README
¶
Godes
Open Source Library to Build Discrete Event Simulation Models in Go/Golang (http://golang.org/)
Copyright (c) 2013-2015 Alex Goussiatiner agoussia@yahoo.com
Features
Godes is the general-purpose simulation library which includes the simulation engine and building blocks for modeling a wide variety of systems at varying levels of details.
All active objects shall implement the RunnerInterface and have Run() method. For each active object Godes creates a goroutine - lightweight thread.
Godes contains set of built-in functions for generating random numbers for commonly used probability distributions. Each of the distrubutions in Godes has one or more parameter values associated with it: Uniform (Min, Max), Normal (Mean and Standard Deviation), Exponential (Lambda), Triangular(Min, Mode, Max)
Godes implements operations with FIFO and LIFO queues
Godes uses BooleanControl variable as a lock for synchronizing execution of multiple runners
The Object calculates and prints statistical parameters for set of samples collected during the simulation.
Library Docs
Advantages
- Godes is easy to learn for the people familiar with the Go and the elementary simulation concept.
- Godes model executes fast as Go compiles to machine code. Its performace is similar to C++ in performance.
- Godes model is multiplatform as Go compiler targets the Linux, Mac OS X, FreeBSD, Microsoft Windows,etc.
- Godes model can be embedded in various computer systems and over the network.
- Speed of the Godes model compilation is high.
- Variety of the IDE with debuggers are available for Go and Godes as well.
- The Godes sumulation model can use all of the GO's features and libraries.
- Code Security - the Godes includes the source code for the library and Go is an open source project supported by Google.
- Godes is free open source software under MIT license.
Installation
$ go get github.com/agoussia/godes
Examples
Example 0. Restaurant.Godes Basics
During the working day the visitors are entering the restaurant at random intervals and immediately get the table. The inter arrival interval is the random variable with uniform distribution from 0 to 70 minutes. The last visitor gets admitted not later than 8 hours after the opening. The simulation itself is terminated when the last visitors enters the restaurant.
package main
import (
"fmt"
"github.com/godes"
)
// the arrival and service are two random number generators for the uniform distribution
var arrival *godes.UniformDistr = godes.NewUniformDistr(true)
// the Visitor is a Runner
// any type of the Runner should be defined as struct
// with the *godes.Runner as anonimous field
type Visitor struct {
*godes.Runner
number int
}
var visitorsCount int = 0
func (vst *Visitor) Run() { // Any runner should have the Run method
fmt.Printf(" %-6.3f \t Visitor # %v arrives \n", godes.GetSystemTime(), vst.number)
}
func main() {
var shutdown_time float64 = 8 * 60
godes.Run()
for {
//godes.Stime is the current simulation time
if godes.GetSystemTime() < shutdown_time {
//the function acivates the Runner
godes.AddRunner(&Visitor{&godes.Runner{}, visitorsCount})
//this advance the system time
godes.Advance(arrival.Get(0, 70))
visitorsCount++
} else {
break
}
}
// waits for all the runners to finish the Run()
godes.WaitUntilDone()
}
/* OUTPUT:
0.000 Visitor # 0 arrives
37.486 Visitor # 1 arrives
98.737 Visitor # 2 arrives
107.468 Visitor # 3 arrives
149.471 Visitor # 4 arrives
207.523 Visitor # 5 arrives
230.922 Visitor # 6 arrives
261.770 Visitor # 7 arrives
269.668 Visitor # 8 arrives
310.261 Visitor # 9 arrives
338.323 Visitor # 10 arrives
397.720 Visitor # 11 arrives
409.123 Visitor # 12 arrives
436.817 Visitor # 13 arrives
447.731 Visitor # 14 arrives
*/
Example 1. Restaurant. Godes Boolean Controls
The restaurant has only one table to sit on. During the working day the visitors are entering the restaurant at random intervals and wait for the table to be available. The inter arrival interval is the random variable with uniform distribution from 0 to 70 minutes.The time spent in the restaurant is the random variable with uniform distribution from 10 to 60 minutes. The last visitor gets admitted not later than 8 hours after the opening. The simulation itself is terminated when the last visitors has left the restaurant.
