoder and multi-pointer network in the decoder to effectively contain memory consumption of the encoder-decoder architecture. To further improve the performance of TSP solutions, Pointerformer employs a feature augmentation method to explore the symmetries of TSP at both training and inference stages as well as an enhanced context embedding approach to include more comprehensive context information in the query. Extensive experiments on a randomly generated benchmark and a public benchmark have shown that, while achieving comparative results on most small-scale TSP instances as state-of-the-art DRL approaches do, Pointerformer can also well generalize to large-scale TSPs.
... Дослідження VRпроблеми дало початок серйозним розробкам у галузі як точних алгоритмів, так і евристик [2]. Свою дуже важливу роль у розвитку теорії маршрутизації відіграла так звана проблема комівояжера [6]. Але найближчим до потреб планування дрібногуртових перевезень віддавна став алгоритм Кларка -Райта (Clarke and Wright) [7,8]. ...
Критично аналізується підхід до маршрутизації дрібногуртових автомобільних перевезень, відомий під іменем «метод Кларка — Райта». Дрібногуртові автомобільні перевезення — дуже особливе явище: у таких перевезеннях може бути задіяна 1/2 наявного рухомого складу, але обсяг перевезень при цьому може становити лише 1/50 загального обсягу. Незважаючи на це, такі перевезення є конче потрібними і завжди вигідними.Та ефективність перевезень істотно залежить від якості їх маршрутизації. Метод Кларка — Райта в класичному своєму варіанті спирається на принцип згортання маятникових маршрутів у кільцевий звізно-розвізний з мінімізацією пробігу автомобіля та раціональнішим використанням його вантажності. У роботі натомість доводиться, що такий підхід до маршрутизації супроводжується неконтрольованим зростанням транспортної роботи без жодного реального зиску. Насправді транспортний засіб перебирає на себе невластиві йому функції складу (він стає коморою на колесах). Це загалом доволі вартісна і погано вмотивована операція. Тож зростання транспортної роботи — це своєрідний штраф за перебирання автомобілем на себе невластивих йому функцій.В аналізованому методі Кларка — Райта насправді значною мірою спотворюється зміст поняття «досконалий план дрібногуртових перевезень». Мінімізуючи пробіг транспортного засобу, ніби вдається зменшити тривалість перевезень. Звісно, час — найцінніший ресурс у житті людини й суспільства. Але збільшення навантаги на транспортний засіб та на дорожню інфраструктуру інтенсифікує їх зношування. Відновлення кондицій транспорту потребує інших ресурсів, за якими обов’язково також стоїть таки час. Але ж цього до уваги від початку ніхто не брав. Звісно, замість пробігу можна оперувати й іншим різновидом «тарифу/ціни/зиску». Але тоді виникає проблема, як об’єктивно визначати тариф/ціну/зиск.Метод Кларка — Райта сприймається доволі природно, якщо мова заходить, скажімо, про «розвезення послуг» — працівник з ремонту побутової техніки, приміром, планує турне (замкнутий тур), аби раціонально за винагороду розвезти свій інтелект, знання, вміння (разом з інструментом у валізі та дрібними запасними частинами. Його філософію мав би сповідувати також і мандрівний продавець (комівояжер), покликаний, приміром, поширити ювелірні прикраси (масою не більше, скажімо, унції).
... As the size of the problem increases, the computation time also increases exponentially. Researchers can get detailed information regarding the types of TSP and solution approaches for these problems from Gutin and Punnen (2006) and Applegate, Bixby, Chvátal, and Cook (2011). ...
Many optimization problems are complex, challenging and take a significant amount of computational effort to solve. These problems have attracted the attention of researchers and they have developed lots of metaheuristic algorithms to use for solving these problems. Most of the developed metaheuristic algorithms are based on some metaphors. For this reason, these algorithms have algorithm-specific parameters to reflect the nature of the inspired metaphor. This violates the algorithm's simplicity and brings extra workload to execute the algorithm. However, the optimization problems can also be solved with simple, useful, metaphor-less and algorithm-specific parameter-less metaheuristic algorithms. So, it is the essential motivation behind this paper. We present a novel metaheuristic algorithm called Discrete Rao Algorithm by updating some components of the generic Rao algorithm to solve the combinatorial optimization problems. To evaluate the performance of the Discrete Rao Algorithm, we carry out experiments on Traveling Salesman Problem which is the well-known combinatorial optimization problem. The experiments are performed on different sized benchmark instances available in the literature. The computational results show that the developed algorithm has obtained high quality solutions in a reasonable computation time and it is competitive with other algorithms in the literature for solving the Traveling Salesman Problem.
