Explore the vast range of titles available.

Data accumulated about customers, products, and website users can not only help interpret the past, it can help predict the future! Probabilistic programming is a programming paradigm in which code models are used to draw probabilistic inferences from data. By applying specialized algorithms, programs assign degrees of probability to conclusions and make it possible to forecast future events like sales trends, computer system failures, experimental outcomes, and other critical concerns. This book explains how to use the PP paradigm to model application domains and express those probabilistic models in code.

This paper considers a class of constrained stochastic composite optimization problems whose objective function is given by the summation of a differentiable (possibly nonconvex) component, together with a certain non-differentiable (but convex) component. In order to solve these problems, we propose a randomized stochastic projected gradient (RSPG) algorithm, in which proper mini-batch of samples are taken at each iteration depending on the total budget of stochastic samples allowed. The RSPG algorithm also employs a general distance function to allow taking advantage of the geometry of the feasible region. Complexity of this algorithm is established in a unified setting, which shows nearly optimal complexity of the algorithm for convex stochastic programming. A post-optimization phase is also proposed to significantly reduce the variance of the solutions returned by the algorithm. In addition, based on the RSPG algorithm, a stochastic gradient free algorithm, which only uses the stochastic zeroth-order information, has been also discussed. Some preliminary numerical results are also provided.

Long‐term multireservoir operations optimization is challenging for existing optimization methods such as stochastic dynamic programming (SDP) and implicit stochastic programming (ISP) suffering from excessive computing time requirements. More difficult is to tackle a risk‐based optimization problem and provide an efficient frontier of the objective function for multireservoir systems. The Fletcher–Ponnambalam (FP) method is an explicit stochastic optimization method suitable for multireservoir operations optimization which faces no curse of dimensionality of SDP and has no need for scenario generations of ISP, thus is extremely fast. Earlier implementations have developed expressions for mean and variance of storages and releases, including deficits and surpluses, to estimate fairly accurate values of the linear and quadratic objective functions when compared with other well‐known methods. This paper introduces analytical derivations of hydropower equations to be used in the recent extension of the FP method and applies it to a long‐term operations optimization problem of a three‐reservoir system in Iran. The objective function is to maximize the expected value of the annual energy, which is a multiplicative nonlinear function of both releases and storage levels. The computational results from simulations for the 60 years of available inflow data for the chosen multireservoir system using the policies derived by the FP, ISP, and SDP methods were compared. The solution qualities were nearly the same, but the FP method has tremendous speedups over the other methods. Secondly, expressions for the variances of monthly energy productions were derived to compute efficient frontier for risk‐return tradeoffs of annual energy to guide decision makers. Plain Language Summary Optimizing long‐term operations for multiple reservoirs is difficult with traditional methods like stochastic dynamic programming (SDP) and implicit stochastic programming (ISP) because they require a lot of computing time. It's even harder to address risk‐based optimization and to create an efficient frontier, which shows the best trade‐offs between different goals. The Fletcher–Ponnambalam (FP) method is a fast and effective solution that doesn't suffer from the complexity issues of SDP and doesn't need scenario generations like ISP. Previous versions of the FP method could accurately estimate values using mean and variance of storage and releases. This paper improves the FP method by introducing new hydropower equations and applies it to optimize a three‐reservoir system in Iran over 60 years of inflow data. The goal is to maximize annual energy, a complex function of water releases and storage levels. The FP method produced solutions comparable in quality to SDP and ISP but was much faster. Additionally, new calculations for monthly energy variance were developed to help make better risk‐return decisions for annual energy production. Key Points Analytic expressions are derived for the moments of energy function to be used in hydropower multireservoir operations optimization, not considered before The derived expressions for the second moment of the generated energy is used to produce efficient frontier easily for annual hydroenergy production function The Fletcher–Ponnambalam (FP) results are better than stochastic dynamic programming and comparable to implicit stochastic programming but much faster than these methods

We introduce a generalized value function of a mixed-integer program, which is simultaneously parameterized by its objective and right-hand side. We describe its fundamental properties, which we exploit through three algorithms to calculate it. We then show how this generalized value function can be used to reformulate two classes of mixed-integer optimization problems: two-stage stochastic mixed-integer programming and multifollower bilevel mixed-integer programming. For both of these problem classes, the generalized value function approach allows the solution of instances that are significantly larger than those solved in the literature in terms of the total number of variables and number of scenarios.

Risk-averse mixed-integer multi-stage stochastic programming problems are challenging, large scale and non-convex optimization problems. In this study, we propose an exact solution algorithm for a type of these problems with an objective of dynamic mean-CVaR risk measure and binary first stage decision variables. The proposed algorithm is based on an evaluate-and-cut procedure and uses lower bounds obtained from a scenario tree decomposition method called as group subproblem approach. We also show that, under the assumption that the first stage integer variables are bounded, our algorithm solves problems with mixed-integer variables in all stages. Computational experiments on risk-averse multi-stage stochastic server location and generation expansion problems reveal that the proposed algorithm is able to solve problem instances with more than one million binary variables within a reasonable time under a modest computational setting.