package main
import (
"fmt"
"github.com/godes"
)
// the arrival and service are two random number generators for the uniform distribution
var arrival *godes.UniformDistr = godes.NewUniformDistr(true)
var service *godes.UniformDistr = godes.NewUniformDistr(true)
// tableBusy is the boolean control variable than can be accessed and changed by number of Runners
var tableBusy *godes.BooleanControl = godes.NewBooleanControl()
// the Visitor is a Runner
// any type of the Runner should be defined as struct // with the *godes.Runner as anonimous field
type Visitor struct {
*godes.Runner
number int
}
var visitorsCount int = 0
func (vst Visitor) Run() { // Any runner should have the Run method
fmt.Printf("%-6.3f \t Visitor %v arrives \n", godes.GetSystemTime(), vst.number)
tableBusy.Wait(false) // this will wait till the tableBusy control becomes false
tableBusy.Set(true) // sets the tableBusy control to true - the table is busy
fmt.Printf("%-6.3f \t Visitor %v gets the table \n", godes.GetSystemTime(), vst.number)
godes.Advance(service.Get(10, 60)) //the function advance the simulation time by the value in the argument
tableBusy.Set(false) // sets the tableBusy control to false - the table is idle
fmt.Printf("%-6.3f \t Visitor %v leaves \n", godes.GetSystemTime(), vst.number)
}
func main() {
var shutdown_time float64 = 8 * 60
godes.Run()
for {
//godes.GetSystemTime() is the current simulation time
if godes.GetSystemTime() < shutdown_time {
//the function acivates the Runner
godes.AddRunner(Visitor{&godes.Runner{}, visitorsCount})
godes.Advance(arrival.Get(0, 70))
visitorsCount++
} else {
break
}
}
godes.WaitUntilDone() // waits for all the runners to finish the Run()
}
/* OUTPUT
0.000 Visitor 0 arrives
0.000 Visitor 0 gets the table
13.374 Visitor 0 leaves
37.486 Visitor 1 arrives
37.486 Visitor 1 gets the table
60.558 Visitor 1 leaves
98.737 Visitor 2 arrives
98.737 Visitor 2 gets the table
107.468 Visitor 3 arrives
146.824 Visitor 2 leaves
146.824 Visitor 3 gets the table
149.471 Visitor 4 arrives
171.623 Visitor 3 leaves
171.623 Visitor 4 gets the table
187.234 Visitor 4 leaves
207.523 Visitor 5 arrives
207.523 Visitor 5 gets the table
230.922 Visitor 6 arrives
245.859 Visitor 5 leaves
245.859 Visitor 6 gets the table
261.770 Visitor 7 arrives
269.668 Visitor 8 arrives
272.368 Visitor 6 leaves
272.368 Visitor 7 gets the table
290.484 Visitor 7 leaves
290.484 Visitor 8 gets the table
310.261 Visitor 9 arrives
333.570 Visitor 8 leaves
333.570 Visitor 9 gets the table
338.323 Visitor 10 arrives
354.874 Visitor 9 leaves
354.874 Visitor 10 gets the table
393.826 Visitor 10 leaves
397.720 Visitor 11 arrives
397.720 Visitor 11 gets the table
409.123 Visitor 12 arrives
436.817 Visitor 13 arrives
447.731 Visitor 14 arrives
455.705 Visitor 11 leaves
455.705 Visitor 13 gets the table
482.955 Visitor 13 leaves
482.955 Visitor 12 gets the table
496.034 Visitor 12 leaves
496.034 Visitor 14 gets the table
555.822 Visitor 14 leaves
*/
Example 2. Restaurant. Godes Queues
During the four working hours the visitors are entering the restaurant at random intervals and form the arrival queue. The inter arrival interval is the random variable with uniform distribution from 0 to 30 minutes. The restaurant employs two waiters who are servicing one visitor in a time. The service time is the random variable with uniform distribution from 10 to 60 minutes. The simulation itself is terminated when
- Simulation time passes the four hours
- Both waiters have finished servicing
- There are no visitors in the arrival queue.