... Traditional TSP algorithms can be classified into three categories, i.e., exact algorithms, approximate algorithms and heuristic algorithms. Concorde (Applegate et al. 2007) is one of the fastest exact solvers. It models TSP as a mixedinteger programming problem, and then adopts a branch and cut algorithm (Padberg and Rinaldi 1991) to search the solution. ...
Traveling Salesman Problem (TSP), as a classic routing optimization problem originally arising in the domain of transportation and logistics, has become a critical task in broader domains, such as manufacturing and biology. Recently, Deep Reinforcement Learning (DRL) has been increasingly employed to solve TSP due to its high inference efficiency. Nevertheless, most of existing end-to-end DRL algorithms only perform well on small TSP instances and can hardly generalize to large scale because of the drastically soaring memory consumption and computation time along with the enlarging problem scale. In this paper, we propose a novel end-to-end DRL approach, referred to as Pointerformer, based on multi-pointer Transformer. Particularly, Pointerformer adopts both reversible residual network in the encoder and multi-pointer network in the decoder to effectively contain memory consumption of the encoder-decoder architecture. To further improve the performance of TSP solutions, Pointerformer employs both a feature augmentation method to explore the symmetries of TSP at both training and inference stages as well as an enhanced context embedding approach to include more comprehensive context information in the query. Extensive experiments on a randomly generated benchmark and a public benchmark have shown that, while achieving comparative results on most small-scale TSP instances as SOTA DRL approaches do, Pointerformer can also well generalize to large-scale TSPs.
... One of the most fundamental problems in combinatorial optimization is the traveling salesperson problem (TSP), formalized as early as 1832 (c.f. [App+07,Ch 1]). In an instance of TSP we are given a set of n cities V along with their pairwise symmetric distances, c : V × V → R ≥0 . ...
We show that the max entropy algorithm can be derandomized (with respect to a particular objective function) to give a deterministic approximation algorithm for metric TSP for some . To obtain our result, we apply the method of conditional expectation to an objective function constructed in prior work which was used to certify that the expected cost of the algorithm is at most times the cost of an optimal solution to the subtour elimination LP. The proof in this work involves showing that the expected value of this objective function can be computed in polynomial time (at all stages of the algorithm's execution).
... In Physics, the Ising model is a well established theoretical framework to describe spin orientations of magnetic members of a magnetic material, among some other things, considering their interactions with one another and some other internal or external physical entities [1]. This model is also used interdisciplinarily to build Quadratic Unconstrained Binary Optimization (QUBO) models [2][3][4] to solve several Combinatorial Optimization Problems (COP) [5], such as traveling salesperson problem (TSP) [6], workflow scheduling (WS) [7], the intersection traffic signal control problem (ITSCP) [8], etc. QUBO helps with the construction of an enunciative square matrix for considered problem. And this matrix is then subjected to an optimization model to fine-tune the problem-specific parameters. ...
Calculating the expected spin configuration of the chain consisting of Ferrum atoms interacting with each other through exchange interaction is fundamentally a configuration optimization problem. Quantum Approximate Optimization Algorithm is a suitable candidate to configure such systems on a quantum device. In this work we have considered Ferrum chains of three different lengths and calculated their most-probable spin configurations using Quantum Approximate Optimization Algorithm. We employed a Quantum Feed Forward Neural Network as the optimizer of Quantum Approximate Optimization Algorithm. We have successfully obtained the expected spin configuration for the longest Ferrum Chain.
... A natural framework for this step is related to Traveling Salesman Problem (TSP), which is a central problem in optimization and computer science, and whose history goes back to at least 1832 [6]. ...
Differential privacy is a mathematical concept that provides an information-theoretic security guarantee. While differential privacy has emerged as a de facto standard for guaranteeing privacy in data sharing, the known mechanisms to achieve it come with some serious limitations. Utility guarantees are usually provided only for a fixed, a priori specified set of queries. Moreover, there are no utility guarantees for more complex - but very common - machine learning tasks such as clustering or classification. In this paper we overcome some of these limitations. Working with metric privacy, a powerful generalization of differential privacy, we develop a polynomial-time algorithm that creates a private measure from a data set. This private measure allows us to efficiently construct private synthetic data that are accurate for a wide range of statistical analysis tools. Moreover, we prove an asymptotically sharp min-max result for private measures and synthetic data for general compact metric spaces. A key ingredient in our construction is a new superregular random walk, whose joint distribution of steps is as regular as that of independent random variables, yet which deviates from the origin logarithmicaly slowly.
