Numerical Computing

You might also like

Download as docx, pdf, or txt
Download as docx, pdf, or txt
You are on page 1of 5

Fundamentals of Accounting (MGT-205)

Title
Furniture business in Gakhar

By Group

Maryam Wasif 19011519-110


Saleha Zahid 19011519-113
Sunaikha 19011519-124
Rubab Tariq 19011519-149

To
Mrs. Sadia Tasaddaque

Bachelor of Computer Science


Faculty of Science
th
March 18 , 2022

Introduction to Numerical Computing


Numerical computing is an approach for solving complex mathematical problems using only
simple arithmetic operations . The approach involves formulation of mathematical models
physical situations that can be solved with arithmetic operations. It requires development,
analysis and use of algorithms.
it is the study of numerical methods that attempt at finding approximate solutions of problems
rather than the exact ones. Numerical analysis finds application in all fields of engineering and
the physical sciences, and in the 21st century also the life and social sciences, medicine,
business and even the arts. Current growth in computing power has enabled the use of more
complex numerical analysis, providing detailed and realistic mathematical models in science
and engineering. Examples of numerical analysis include: ordinary differential equations as
found in celestial mechanics (predicting the motions of planets, stars and galaxies), numerical
linear algebra in data analysis, and stochastic differential equations and Markov chains for
simulating living cells in medicine and biology.

Applications of Numerical Computing:


There’s so many uses for numerical methods, it is impossible to list them all. But essentially, we
can cover first the basic math problems they can be used for, which are often in Engineering
Field
● Computing integrals and derivatives
● Solving differential equations
● Building models based on data, be it through interpolation, Least Square, or other
methods
● Root finding and numerical optimization
● Estimating the solution to a set of linear and nonlinear equations
● Computational geometry
There’s other areas I haven’t listed, but that’s some of the common fundamental uses. With
respect to real world problems, here are some examples where numerical methods are used:
● Development and computation of optimal control algorithms
● Development of high fidelity simulations to model viscous flow around a race car to see
if the wing designs generate sufficient downforce
● Machine learning algorithms, like estimating optimal weights of parametric models
using only subsets of the full dataset (like stochastic gradient descent)
● Photorealistic renderer
● Design optimization based on simulation and multi-objective optimization
formulations
● Game Engines
● Filtering of noisy data based on an approximately expected model of the dynamics
(Kalman Filter, Particle Filter, etc.)

With respect to Computer Science, here are some examples where numerical methods are
used:
● Systems that predict things
● Simulations
● Systems that analyze user behaviors
● Systems that calculate the safest decision to take based on numerical data
● Some algorithms related to computer graphics
● Some algorithms used in videogames to make the AI take smart decisions
● one such Practical application is in machine learning where you have to build a model
from huge amount of data and this model is actually a precise model which can classify
your test cases into some category or can output you some value.
Here are some more uses of numerical Computing in Artificial Intelligence:

1. Learning/Parameter Estimation

The goal of a learning machine is often formalized in terms of an optimization problem, i.e.,
minimization or maximization of a mathematical function. Numerical methods are crucial when
an analytical solution to the optimization does not exist. The most popular numerical
optimization algorithms in ML use derivative information (for example, gradient descent,
accelerated gradient descent methods including Momentum, Nestorov method etc., Newton's
method, L-BFGS). Stochastic approximation algorithms (for example, stochastic gradient
descent) are also popular and have become especially relevant with the proliferation of large
datasets.
The Expectation Maximization (EM) algorithm and variants are popular numerical methods for
estimation in models with hidden/latent variables (for example, Gaussian mixture model,
Hidden Markov Random Field, Structural Equation Model with latent variables).

These algorithms are widely used in regression models (for example, neural
network model with real valued output, large scale linear regression model),
classification models (for example, logistic regression and multinomial
regression), clustering models (for example, k-means clustering, Gaussian mixture
model), structure learning models (for example, Hidden Markov Model, Conditional Random
Field) and reinforcement learning models.

2. Inference
Inference in large probability distributions such as Markov Random Fields (example: Boltzmann
Machine is a special case of a MRF) is often intractable because the normalization constant is
computationally intractable. Hence, to be able to answer questions such as "what is the most
probable value under this distribution?" requires relying upon numerical methods such as
Monte Carlo Markov Chain methods (for example, Metropolis Hastings algorithm, Metropolis
algorithm, Gibbs sampler) which can provide approximations.

3. Numerical Integration
Computing large integrals (or sums) is a common operation in ML. For example, Bayesian
Statistics often requires computing very large multidimensional integrals.
Computing expectations is another common operation in ML and often requires computing
large integrals (examples include EM type algorithms, MRF learning algorithms).
Numerical methods such as Monte Carlo methods are commonly called upon in these cases.

4. Numerical Linear Algebra


Matrix operations such as matrix inversion, computing eigenvalues, and matrix factorization are
essential across all of areas of ML, and require numerical computing. For example, Principal
Component Analyses (PCA) and related methods such as Canonical Correlation Analysis (MCA),
Multiple Correspondence Analysis (MCA) and Latent Semantic Analysis (LSA) all require
numerically computing the Singular Value Decomposition (SVD) of a matrix (or Generalized
SVD).

5. Finite Precision Arithmetic


The digital computer limits our computations to finite precision arithmetic and hence it
becomes important to develop algorithms which protect against common issues arising from
this limitation such as rounding errors leading to overflow or underflow. A good example to
study is the implementation of the softmax function which is commonly used in ML.
We will talk about some major applications of Numerical Analysis in daily-day life that
are both intriguing and easy to understand. Trying to figure out these applications will
give you a sound understanding of the concept in general. 
1.Making Weather Predictions

Advanced computer simulations have made it possible to make weather predictions by


computing numerical data from weather forecasting equipment such as weather
satellites. This is done by making a mathematical model of a particular location and
using computer based Numerical Analysis to obtain precise numerical values that are
used for determining weather changes.
2. Car Safety Enhancement
Car manufacturers also use Numerical Analysis to make numerical models of car crash safety
simulations. These models serve as the platform to unravel optimal results that signify various
aspects of a car crash test. These are then fed to an advanced computer to gain insights into
improving car safety. Mostly, these are partial differential equations. 
3. Machine Learning 
Just like other domains, Machine Learning tends to use Numerical Analysis Questions and
Answers for optimization of numerical functions that it is presented with. A few examples of the
same are Newton’s Method and Nestorov Method. Machine learning needs numerical analysis
for problems that don’t have an analytically optimize-able solution. Artificial Intelligence is yet
another field where machine learning is applied by the use of Numerical Analysis in stochastic
environments.   
4. Spacecraft Dynamics
Space travel also uses numerical methods to make approximations on the spacecraft
trajectories. Space has no gravity in it and therefore the dynamics of motion operate very
differently than on Earth. Numerical Analysis comes handy in making these estimations with the
help of linear as well as non-linear equations that are computed by onboard computers. Solving
simple differential equations is the key in this application of Numerical Analysis. 
5. Price Estimation by Airline Companies 
It might not seem obvious but Airline companies also use Numerical Methods to optimize their
ticket prices, keep a check on their fuel needs, payroll and many other activities. Sophisticated
Numerical Analysis technologies have made it possible to generate accurate approximations of
quantities that can’t be computed otherwise. A field of study stemming from this application;
operations research also deals in finding optimised results for some daily day issues. This is
done making intelligent algorithms that work on Numerical Analysis Questions and
Answers to solve problems. 

You might also like