The model calculates the average (arithmetic mean) of the visitors waiting time
package main
import (
"fmt"
"github.com/godes"
)
// the arrival and service are two random number generators for the uniform distribution
var arrival *godes.UniformDistr = godes.NewUniformDistr(true)
var service *godes.UniformDistr = godes.NewUniformDistr(true)
// true when waiter should act
var waitersSwt *godes.BooleanControl = godes.NewBooleanControl()
// FIFO Queue for the arrived
var visitorArrivalQueue *godes.FIFOQueue = godes.NewFIFOQueue("arrivalQueue")
// the Visitor is a Passive Object
type Visitor struct {
id int
}
// the Waiter is a Runner
type Waiter struct {
*godes.Runner
id int
}
var visitorsCount int = 0
var shutdown_time float64 = 4 * 60
func (waiter *Waiter) Run() {
for {
waitersSwt.Wait(true)
if visitorArrivalQueue.Len() > 0 {
visitor := visitorArrivalQueue.Get()
if visitorArrivalQueue.Len() == 0 {
waitersSwt.Set(false)
}
fmt.Printf("%-6.3f \t Visitor %v is invited by waiter %v \n", godes.GetSystemTime(), visitor.(Visitor).id, waiter.id)
godes.Advance(service.Get(10, 60)) //advance the simulation time by the visitor service time
fmt.Printf("%-6.3f \t Visitor %v leaves \n", godes.GetSystemTime(), visitor.(Visitor).id)
}
if godes.GetSystemTime() > shutdown_time && visitorArrivalQueue.Len() == 0 {
fmt.Printf("%-6.3f \t Waiter %v ends the work \n", godes.GetSystemTime(), waiter.id)
break
}
}
}
func main() {
for i := 0; i < 2; i++ {
godes.AddRunner(&Waiter{&godes.Runner{}, i})
}
godes.Run()
for {
visitorArrivalQueue.Place(Visitor{visitorsCount})
fmt.Printf("%-6.3f \t Visitor %v arrives \n", godes.GetSystemTime(), visitorsCount)
waitersSwt.Set(true)
godes.Advance(arrival.Get(0, 30))
visitorsCount++
if godes.GetSystemTime() > shutdown_time {
break
}
}
waitersSwt.Set(true)
godes.WaitUntilDone() // waits for all the runners to finish the Run()
fmt.Printf("Average Waiting Time %6.3f \n", visitorArrivalQueue.GetAverageTime())
}
/* OUTPUT
0.000 Visitor 0 arrives
0.000 Visitor 0 is invited by waiter 0
13.374 Visitor 0 leaves
16.066 Visitor 1 arrives
16.066 Visitor 1 is invited by waiter 1
39.137 Visitor 1 leaves
42.316 Visitor 2 arrives
42.316 Visitor 2 is invited by waiter 1
46.058 Visitor 3 arrives
46.058 Visitor 3 is invited by waiter 0
64.059 Visitor 4 arrives
70.857 Visitor 3 leaves
70.857 Visitor 4 is invited by waiter 0
86.468 Visitor 4 leaves
88.938 Visitor 5 arrives
88.938 Visitor 5 is invited by waiter 0
90.403 Visitor 2 leaves
98.966 Visitor 6 arrives
98.966 Visitor 6 is invited by waiter 1
112.187 Visitor 7 arrives
115.572 Visitor 8 arrives
125.475 Visitor 6 leaves
125.475 Visitor 7 is invited by waiter 1
127.275 Visitor 5 leaves
127.275 Visitor 8 is invited by waiter 0
132.969 Visitor 9 arrives
143.591 Visitor 7 leaves
143.591 Visitor 9 is invited by waiter 1
144.995 Visitor 10 arrives
164.895 Visitor 9 leaves
164.895 Visitor 10 is invited by waiter 1
170.361 Visitor 8 leaves
170.451 Visitor 11 arrives
170.451 Visitor 11 is invited by waiter 0
175.338 Visitor 12 arrives
187.207 Visitor 13 arrives
191.885 Visitor 14 arrives
203.848 Visitor 10 leaves
203.848 Visitor 12 is invited by waiter 1
213.596 Visitor 15 arrives
228.436 Visitor 11 leaves
228.436 Visitor 13 is invited by waiter 0
231.098 Visitor 12 leaves
231.098 Visitor 14 is invited by waiter 1
231.769 Visitor 16 arrives
241.515 Visitor 13 leaves
241.515 Visitor 15 is invited by waiter 0
287.864 Visitor 15 leaves
287.864 Visitor 16 is invited by waiter 0
290.886 Visitor 14 leaves
290.886 Waiter 1 ends the work
330.903 Visitor 16 leaves
330.903 Waiter 0 ends the work
Average Waiting Time 15.016
*/
Example 3. Restaurant. Multiple Runs
This is the same process as in Example 2. Simulation is repeated 5 times.
package main
import (
"fmt"
"github.com/godes"
)
// the arrival and service are two random number generators for the uniform distribution
var arrival *godes.UniformDistr = godes.NewUniformDistr(true)
var service *godes.UniformDistr = godes.NewUniformDistr(true)
// true when waiter should act
var waitersSwt *godes.BooleanControl = godes.NewBooleanControl()
// FIFO Queue for the arrived
var visitorArrivalQueue *godes.FIFOQueue = godes.NewFIFOQueue("0")
// the Visitor is a Passive Object
type Visitor struct {
id int
}
// the Waiter is a Runner
type Waiter struct {
*godes.Runner
id int
}
var visitorsCount int = 0
var shutdown_time float64 = 4 * 60
func (waiter *Waiter) Run() {
for {
waitersSwt.Wait(true)
if visitorArrivalQueue.Len() > 0 {
visitorArrivalQueue.Get()
if visitorArrivalQueue.Len() == 0 {
waitersSwt.Set(false)
}
godes.Advance(service.Get(10, 60)) //advance the simulation time by the visitor service time
}
if godes.GetSystemTime() > shutdown_time && visitorArrivalQueue.Len() == 0 {
break
}
}
}
func main() {
for runs := 0; runs < 5; runs++ {
for i := 0; i < 2; i++ {
godes.AddRunner(&Waiter{&godes.Runner{}, i})
}
godes.Run()
for {
visitorArrivalQueue.Place(Visitor{visitorsCount})
waitersSwt.Set(true)
godes.Advance(arrival.Get(0, 30))
visitorsCount++
if godes.GetSystemTime() > shutdown_time {
break
}
}
waitersSwt.Set(true)
godes.WaitUntilDone() // waits for all the runners to finish the Run()
fmt.Printf(" Run # %v \t Average Time in Queue=%6.3f \n", runs, visitorArrivalQueue.GetAverageTime())
//clear after each run
waitersSwt.Clear()
visitorArrivalQueue.Clear()
godes.Clear()
}
}
/* OUTPUT
Run # 0 Average Time in Queue=15.016
Run # 1 Average Time in Queue=17.741
Run # 2 Average Time in Queue=49.046
Run # 3 Average Time in Queue=30.696
Run # 4 Average Time in Queue=14.777
*/
Example 4. Machine Shop. Godes Interrupt and Resume Feature.