... The traveling salesman problem (TSP) (Applegate, Bixby, Chvátal, & Cook, 2007) had crucial practical significance in optimizing transport routes, such as logistics distribution and vehicle paths. The TSP can be summarized as follows: consider that a series of cities and the distance between each pair are given, solve the shortest loop that visits each city once, and returns to the starting city. ...
The Traveling Salesman Problem (TSP) is a classic combinatorial optimization problem with various industrial applications. This paper proposes a discrete cuckoo search algorithm based on random walk and cluster analysis to solve the traveling salesman problem (TSP). Although the original cuckoo search algorithm is not suitable to solve the TSP, this algorithm is modified to solve the TSP. Lévy flights and random preference walk in the original cuckoo search algorithm are replaced with novel tools, including the local adjustment operator and discrete random walk. The former is utilized to preserve superior solutions found by the algorithm, while the latter is employed to maintain the diversity of the population. For large-scale TSP problems, the k-means algorithm is first utilized to divide cities into k categories for optimization, while random algorithms are adopted to combine them. A simple 2-opt operator is utilized as the local optimization operator to accelerate the algorithm's convergence rate. Multiple groups of standard TSPLIB datasets are chosen to compare the proposed method with state-of-the-art optimization algorithms that also use the 2-opt/k-opt optimization operator. According to the experimental results, the stability and accuracy of the proposed algorithm are superior to the state-of-the-art optimization algorithms. Regarding the solution accuracy, the algorithm improves 0.39%, 0.29%, 0.84%, 0.4%, 0.07% and 2.22% in the optimal solution over discrete cuckoo search (DCS), discrete bat algorithm (DBA), discrete sine-cosine algorithm (DSCA), random-key cuckoo search (RKCS) and discrete spider monkey optimization (DSMO) on the standard test cases, and 0.58%, 0.19%, 0.85%, 0.43%, 0.36% and 5.83% in the average optimal solution. Regarding stability, the variance of the optimal solutions of RKCS, DSMO, and novel discrete differential evolution (NDEE) for 30 independent runs is 38.5, 52.6, and 59.4 times higher than that of the algorithm in this paper. Finally, the proposed algorithm optimizes air knife migration during glass cutting. Accordingly, the path for air knife migration is reduced by 8556mm.
... The traveling salesperson polytope was introduced by Dantzig, Fulkerson, and Johnson in their classic work on solving the traveling salesperson problem for 49 US cities by integer linear programming [13]. State-of-the-art exact algorithms for the traveling salesperson problem are based on a partial description of the facets of the traveling salesperson polytope and the branch and cut method for integer linear programming [2]. ...
Pyramidal tours with step-backs are Hamiltonian tours of a special kind: the salesperson starts in city 1, then visits some cities in ascending order, reaches city n, and returns to city 1 visiting the remaining cities in descending order. However, in the ascending and descending direction, the order of neighboring cities can be inverted (a step-back). It is known that on pyramidal tours with step-backs the traveling salesperson problem can be solved by dynamic programming in polynomial time. We define the polytope of pyramidal tours with step-backs as the convex hull of the characteristic vectors of all possible pyramidal tours with step-backs in a complete directed graph. The 1-skeleton of is the graph whose vertex set is the vertex set of the polytope, and the edge set is the set of geometric edges or one-dimensional faces of the polytope. We present a linear-time algorithm to verify vertex adjacencies in 1-skeleton of the polytope and estimate the diameter and the clique number of 1-skeleton: the diameter is bounded above by 4 and the clique number grows quadratically in the parameter n.
... The PAR10-score of an algorithm A on an instance I is the arithmetic mean of the runtimes of R independent runs of A on I where failure runs are penalized with 10 · T , T being the cutoff-time; a run is termed a failure if the algorithm did not find the optimum within the cutoff-time. Note that the objective function evaluation requires a single run of the exact TSP solver concorde [37] to calculate an optimal tour. We set R = 3 and T = 180[s] and run each evolving algorithm with 48h wall-time. . ...