The home health care (HHC) covers a wide range of health care services carried out in patients’ home in case of illness, injury or aging. Each caregiver should as far as possible adhere to the schedule set by the decision maker. However, unforeseen events would sometimes occur and delay the delivery of care services, which will qualify the service as poor or even risky. Deterministic models ignore this uncertainty, which can arise at any time and will therefore lead to non-compliance with the predefined schedule. Furthermore, patients need several care activities per day, and some of them require to be simultaneous by their nature such as dressing, getting out of bed and bathing. In this work, a stochastic programming model with recourse (SPR model) is proposed to deal with the home health care routing and scheduling problem (HHCRSP) where uncertainties in terms of traveling and caring times that may occur as well as synchronization of services are considered. The objective is to minimize the transportation cost and the expected value of recourse, which is estimated using Monte Carlo simulation. The recourse is defined as a penalty cost for patients’ delayed services and a remuneration for caregivers’ extra working time. The deterministic model is solved by CPLEX, the genetic algorithm (GA) and the general variable neighborhood search (GVNS) based heuristics. The SPR model is solved by Monte Carlo simulation embedded into the GA. Computational results highlight the efficiency of GVNS and GA based heuristics and the complexity of the SPR model in terms of CPU running times.

In this paper, we introduce an approach for constructing uncertainty sets for robust optimization using new deviation measures for random variables termed the forward and backward deviations . These deviation measures capture distributional asymmetry and lead to better approximations of chance constraints. Using a linear decision rule, we also propose a tractable approximation approach for solving a class of multistage chance-constrained stochastic linear optimization problems. An attractive feature of the framework is that we convert the original model into a second-order cone program, which is computationally tractable both in theory and in practice. We demonstrate the framework through an application of a project management problem with uncertain activity completion time.

Multi-stage stochastic linear programs (MSLPs) are notoriously hard to solve in general. Linear decision rules (LDRs) yield an approximation of an MSLP by restricting the decisions at each stage to be an affine function of the observed uncertain parameters. Finding an optimal LDR is a static optimization problem that provides an upper bound on the optimal value of the MSLP, and, under certain assumptions, can be formulated as an explicit linear program. Similarly, as proposed by Kuhn et al. (Math Program 130(1):177–209, 2011) a lower bound for an MSLP can be obtained by restricting decisions in the dual of the MSLP to follow an LDR. We propose a new approximation approach for MSLPs, two-stage LDRs. The idea is to require only the state variables in an MSLP to follow an LDR, which is sufficient to obtain an approximation of an MSLP that is a two-stage stochastic linear program (2SLP). We similarly propose to apply LDR only to a subset of the variables in the dual of the MSLP, which yields a 2SLP approximation of the dual that provides a lower bound on the optimal value of the MSLP. Although solving the corresponding 2SLP approximations exactly is intractable in general, we investigate how approximate solution approaches that have been developed for solving 2SLP can be applied to solve these approximation problems, and derive statistical upper and lower bounds on the optimal value of the MSLP. In addition to potentially yielding better policies and bounds, this approach requires many fewer assumptions than are required to obtain an explicit reformulation when using the standard static LDR approach. A computational study on two example problems demonstrates that using a two-stage LDR can yield significantly better primal policies and modestly better dual policies than using policies based on a static LDR.

In this paper, we consider a stochastic Nash game in which each player minimizes a parameterized expectation-valued convex objective function. In deterministic regimes, proximal best-response (BR) schemes have been shown to be convergent under a suitable spectral property associated with the proximal BR map. However, a direct application of this scheme to stochastic settings requires obtaining exact solutions to stochastic optimization problems at each iteration. Instead, we propose an inexact generalization of this scheme in which an inexact solution to the BR problem is computed in an expected-value sense via a stochastic approximation (SA) scheme. On the basis of this framework, we present three inexact BR schemes: (i) First, we propose a synchronous inexact BR scheme where all players simultaneously update their strategies. (ii) Second, we extend this to a randomized setting where a subset of players is randomly chosen to update their strategies while the other players keep their strategies invariant. (iii) Third, we propose an asynchronous scheme, where each player chooses its update frequency while using outdated rival-specific data in updating its strategy. Under a suitable contractive property on the proximal BR map, we proceed to derive almost sure convergence of the iterates to the Nash equilibrium (NE) for (i) and (ii) and mean convergence for (i)–(iii). In addition, we show that for (i)–(iii), the generated iterates converge to the unique equilibrium in mean at a linear rate with a prescribed constant rather than a sublinear rate. Finally, we establish the overall iteration complexity of the scheme in terms of projected stochastic gradient (SG) steps for computing an ɛ -NE 2 (or ɛ -NE ∞ ) and note that in all settings, the iteration complexity is O ( 1 / ɛ 2 ( 1 + c ) + δ ) , where c = 0 in the context of (i), and c > 0 represents the positive cost of randomization in (ii) and asynchronicity and delay in (iii). Notably, in the synchronous regime, we achieve a near-optimal rate from the standpoint of solving stochastic convex optimization problems by SA schemes. The schemes are further extended to settings where players solve two-stage stochastic Nash games with linear and quadratic recourse. Finally, preliminary numerics developed on a multiportfolio investment problem and a two-stage capacity expansion game support the rate and complexity statements.