A workshop has n identical machines. A stream of jobs (enough to keep the machines busy) arrives. Each machine breaks down periodically. Repairs are carried out by one repairman. The repairman continues them when he is done with the machine repair. The workshop works continuously.
package main
import (
"fmt"
"github.com/godes"
)
const PT_MEAN = 10.0 // Avg. processing time in minutes
const PT_SIGMA = 2.0 // Sigma of processing time
const MTTF = 300.0 // Mean time to failure in minutes
const REPAIR_TIME = 30.0 // Time it takes to repair a machine in minutes
const REPAIR_TIME_SIGMA = 1.0 // Sigma of repair time
const NUM_MACHINES = 10
const SHUT_DOWN_TIME = 4 * 7 * 24 * 60
// random generator for the processing time - normal distribution
var processingGen *godes.NormalDistr = godes.NewNormalDistr(true)
// random generator for the time until the next failure for a machine - exponential distribution
var breaksGen *godes.ExpDistr = godes.NewExpDistr(true)
// true when repairman is available for carrying a repair
var repairManAvailableSwt *godes.BooleanControl = godes.NewBooleanControl()
type Machine struct {
*godes.Runner
partsCount int
number int
finished bool
}
func (machine *Machine) Run() {
for {
godes.Advance(processingGen.Get(PT_MEAN, PT_SIGMA))
machine.partsCount++
if godes.GetSystemTime() > SHUT_DOWN_TIME {
machine.finished = true
break
}
}
fmt.Printf(" Machine # %v %v \n", machine.number, machine.partsCount)
}
type MachineRepair struct {
*godes.Runner
machine *Machine
}
func (machineRepair *MachineRepair) Run() {
machine := machineRepair.machine
for {
godes.Advance(breaksGen.Get(1 / MTTF))
if machine.finished {
break
}
interrupted := godes.GetSystemTime()
godes.Interrupt(machine)
repairManAvailableSwt.Wait(true)
if machine.finished {
break
}
repairManAvailableSwt.Set(false)
godes.Advance(processingGen.Get(REPAIR_TIME, REPAIR_TIME_SIGMA))
if machine.finished {
break
}
//release repairman
repairManAvailableSwt.Set(true)
//resume machine and change the scheduled time to compensate delay
godes.Resume(machine, godes.GetSystemTime()-interrupted)
}
}
func main() {
godes.Run()
repairManAvailableSwt.Set(true)
var m *Machine
for i := 0; i < NUM_MACHINES; i++ {
m = &Machine{&godes.Runner{}, 0, i, false}
godes.AddRunner(m)
godes.AddRunner(&MachineRepair{&godes.Runner{}, m})
}
godes.WaitUntilDone()
}
/* OUTPUT
Machine # 1 3382
Machine # 7 3255
Machine # 4 3343
Machine # 5 3336
Machine # 6 3248
Machine # 0 3369
Machine # 2 3378
Machine # 9 3342
Machine # 8 3362
Machine # 3 3213
*/
Example 5. Bank Counter. Godes Wait with Timeout Feature.
This example models a bank counter and customers arriving at random times. Each customer has a certain patience. It waits to get to the counter until she’s at the end of her tether. If she gets to the counter, she uses it for a while before releasing it.