Generating instances of different properties is key to algorithm selection methods that differentiate between the performance of different solvers for a given combinatorial optimization problem. A wide range of methods using evolutionary computation techniques has been introduced in recent years. With this paper, we contribute to this area of research by providing a new approach based on quality diversity (QD) that is able to explore the whole feature space. QD algorithms allow to create solutions of high quality within a given feature space by splitting it up into boxes and improving solution quality within each box. We use our QD approach for the generation of TSP instances to visualize and analyze the variety of instances differentiating various TSP solvers and compare it to instances generated by a -EA for TSP instance generation.
We study small traveling salesman problems (TSPs) because current quantum computers can find optional solutions for TSPs with up to 14 cities. Also, we study small TSPs because TSPs have been recommended to be benchmarks to measure quantum optimization on all types of quantum hardware. This means comparisons of quantum data about small TSPs. We extent previous numerical results that were reported in “Small Traveling Salesman Problems” for 6, 8 and 10 cities. The new results in this paper are for 10 – 14 cities in symmetric TSPs. The data for this new range of cities is consistent with the previous data and can be the basis for estimates of results from quantum computers that are upgraded to handle more than 14 cities. The work and analysis suggest two conjectures that we discuss. The paper also contains an annotated survey of recent publications about TSPs.
A Hamiltonian decomposition of a regular graph is a partition of its edge set into Hamiltonian cycles. We consider the second Hamiltonian decomposition problem: for a 4-regular multigraph, find 2 edge-disjoint Hamiltonian cycles different from the given ones. This problem arises in polyhedral combinatorics as a sufficient condition for non-adjacency in the 1-skeleton of the traveling salesperson polytope. We introduce two integer linear programming models for the problem based on the classical Dantzig-Fulkerson-Johnson and Miller-Tucker-Zemlin formulations for the traveling salesperson problem. To enhance the performance on feasible problems, we supplement the algorithm with a variable neighborhood descent heuristic w.r.t. two neighborhood structures and a chain edge fixing procedure. Based on the computational experiments, the Dantzig-Fulkerson-Johnson formulation showed the best results on directed multigraphs, while on undirected multigraphs, the variable neighborhood descent heuristic was especially effective.
This research conducts a comparative analysis of four Ant Colony Optimization (ACO) variants -- Ant System (AS), Rank-Based Ant System (ASRank), Max-Min Ant System (MMAS), and Ant Colony System (ACS) -- for solving the Traveling Salesman Problem (TSP). Our findings demonstrate that algorithm performance is significantly influenced by problem scale and instance type. ACS excels in smaller TSP instances due to its rapid convergence, while PACS proves more adaptable for medium-sized problems. MMAS consistently achieves competitive results across all scales, particularly for larger instances, due to its ability to avoid local optima. ASRank, however, struggles to match the performance of the other algorithms. This research provides insights into the strengths and weaknesses of these ACO variants, guiding the selection of the most suitable algorithm for specific TSP applications.
Boolean satisfiability (SAT) problems are routinely solved by SAT solvers in real-life applications, yet solving time can vary drastically between solvers for the same instance. This has motivated research into machine learning models that can predict, for a given SAT instance, which solver to select among several options. Existing SAT solver selection methods all rely on some hand-picked instance features, which are costly to compute and ignore the structural information in SAT graphs. In this paper we present GraSS, a novel approach for automatic SAT solver selection based on tripartite graph representations of instances and a heterogeneous graph neural network (GNN) model. While GNNs have been previously adopted in other SAT-related tasks, they do not incorporate any domain-specific knowledge and ignore the runtime variation introduced by different clause orders. We enrich the graph representation with domain-specific decisions, such as novel node feature design, positional encodings for clauses in the graph, a GNN architecture tailored to our tripartite graphs and a runtime-sensitive loss function. Through extensive experiments, we demonstrate that this combination of raw representations and domain-specific choices leads to improvements in runtime for a pool of seven state-of-the-art solvers on both an industrial circuit design benchmark, and on instances from the 20-year Anniversary Track of the 2022 SAT Competition.
