Simulation, Statistics & Data Analysis


At Tempest, our core expertise is the development and application of quantitative methods to solve problems. We are dedicated to providing our clients with technical support throughout the product design and development process. Quantitative modeling not only reduces both the expense and risk of development but also enhances product performance through improved design. Let us bring our experience to your team and aid you in the rapid and efficient development of ideas into products.

Public Health, Social Sciences, and Policy

Public policy challenges of today require in-depth understanding of societal dynamics. Implementing an expensive, time consuming, complex policy without a priori information on its impacts can have far-reaching negative consequences. Mathematical modeling and supporting data collection and analysis can provide important insights and needed risk mitigation. Moreover, models can often guide policy development and implementation for improved results.

Surveys are an important tool for the collection of data for public health and social systems. Poorly designed surveys can obscure the situation or even provide erroneous data. Particularly in the case of public opinion research, instruments must be constructed with great care and results must be interpreted. Tempest personnel work with clients to design and administer effective surveys that produce insights into social systems.

A highly efficient means of data collection is the web-based survey instrument. However, an open-to-the-public survey instrument often collects only convenience samples and at worst can be manipulated to mislead researchers and policy makers. Working closely with clients to design appropriate samples, our web-survey technology provides a secure platform for data collection. Selected participants are given secure means of accessing the survey and the resulting database protects privacy and data integrity.

The goal of statistical modeling is to provide a medium for making rational decisions in the face of uncertainty. From data mining to economic forecasting to clinical trials analysis, statistical methods form the framework for interpreting and capitalizing on the data at hand. The best models are built from combining an intimate knowledge of the process and the modeling methodology. Mathematical magic and black-box software are tools in the modeling process that must be wielded judiciously. Examples of applications we have worked on include:

  • Assessing insurgent event likelihoods;
  • Determining time-varying yield rates in college admissions;
  • Quantifying and improving effectiveness of credit card retention programs;
  • Computing the likelihood of hurricanes for homeowner insurance rate analysis; and
  • Categorizing drinking behaviors among college students.

Mathematical models provide ways to explore societies as systems. From simple compartmental deterministic models to sophisticated agent-based simulations, models can provide key insights into the impact of policy decisions on populations. Our expertise in mathematical modeling for policy spans the spectrum from back-of-the-envelope to detailed computational simulations. Problem domains in which we have applied mathematical models of societies include:

  • Insurgency dynamics and influence operations;
  • College drinking dynamics and the minimum legal drinking age;
  • The complexities of building evacuation in case of emergency;
  • Optimal treatment of rice fields with mosquitofish; and
  • Societal response to energy availability reduction.

Engineering and Physical Sciences
The modeling of physical systems is both an art and a science – balancing realism with tractability. An integrated approach to engineering product development requires a hand-in-hand treatment of planning and design with experimentation and data analysis. At Tempest, we have the broad-based experience required to undertake the modeling necessary for design, the experimental planning and data analysis that validates the model, and the control and optimization that provides optimal system performance.
Data analysis tasks for engineering product development can range from simple calibration to advanced signal and image processing, with experimental design requiring a great deal of prior modeling and simulation. Tempest personnel have expertise across the board in time and frequency domain tools, nonstationary statistical techniques and experimental design. We have conducted laboratory and field experiments with hardware in the loop for beam control in strategic and tactical lasers, multi-phase fluid flow component quantification, mechanical and optical jitter characterization and control, and image-based, active-passive lidar, laser, and infrared tracking.
Physics based modeling is used to predict performance of systems before they are built, simulate processes to allow determination of optimal control, analyze performance failures, and generally to perform experiments without the necessity of building physical devices. It is also used to derive insight into the behavior of a system. Only experience can effectively drive the modeling process. Examples of applications we have worked on include:

  • Developing a non-invasive medical device to measure the electro-mechanical response of cartilage in order to detect the early onset of arthritis;
  • Determining the transport of subsurface contaminants from highly uncertain and sparse measurements;
  • Controlling vibration in lightly damped flexible structures;
  • Modeling of optical turbulence for image-based tracking; and
  • Fluid flow modeling for oil recovery, ballistic trajectory manipulation, and cardio-vascular system performance analysis.

Control and optimization involve the determination of the ‘best’ solution, a task that arises in almost every branch of science and technology. Frequently, mathematical models are built solely for the purpose of applying optimization techniques. Solution methods are a mixture of heuristics and analysis. An understanding of the theoretical basis of an algorithm and a background of practical experience are important to the solution of real life optimization problems. In many planning situations, optimization approaches must include a game approach when other decision makers (possibly adversarial) are involved.
The use of optimization algorithms spans the realms of aero-engine design, from manufacturing scheduling and warehouse organization, to satellite communications and rocket fuel tank design. Examples of applications we have worked on include:

  • Determining optimal pointing directions for weapon systems;
  • Designing sensing plans for unmanned air and ground vehicles that support troop movements;
  • Optimizing portfolio positions and hedging strategies for investment planning;
  • Allocating tactical resources in missile defense systems; and
  • Defining optimal influence operation levels in the presence of intelligent adversaries.

Control applications often require high speed real time implementation. Moreover, robustness and risk-mitigation concerns bring about the need for adaptive estimation of crucial system parameters. In applications ranging from vibration control in mechanical systems to tracking and adaptive optics in laser beam control, we have developed real-time software (implemented in laboratory and field experiments) that provides high bandwidth control and adaptive estimation capability.