package main
import (
"fmt"
"github.com/godes"
)
const NEW_CUSTOMERS = 5 // Total number of customers
const INTERVAL_CUSTOMERS = 12.00 // Generate new customers roughly every x minites
const SERVICE_TIME = 12.0
const MIN_PATIENCE = 1 // Min. customer patience
const MAX_PATIENCE = 3 // Max. customer patience
// random generator for the arrival interval - expovariate distribution
var arrivalGen *godes.ExpDistr = godes.NewExpDistr(true)
// random generator for the patience time time - uniform distribution
var patienceGen *godes.UniformDistr = godes.NewUniformDistr(true)
// random generator for the service time - expovariate distribution
var serviceGen *godes.ExpDistr = godes.NewExpDistr(true)
// true when Counter
var counterAvailable *godes.BooleanControl = godes.NewBooleanControl()
type Customer struct {
*godes.Runner
name int
}
func (customer *Customer) Run() {
arrivalTime := godes.GetSystemTime()
patience := patienceGen.Get(MIN_PATIENCE, MAX_PATIENCE)
fmt.Printf(" %6.3f Customer %v : Here I am My patience=%6.3f \n", godes.GetSystemTime(), customer.name, patience)
counterAvailable.WaitAndTimeout(true, patience)
if !counterAvailable.GetState() {
fmt.Printf(" %6.3f Customer %v : Reneged after %6.3f \n", godes.GetSystemTime(), customer.name, godes.GetSystemTime()-arrivalTime)
} else {
counterAvailable.Set(false)
fmt.Printf(" %6.3f Customer %v : Waited %6.3f \n", godes.GetSystemTime(), customer.name, godes.GetSystemTime()-arrivalTime)
godes.Advance(serviceGen.Get(1 / SERVICE_TIME))
fmt.Printf(" %6.3f Customer %v : Finished \n", godes.GetSystemTime(), customer.name)
counterAvailable.Set(true)
}
}
func main() {
counterAvailable.Set(true)
godes.Run()
for i := 0; i < NEW_CUSTOMERS; i++ {
godes.AddRunner(&Customer{&godes.Runner{}, i})
godes.Advance(arrivalGen.Get(1 / INTERVAL_CUSTOMERS))
}
godes.WaitUntilDone()
}
/* OUTPUT
0.000 Customer 0 : Here I am My patience= 1.135
0.000 Customer 0 : Waited 0.000
5.536 Customer 0 : Finished
11.141 Customer 1 : Here I am My patience= 1.523
11.141 Customer 1 : Waited 0.000
34.482 Customer 2 : Here I am My patience= 2.523
36.123 Customer 1 : Finished
36.123 Customer 2 : Waited 1.641
38.869 Customer 2 : Finished
39.943 Customer 3 : Here I am My patience= 1.592
39.943 Customer 3 : Waited 0.000
49.113 Customer 4 : Here I am My patience= 1.224
50.337 Customer 4 : Reneged after 1.224
59.948 Customer 3 : Finished
*/
Example 6. Bank. Sngle Run, FIFO Queue, Parallel Resources, StatCollector
A bank employs three tellers and the customers form a queue for all three tellers. The doors of the bank close after eight hours. The simulation is ended when the last customer has been served.
Execute single simulation run, calculate average, standard deviation, confidence interval, lower and upper bounds, minimum and maximum values for the following performance measures: total elapsed time, queue length, queueing time service time.
- FIFO Queue. The customer object is placed in the FIFO arrival queue as soon as the customer is created.
- Parallel Resources. The application constructs Tellers object to model tellers as a set of resources. The object 'provides' tellers to the customer located in the Queue head and "releases" the teller when customer is serviced. Maximum 3 tellers can be provided simultaneously. The interlocking between catching request is performed using godes BooleanControl object.
- Collection and processing of statistics. While finishing a customer run the application creates data arrays for each measure. At the end of simulation, the application creates StatCollector object and performs descriptive statistical analysis. The following statistical parameters are calculated for each measure array: Observ - number of observations, Average - average (mean) value, Std Dev- standard deviation, L-Bound-lower bound of the confidence interval with 95% probability, U-Bound-upper bound of the confidence interval with 95% probability, Minimum value,Maximum value
package main
import (
"fmt"
"github.com/godes"
)
//Input Parameters
const (
ARRIVAL_INTERVAL = 0.5
SERVICE_TIME = 1.3
SHUTDOWN_TIME = 8 * 60.