This tutorial presents a novel search system—the Attractor-Based Search System (ABSS)—that can solve the Traveling Salesman Problem much efficiently with optimality guarantee. From the perspective of dynamical systems, a heuristic local search algorithm for an NP-complete combinatorial problem is a discrete dynamical system. In a local search system, an attractor drives the search trajectories into the vicinity of a globally optimal point in the solution space, and the convergence of local search trajectories make the search system become a global and deterministic system. The attractor contains a small set of the most promising solutions to the problem. The attractor can reduce the problem size exponentially, and thus make the exhaustive search feasible. Therefore, this new search paradigm is called optimizing with attractor. The ABSS consists of two search phases: local search phase and exhaustive search phase. Local search process is used to quickly construct the attractor in the solution space, and exhaustive search process is used to search completely the attractor to identify the optimal solution. Therefore, the exact optimal solution can be found quickly by combining local search and exhaustive search. This tutorial introduces the concept of attractor in a local search system, and describes the process of optimizing with attractor, using the Traveling Salesman Problem as the study platform.
The potholes in the road cause substantial monetary and physical damage to the ongoing traffic. Also, it requires extensive maintenance, especially in areas where the temperature may go down below freezing point. One of the major causes of potholes is the rain or running water that is accumulated in the cracks and later on due to low temperature, changes into ice. Then, ice forms a larger volume for the same amount of water after expanding and causes cracks to expand and at some point become a pothole. They are also the cause of traffic congestion. Therefore, pothole repair needs urgent attention and is one of the routine road maintenance tasks. Kansas City is one of the cities with established social data networks for residents to request road services to mitigate the problem. Although the policymakers have not ignored the issue and rudimentary patching policies are in place, unfortunately, that does not provide efficient road maintenance routes. This paper proposes a Hierarchical Clustering Algorithm for Road Service Routing Enhancement (HiCARE). HiCARE is a practical framework, that makes use of open data in order to optimize the route for maintenance. The optimization considers the time and date of reported potholes, locations, traffic situations, weather conditions, type of patch or other repairs needed, and crew availability, to mention some. This research has characterized spatiotemporal pothole density by analyzing the past 16 years of pothole data from the Open Data KC 311 in Kansas City. HiCARE enhances the NP-hard Traveling Salesperson Problem (TSP) by classifying potholes into layers of clusters. HiCARE finds a cluster’s shortest possible pothole route by identifying each cluster group’s entrance and exit pothole points. Moreover, it modifies the final routes to skip any local minima. The empirical research and analysis indicate that HiCARE significantly reduces the traversing distance and is faster in computation time when compared to typical TSP heuristic algorithms for daily resolution scheduling.
The Covering Salesman Problem (CSP) is a generalization of the Traveling Salesman Problem in which the tour is not required to visit all vertices, as long as all vertices are covered by the tour. The objective of CSP is to find a minimum length Hamiltonian cycle over a subset of vertices that covers an undirected graph. In this paper, valid inequalities from the generalized traveling salesman problem are applied to the CSP in addition to new valid inequalities that explore distinct aspects of the problem. A branch-and-cut framework assembles exact and heuristic separation routines for integer and fractional CSP solutions. Computational experiments show that the proposed framework outperformed methodologies from literature with respect to optimality gaps. Moreover, optimal solutions were proven for several previously unsolved instances.
Autonomous exploration in large-scale and complex environments is a challenging task. As the size of the environment increases, the significant overhead of exploration algorithms could overwhelm the computational capability of mobile platforms, prohibiting timely response to environmental changes. Meanwhile, the quality of exploration paths becomes increasingly important in larger scenes, as poorly selected paths greatly reduce efficiency. In this letter, a systematic framework is proposed to explore large-scale unknown environments. To enable high-frequency planning, a fast preprocessing of environmental information is presented, providing fundamental information to support high-frequency path planning. An path optimization formulation that comprehensively considers key factors about fast exploration is introduced. Further, an heuristic algorithm is devised to solve the NP-hard optimization problem, which empirically finds optimal solution in real time. Simulation results show the run time of our method is significantly shorter than existing ones. Our method completes exploration with the least time and shortest movement distance compared to current state-of-the-art methods.
the performance of the Plant Propagation Algorithm, a crossoverless
metaheuristic, when applied to the Euclidean traveling salesman
problem. However, with large volumes of experimental data, a
clearly characterizable relation emerges, facilitating both an exact
parameterization as well as opening up perspectives on a precisely
predictable outcome. These results directly question its position
between exact, approximation, and PTAS algorithms.
Although the metaheuristic is thereby officially parameter sensitive
when deployed to the Euclidean traveling salesman problem, it
does appear to be largely insensitive to different problem instances.
This is very different from earlier results on continuous optimization,
which were completely opposite. Why is that, and what does
it mean for our practices in algorithmic benchmarking?
In any N-city travelling salesman problem there are (N-1)!/2 possible tours. We use the Metropolis algorithm to generate a sequence of such tours. This sequence may be viewed as the random evolution of a physical system in contact with a heat-bath. As the temperature is lowered, the tours gererated approach the optimal tour. It appears for large N one arrives within a few percent of the optimal solution in better than quadratic time.
A Greek manufacturer of printed circuit boards achieved a significant improvement in the performance of a programmable drilling machine by using a traveling salesman heuristic. Most of the development work was carried out on the company's small personal computer using BASIC. In order to tackle larger drilling tasks and to facilitate the input process, the BASIC program was translated into machine code and a drilling machine-computer interface was designed, leading finally to the manufacture of a drilling optimization device.
We introduce a meta-heuristic to combine simulated annealing with local search methods for CO problems. This new class of Markov chains leads to significantly more powerful optimization methods than either simulated annealing or local search. The main idea is to embed deterministic local search techniques into simulated annealing so that the chain explores only local optima. It makes large, global changes, even at low temperatures, thus overcoming large barriers in configuration space. We have tested this meta-heuristic for the traveling salesman and graph partitioning problems. Tests on instances from public libraries and random ensembles quantify the power of the method. Our algorithm is able to solve large instances to optimally, improving upon local search methods very significantly. For the traveling salesman problem with randomly distributed cities in a square, the procedure improves on 3-opt by 1.6%, and on Lin-Kernighan local search by 1.3%. For the partitioning of sparse random graphs of average degree equal to 5, the improvement over Kernighan-Lin local search is 8.9%. For both CO problems, we obtain new best heuristics.
This paper describes a fast and scalable strategy for constructing a radiation hybrid (RH) map from data on different RH panels. The maps on each panel are then integrated to produce a single RH map for the genome. Recurring problems in using maps from several sources are that the maps use different markers, the maps do not place the overlapping markers in same order, and the objective functions for map quality are incomparable. We use methods from combinatorial optimization to develop a strategy that addresses these issues. We show that by the standard objective functions of obligate chromosome breaks and maximum likelihood, software for the traveling salesman problem produces RH maps with better quality much more quickly than using software specifically tailored for RH mapping. We use known algorithms for the longest common subsequence problem as part of our map integration strategy. We demonstrate our methods by reconstructing and integrating maps for markers typed on the Genebridge 4 (GB4) and the Stanford G3 panels publicly available from the RH database. We compare map quality of our integrated map with published maps for GB4 panel and G3 panel by considering whether markers occur in the same order on a map and in DNA sequence contigs submitted to GenBank. We find that all of the maps are inconsistent with the sequence data for at least 50% of the contigs, but our integrated maps are more consistent. The map integration strategy not only scales to multiple RH maps but also to any maps that have comparable criteria for measuring map quality. Our software improves on current technology for doing RH mapping in areas of computation time and algorithms for considering a large number of markers for mapping. The essential impediments to producing dense high-quality RH maps are data quality and panel size, not computation.
Polyhedral theory / M. Grötschel ; M. W. Padberg. - In: The travelling salesman problem / ed. by E. L. Lawler ... - Chichester u. a. : Wiley, 1985. - S. 251- 305.
Polyhedral computations / M. W. Padberg ; M. Grötschel. - In: The travelling salesman problem / ed. by E. L. Lawler ... - Chichester u. a. : Wiley, 1985. - S. 307-360.
T his paper is concerned with finding the shortest closed path joining n points where the distances between all pairs of points are given.
For a set of points in a metric space we establish that the shortest path will consist of a single loop circuit that will never cross itself. The problem is formally stated in Linear Programming terms. The difficulty in the Linear Programming formulation is to avoid the appearance of sub‐cycles. This has been effected by making the programme dynamic. The computational cost of a complete enumeration of all possible circuits or the cost of solving the Linear Programme being prohibitive for a large number of points, we attempt a practicable computing routine. The method proceeds by writing down the set of distances as a square matrix and selecting from the matrix a set of n links which will form an initial single loop circuit: the circuit is changed by simultaneously removing three links and introducing three new links without destroying the single loop form of the circuit.
Related methods and problems are discussed.
We generalize a method for constructing facets for the convex hull of integer solutions to set packing problems to arbitrary zero-one problems having nonnegative constraint-matrices. A particular class of facets is obtained explicitly and illustrated by a numerical example.