)
// the arrival and service are two random number generators for the exponential distribution
var arrival *godes.ExpDistr = godes.NewExpDistr(true)
var service *godes.ExpDistr = godes.NewExpDistr(true)
// true when any counter is available
var counterSwt *godes.BooleanControl = godes.NewBooleanControl()
// FIFO Queue for the arrived customers
var customerArrivalQueue *godes.FIFOQueue = godes.NewFIFOQueue("0")
var tellers *Tellers
var measures [][]float64
var titles = []string{
"Elapsed Time",
"Queue Length",
"Queueing Time",
"Service Time",
}
var availableTellers int = 0;
// the Tellers is a Passive Object represebting resource
type Tellers struct {
max int
}
func (tellers *Tellers) Catch(customer *Customer) {
for {
counterSwt.Wait(true)
if customerArrivalQueue.GetHead().(*Customer).id == customer.id {
break
} else {
godes.Yield()
}
}
availableTellers++
if availableTellers == tellers.max {
counterSwt.Set(false)
}
}
func (tellers *Tellers) Release() {
availableTellers--
counterSwt.Set(true)
}
// the Customer is a Runner
type Customer struct {
*godes.Runner
id int
}
func (customer *Customer) Run() {
a0 := godes.GetSystemTime()
tellers.Catch(customer)
a1 := godes.GetSystemTime()
customerArrivalQueue.Get()
qlength := float64(customerArrivalQueue.Len())
godes.Advance(service.Get(1. / SERVICE_TIME))
a2 := godes.GetSystemTime()
tellers.Release()
collectionArray := []float64{a2 - a0, qlength, a1 - a0, a2 - a1}
measures = append(measures, collectionArray)
}
func main() {
measures = [][]float64{}
tellers = &Tellers{3}
godes.Run()
counterSwt.Set(true)
count := 0
for {
customer := &Customer{&godes.Runner{}, count}
customerArrivalQueue.Place(customer)
godes.AddRunner(customer)
godes.Advance(arrival.Get(1. / ARRIVAL_INTERVAL))
if godes.GetSystemTime() > SHUTDOWN_TIME {
break
}
count++
}
godes.WaitUntilDone() // waits for all the runners to finish the Run()
collector := godes.NewStatCollector(titles, measures)
collector.PrintStat()
fmt.Printf("Finished \n")
}
/* OUTPUT
Variable # Average Std Dev L-Bound U-Bound Minimum Maximum
Elapsed Time 944 2.591 1.959 2.466 2.716 0.005 11.189
Queue Length 944 2.411 3.069 2.215 2.607 0.000 13.000
Queueing Time 944 1.293 1.533 1.195 1.391 0.000 6.994
Service Time 944 1.298 1.247 1.219 1.378 0.003 7.824
*/
Example 7. Bank. Multiple Runs, FIFO Queue, Parallel Resources, StatCollector
See example 6.
Execute multiple simulation runs, calculate Average, Standard Deviation, confidence intervall lower and upper bounds,minimu and maximum for the following performance measures: total elapsed time, queue length, queueing time, service time.
package main
import (
"fmt"
"github.com/godes"
)
//Input Parameters
const (
ARRIVAL_INTERVAL = 0.5
SERVICE_TIME = 1.3
SHUTDOWN_TIME = 8 * 60.
INDEPENDENT_RUNS = 100
)
// the arrival and service are two random number generators for the exponential distribution
var arrival *godes.ExpDistr = godes.NewExpDistr(true)
var service *godes.ExpDistr = godes.NewExpDistr(true)
// true when any counter is available
var counterSwt *godes.BooleanControl = godes.NewBooleanControl()
// FIFO Queue for the arrived customers
var customerArrivalQueue *godes.FIFOQueue = godes.NewFIFOQueue("0")
var tellers *Tellers
var statistics [][]float64
var replicationStats [][]float64
var titles = []string{
"Elapsed Time",
"Queue Length",
"Queueing Time",
"Service Time",
}
var availableTellers int = 0
// the Tellers is a Passive Object represebting resource
type Tellers struct {
max int
}
func (tellers *Tellers) Catch(customer *Customer) {
for {
counterSwt.Wait(true)
if customerArrivalQueue.GetHead().(*Customer).GetId() == customer.GetId() {
break
} else {
godes.Yield()
}
}
availableTellers++
if availableTellers == tellers.max {
counterSwt.Set(false)
}
}
func (tellers *Tellers) Release() {
availableTellers--
counterSwt.Set(true)
}
// the Customer is a Runner
type Customer struct {
*godes.Runner
id int
}
func (customer *Customer) Run() {
a0 := godes.GetSystemTime()
tellers.Catch(customer)
a1 := godes.GetSystemTime()
customerArrivalQueue.Get()
qlength := float64(customerArrivalQueue.Len())
godes.Advance(service.Get(1. / SERVICE_TIME))
a2 := godes.GetSystemTime()
tellers.Release()
collectionArray := []float64{a2 - a0, qlength, a1 - a0, a2 - a1}
replicationStats = append(replicationStats, collectionArray)
}
func (customer *Customer) GetId() int{
return customer.id
}
func main() {
statistics = [][]float64{}
tellers = &Tellers{3}
for i := 0; i < INDEPENDENT_RUNS; i++ {
replicationStats = [][]float64{}
godes.Run()
counterSwt.Set(true)
customerArrivalQueue.Clear()
count := 0
for {
customer := &Customer{&godes.Runner{}, count}
customerArrivalQueue.Place(customer)
godes.AddRunner(customer)
godes.Advance(arrival.Get(1. / ARRIVAL_INTERVAL))
if godes.GetSystemTime() > SHUTDOWN_TIME {
break
}
count++
}
godes.WaitUntilDone() // waits for all the runners to finish the Run()
godes.Clear()
replicationCollector := godes.NewStatCollector(titles, replicationStats)
collectionArray := []float64{
replicationCollector.GetAverage(0),
replicationCollector.GetAverage(1),
replicationCollector.GetAverage(2),
replicationCollector.GetAverage(3),
}
statistics = append(statistics, collectionArray)
}
collector := godes.NewStatCollector(titles, statistics)
collector.PrintStat()
fmt.Printf("Finished \n")
}
/* OUTPUT
Variable # Average Std Dev L-Bound U-Bound Minimum Maximum
Elapsed Time 100 3.672 1.217 3.433 3.910 1.980 8.722
Queue Length 100 4.684 2.484 4.197 5.171 1.539 14.615
Queueing Time 100 2.368 1.194 2.134 2.602 0.810 7.350
Service Time 100 1.304 0.044 1.295 1.312 1.170 1.432
Finished
*/
Documentation
¶
Overview ¶
Copyright 2015 Alex Goussiatiner. All rights reserved. Use of this source code is governed by a MIT license that can be found in the LICENSE file.