We present three previously unknown classes of facets for the symmetric traveling salesman polytope STSP(8) on 8 nodes which we found using a computer code. These new inequalities now allow for the complete linear description of STSP(8).
We report the solution to optimality of a 532-city symmetric traveling salesman problem involving the optimization over 141,246 zero-one variables. The results of an earlier study by Crowder and Padberg [1] are cross-validated. In this note we briefly outline the methodology, algorithms and software system that we developed to obtain these results.
The paper presents new lower bounding and reduction procedures for the symmetric travelling salesman problem. Lower bounds are obtained through the solution of the linear program corresponding to the 2-matching relaxation and improved through a restricted Lagrangean 2-matching approach. A computational comparison of the new bounds with bounds obtained from the well-known Lagrangean 1-tree relaxation is also presented.
We introduce a new class of valid inequalities for the polytope of the symmetric travelling salesman problem. We also give complete characterizations of the polytope for 6 and 7 cities. For the latter case, the new inequalities are needed. These results are related to work of R. Z. Norman [“On the convex polyhedra of the symmetric travelling salesman problem”, Bull. Amer. Math. Soc. 61, pp. 559 (1955)].
This paper focusses on an often encountered constraint in real-life cutting-stock problems. The constraints require that pieces corresponding to the same order are not spread too much over the production run. This elimination of order spread is called pattern allocation or cutting sequencing. In this paper, a two-stage procedure to solve the two-dimensional pattern-allocation problem is suggested. The first stage consists of solving the cutting-stock problem without the sequencing constraint. In the second stage a sequencing problem is used for the ordering of the cutting patterns in an optimal or near-optimal way. The sequencing problem is formulated as a travelling-salesman model, and the model is solved by Lin's 3-optimal method. Computational experience is reported from a case study in the glass industry.
The author recently proposed a class of cutting planes for integer programs called Fenchel cuts which distinguish themselves from more conventional cuts in that they are generated by directly seeking to solve the separation problem rather than by using explicit knowledge of the polyhedral structure of the integer program. An algorithm for generating Fenchel cuts is presented and described in detail for the separation problem associated with knapsack polyhedra. Computational results are presented for a collection of real-world integer programs to demonstrate the effectiveness of the cutting planes.
We address the problem of wallpapering a room so as to minimize the paper wasted. We show that the problem is equivalent to finding a shortest hamiltonian chain in a highly structured graph. When the chain connects two equivalent nodes traveling salesman problem, the "nearest-neighbor" technique yields an optimal solution.
The traveling-salesman problem consists in selecting from a total number of closed-circuit routes all passing through a number N of given points the one shortest route. To consider all the possible circuits would lead to excessive computations. However, this problem may be solved practically by using a simple, essentially intuitive method.
We define a family of valid inequalities for the Symmetric Travelling Salesman Polytope which are defined on two nested sets of vertices of the graph. These inequalities generalize the comb inequalities of Chvàtal, Grötschel and Padberg, the clique tree inequalities of Grötschel and Pulleyblank, the path inequalities of Cornuéjols, Fonlupt and Naddef and the hyperstar inequalities of Fleischman. This is the largest known family of valid inequalities known so far. Facet inducing inequalities for the Symmetric Travelling Salesman Polytope contained in this class and in no other one are given proving that this is a proper generalization of the previously mentioned families.
We define a new family of valid inequalities for the Symmetric Traveling Salesman Polytope (STSP(n)), that is, the convex hull of the incidence vectors of all Hamiltonian cycles of a complete graph with n nodes. The smallest value of n for which the inequalities are defined is 8. We call these inequalities crown inequalities and we prove that they are facet-inducing for STSP(n), for n greater-than-or-equal-to 8. We describe some extensions of the crown inequalities and we prove that they induce facets of STSP(n), as well. Finally, we discuss some issues on the possible applications of these inequalities in a polyhedral cutting-plane algorithm.
An O(n) space data structure for performing Lin-type p-opt exchanges in O(log n) time for a traveling salesman tour on n nodes is presented.
A graph G is called hypohamiltonian (hypotraceable) if it does not contain a hamiltonian cycle (chain) but if every vertex-deleted subgraph G − v contains a hamiltonian cycle (chain). It is shown that certain classes of these graphs induce facets of the monotone symmetric travelling salesman polytope, i.e. the convex hull of the incidence vectors of all tours and subsets of tours. These results indicate that it is quite unlikely that an explicit complete characterization of this polytope can be obtained.
We describe a new implementation of Edmonds’ blossom algorithm for computing minimum weight perfect matchings. Combining this with specialized pricing techniques, we obtain a solution method for large- scale graphs. We report on the solution of a set of geometric test problems (complete graphs, described as points in the plane), the largest having 101,230 nodes.
Linear programming used to reduce the combinatorial magnitude of travelling-salesman-problems. To illustrate the method, a step-by-step solution of Barachet’s ten-city example is presented.
The traveling-salesman problem has been the target of a substantial number of computational algorithms over the last two decades. Reported computational experience with these algorithms varies widely; authors, however, have generally failed to explain this variation adequately, or to offer predictive theories for their approaches. This paper (a) develops an underlying theory for the problem, (b) predicts pathological performance of some existing techniques, and (c) presents two algorithms, based upon the theory, with predictable polynomial growth in expected computation time and resistence to pathological problems.
The network-flow problem, originally posed by T . Harris of the Rand Corporation, has been discussed from various viewpoints in ( 1; 2; 7; 16 ). The problem arises naturally in the study of transportation networks; it may be stated in the following way. One is given a network of directed arcs and nodes with two distinguished nodes, called source and sink, respectively. All other nodes are called intermediate. Each directed arc in the network has associated with it a nonnegative integer, its flow capacity. Source arcs may be assumed to be directed away from the source, sink arcs into the sink. Subject to the conditions that the flow in an arc is in the direction of the arc and does not exceed its capacity, and that the total flow into any intermediate node is equal to the flow out of it, it is desired to find a maximal flow from source to sink in the network, i.e., a flow which maximizes the sum of the flows in source (or sink) arcs.
Thus, if we let P 1 be the source, P n the sink, we are required to find x ij ( i , j =1, . . . , w ) which maximize
The classical problems reviewed are the traveling salesman problem, minimal spanning tree, minimal matching, greedy matching, minimal triangulation, and others. Each optimization problem is considered for finite sets of points of ℝ d , and the feature of principal interest is the value of the associated objective function. Special attention is given to the asymptotic behavior of this value under probabilistic assumptions, but both probabilistic and worst case analyses are surveyed.
A technique for generating cutting planes for integer programs is introduced that is based on the ability to optimize a linear function on a polyhedron rather than explicit knowledge of the underlying polyhedral structure of the integer program. The theoretical properties of the cuts and their relationship to Lagrangian relaxation are discussed, the cut generation procedure is described, and computational results are presented.
We give a general result showing when a valid inequality or facet of an integer polytope can be obtained from a valid inequality or facet in a subset of the variables.
Many important cutting planes have been discovered for the traveling salesman problem. Among them are the clique tree inequalities and the more general bipartition inequalities. Little is known in the way of exact algorithms for separating these inequalities in polynomial time in the size of the fractional point x * which is being separated. An algorithm is presented here that separates bipartition inequalities of a fixed number of handles and teeth, which includes clique tree inequalities of a fixed number of handles and teeth, in polynomial time.
The linear programming cutting plane approach for solving the travelling salesman problem has recently proven to be highly successful, cf. Crowder and Padberg (Crowder, H. P., M. W. Padberg. 1980. Solving large-scale symmetric travelling salesman problems to optimality. Management Sci. 26 495–509.), Grötschel (Grötschel, M. 1980a. On the symmetric travelling salesman problem: Solution of a 120 city problem. Math. Programming Stud. 12 61–77.), Padberg and Hong (Padberg, M. W., S. Hong. 1980. On the symmetric travelling salesman problem: A computational study. Math. Programming Stud. 12 78–107.). One of the reasons for this success is certainly the fact that instead of ordinary cutting planes (Gomory-cuts etc.) problem-specific cutting planes could be used which define facets of the underlying integer programming polytopes.
In this paper we shall define a new class of inequalities (clique tree inequalities) valid for the travelling salesman polytope which properly contains many of the known classes of inequalities (like subtour elimination constraints, 2-matching constraints, comb inequalities), and we show that all these new inequalities induce facets of the travelling salesman polytope. Since the general structure of these new inequalities is quite simple we hope that it will be possible to use the inequalities efficiently in cutting plane procedures for the travelling salesman problem.
In this history of LP solvers, I describe my efforts from the early days of working with card-programmed calculators and mainframes up to the large systems in use today. The improvements in solvers paralleled the improvements in computers, and what were large problems in the early days are small problems today.