Godes is the general-purpose simulation library which includes the simulation engine and building blocks for modeling a wide variety of systems at varying levels of details.
Godes Main Features:
1.Active Objects: All active objects in Godes shall implement the RunnerInterface
2.Random Generators: Godes contains set of built-in functions for generating random numbers for commonly used probability distributions. Each of the distrubutions in Godes has one or more parameter values associated with it:Uniform (Min, Max), Normal (Mean and Standard Deviation), Exponential (Lambda), Triangular(Min, Mode, Max)
3.Queues: Godes implements operations with FIFO and LIFO queues
4.BooleanControl : Godes uses BooleanControl variables as a locks for syncronizing execution of multiple runners
5.StatCollector: The object calculates and prints statistical parameters for set of samples collected during the simulation.
See examples for usage.
Index ¶
- func AddRunner(runner RunnerInterface)
- func Advance(interval float64)
- func Clear()
- func GetCurComputerTime() int64
- func GetSystemTime() float64
- func Interrupt(runner RunnerInterface)
- func Mean(nums []float64) (mean float64)
- func MinMax(nums []float64) (minimum float64, maximum float64)
- func NormalConfidenceInterval(nums []float64) (lower float64, upper float64)
- func Resume(runner RunnerInterface, timeChange float64)
- func Run()
- func StandardDeviation(nums []float64) (dev float64)
- func Verbose(v bool)
- func WaitUntilDone()
- func Yield()
- type BooleanControl
- type ExpDistr
- type FIFOQueue
- type LIFOQueue
- type NormalDistr
- type Queue
- type Runner
- type RunnerInterface
- type StatCollector
- func (collector *StatCollector) GetAverage(measureInd int) float64
- func (collector *StatCollector) GetLowBoundCI(measureInd int) float64
- func (collector *StatCollector) GetMaximum(measureInd int) float64
- func (collector *StatCollector) GetMinimum(measureInd int) float64
- func (collector *StatCollector) GetSize(measureInd int) int
- func (collector *StatCollector) GetStandardDeviation(measureInd int) float64
- func (collector *StatCollector) GetStat(measureInd int) (int64, float64, float64, float64, float64, float64, float64)
- func (collector *StatCollector) GetUpperBoundCI(measureInd int) float64
- func (collector *StatCollector) Print(statWriter *bufio.Writer, measuresSwt bool, avgSwt bool, stdDevSwt bool, ...) (err error)
- func (collector *StatCollector) PrintStat()
- type TimeoutRunner
- type TriangularDistr
- type UniformDistr
Constants ¶
This section is empty.
Variables ¶
This section is empty.
Functions ¶
func GetCurComputerTime ¶
func GetCurComputerTime() int64
func NormalConfidenceInterval ¶
NormalConfidenceInterval returns Confidence Interval
func Resume ¶
func Resume(runner RunnerInterface, timeChange float64)
Resume restarts the runner execution
func StandardDeviation ¶
StandardDeviation returns STD
func WaitUntilDone ¶
func WaitUntilDone()
WaitUntilDone stops the main goroutine and waits until all the runners finished executing the Run()
Types ¶
type BooleanControl ¶
type BooleanControl struct {
// contains filtered or unexported fields
}
BooleanControl is a boolean control variable
func NewBooleanControl ¶
func NewBooleanControl() *BooleanControl
NewBooleanControl constructs a BooleanControl
func (*BooleanControl) Clear ¶
func (bc *BooleanControl) Clear()
Clear sets the bc value to default false
func (*BooleanControl) GetState ¶
func (bc *BooleanControl) GetState() bool
getState returns value of bc
func (*BooleanControl) Wait ¶
func (bc *BooleanControl) Wait(b bool)
Wait stops the runner untill the BooleanControll bc is set to true
func (*BooleanControl) WaitAndTimeout ¶
func (bc *BooleanControl) WaitAndTimeout(b bool, timeOut float64)
Wait stops the runner untill the BooleanControll bc is set to true or timeout
type ExpDistr ¶
type ExpDistr struct {
// contains filtered or unexported fields
}
ExpDistr represents the generator for the exponential distribution
func NewExpDistr ¶
NewExpDistr initiats the generator for the exponential distribution If repetition flag is true, the generator will generate the same sequences for every execution
type NormalDistr ¶
type NormalDistr struct {
// contains filtered or unexported fields
}
NormalDistr represents the generator for the normal distribution
func NewNormalDistr ¶
func NewNormalDistr(repetion bool) *NormalDistr
NewNormalDistr initiats the generator for the normal distribution
type Queue ¶
type Queue struct {
// contains filtered or unexported fields
}
Queue represents a FIFO or LIFO queue
func (*Queue) Get ¶
func (q *Queue) Get() interface{}
Get returns an object and removes it from the queue
func (*Queue) GetAverageNumber ¶
GetAverageTime is average elapsed time for an object in the queue
func (*Queue) GetAverageTime ¶
GetAverageTime is average elapsed time for an object in the queue
type Runner ¶
type Runner struct {
// contains filtered or unexported fields
}
func (*Runner) GetMovingTime ¶
func (*Runner) IsShedulled ¶
type RunnerInterface ¶
type RunnerInterface interface {
Run()
// contains filtered or unexported methods
}
type StatCollector ¶
type StatCollector struct {
// contains filtered or unexported fields
}
StatCollector is a wrapper which contains set of samples for statistical analyses
func NewStatCollector ¶
func NewStatCollector(measures []string, samples [][]float64) *StatCollector
NewStatCollector creates a wrapper for samples data
func (*StatCollector) GetAverage ¶
func (collector *StatCollector) GetAverage(measureInd int) float64
GetAverage returns average of a sample
func (*StatCollector) GetLowBoundCI ¶
func (collector *StatCollector) GetLowBoundCI(measureInd int) float64
GetLowBoundCI returns low bound of confidence interval for sample
func (*StatCollector) GetMaximum ¶
func (collector *StatCollector) GetMaximum(measureInd int) float64
GetMaximum returns maximum value for a sample
func (*StatCollector) GetMinimum ¶
func (collector *StatCollector) GetMinimum(measureInd int) float64
GetMinimum returns minimum value for a sample
func (*StatCollector) GetSize ¶
func (collector *StatCollector) GetSize(measureInd int) int
GetSize returns size of a sample
func (*StatCollector) GetStandardDeviation ¶
func (collector *StatCollector) GetStandardDeviation(measureInd int) float64
GetStandardDeviation returns standard deviation of a sample
func (*StatCollector) GetStat ¶
func (collector *StatCollector) GetStat(measureInd int) (int64, float64, float64, float64, float64, float64, float64)
GetStat returns size of the sample, average, standard deviation, low bound and uppe bounds of confidence inteval, minimum and maximum values
func (*StatCollector) GetUpperBoundCI ¶
func (collector *StatCollector) GetUpperBoundCI(measureInd int) float64
GetUpperBoundCI returns upper bound of confidence interval for sample
func (*StatCollector) Print ¶
func (collector *StatCollector) Print(statWriter *bufio.Writer, measuresSwt bool, avgSwt bool, stdDevSwt bool, lBooundSwt bool, uBoundSwt bool, minSwt bool, maxSwt bool) (err error)
Print calculates statistical parameters and output them to *bufio.Writer parameters flags are allowed
func (*StatCollector) PrintStat ¶
func (collector *StatCollector) PrintStat()
PrintStat calculates and prints statistical parameters
type TimeoutRunner ¶
type TimeoutRunner struct {
*Runner
// contains filtered or unexported fields
}
func (*TimeoutRunner) Run ¶
func (timeOut *TimeoutRunner) Run()
type TriangularDistr ¶
type TriangularDistr struct {
// contains filtered or unexported fields
}
TriangularDistr represents the generator for the triangular distribution
func NewTriangularDistr ¶
func NewTriangularDistr(repetion bool) *TriangularDistr
NewTriangularDistr initiats the generator for the triangular distribution If repetition flag is true, the generator will generate the same sequences for every execution
type UniformDistr ¶
type UniformDistr struct {
// contains filtered or unexported fields
}
UniformDistr represents the generator for the uniform distribution
func NewUniformDistr ¶
func NewUniformDistr(repetion bool) *UniformDistr
NewUniformDistr initiats the generator for the uniform distribution
Source Files
Directories