# Why do I get the same parameter after the OptimizationProblem

**URL:** <https://discourse.julialang.org/t/why-do-i-get-the-same-parameter-after-the-optimizationproblem/110243>\
**Category:** General Usage\
**Tags:** question\
**Created:** [February 15, 2024, 9:02am UTC](https://discourse.julialang.org/t/why-do-i-get-the-same-parameter-after-the-optimizationproblem/110243 "2024-02-15T09:02:29Z")\
**Posts on this page:** 20\
**Page:** 1

<div class="post-metadata">

**Author:** ![whucsu](https://sea2.discourse-cdn.com/julialang/user_avatar/discourse.julialang.org/whucsu/32/207015_2.png) [@whucsu](https://discourse.julialang.org/u/whucsu)\
**Post date:** [February 15, 2024, 9:02am UTC](https://discourse.julialang.org/t/why-do-i-get-the-same-parameter-after-the-optimizationproblem/110243/1 "2024-02-15T09:02:29Z")

</div>

Dear All,  
I got some questions when using the [DiffEqParamEstim.jl](https://docs.sciml.ai/DiffEqParamEstim/stable/)

my goal is to get the estimated parameter accroding to the Synthetic Data.  
but I got the same parameter as to the firtst “guess” parameter.

the core code is as below:

u0 = [8.0, 8.0, 8.0, 265.0, 0.0, 0.0, 0.0]

p = [0.3, 0.45, 0.06, 0.033, 1.11, 1.13, 11.0, 1.5, 0.03]

tspan = (0, τ \* 30)

prob = ODEProblem(Nik\_ODE, u0, tspan, p) #ODEProblem(Nik\_ODE, u0, tspan, p)

@time sol = solve(prob, Rodas5P(autodiff = false), adaptive = false, dt = 0.00225, reltol = 1e-12, abstol = 1e-12)

using RecursiveArrayTools # for VectorOfArray  
using Optimization, ForwardDiff, OptimizationOptimJL, OptimizationBBO,DiffEqParamEstim  
t = collect(range(6 \* τ, stop = 7 \* τ, length = 200))#collect(start,step,stop) #6 \* τ, 7 \* τ  
randomized = VectorOfArray([(sol(t[i]) + 0.01randn(7)) for i in 1:length(t)])  
data = convert(Array, randomized)  
cost\_function = build\_loss\_objective(prob, Rodas5P(), L2Loss(t, data),  
Optimization.AutoForwardDiff(),  
maxiters = 100000, verbose = false,save\_idxs = [1,4]) #idxs 2 too low ;save\_idxs=[1, 2]

#the original p parameter is :[0.3, 0.45, 0.06, 0.033, 1.11, 1.13, 11.0, 1.5, 0.03]

optprob = Optimization.OptimizationProblem(cost\_function,[0.1, 0.4, 0.05, 0.023, 1.31, 1.23, 11.5, 1.6, 0.03]) # little change around the p

optsol = solve(optprob,BFGS())#Rodas5P(),BFGS()

the result showed the same as the "[0.1, 0.4, 0.05, 0.023, 1.31, 1.23, 11.5, 1.6, 0.03]

ask for help…  
=== the full code is 🙂 new users can not upload file…sorry  
using DifferentialEquations, Plots, LinearAlgebra, Distributions, OffsetArrays, Random, LaTeXStrings

function Valve(R, deltaP)  
q = 0.0  
if (-deltaP) \< 0.0  
q = deltaP/R  
else  
q = 0.0  
end  
return q

end

function ShiElastance(t, Eₘᵢₙ, Eₘₐₓ, τ, τₑₛ, τₑₚ, Eshift)  
τₑₛ = τₑₛ_τ  
τₑₚ = τₑₚ_τ  
#τ = 4/3(τₑₛ+τₑₚ)  
tᵢ = rem(t + (1 - Eshift) \* τ, τ)

```
Eₚ = (tᵢ <= τₑₛ) * (1 - cos(tᵢ / τₑₛ * pi)) / 2 +
     (tᵢ > τₑₛ) * (tᵢ <= τₑₚ) * (1 + cos((tᵢ - τₑₛ) / (τₑₚ - τₑₛ) * pi)) / 2 +
     (tᵢ <= τₑₚ) * 0

E = Eₘᵢₙ + (Eₘₐₓ - Eₘᵢₙ) * Eₚ

return E

```

end

function DShiElastance(t, Eₘᵢₙ, Eₘₐₓ, τ, τₑₛ, τₑₚ, Eshift)

```
τₑₛ = τₑₛ*τ
τₑₚ = τₑₚ*τ
#τ = 4/3(τₑₛ+τₑₚ)
tᵢ = rem(t + (1 - Eshift) * τ, τ)

DEₚ = (tᵢ <= τₑₛ) * pi / τₑₛ * sin(tᵢ / τₑₛ * pi) / 2 +
      (tᵢ > τₑₛ) * (tᵢ <= τₑₚ) * pi / (τₑₚ - τₑₛ) * sin((τₑₛ - tᵢ) / (τₑₚ - τₑₛ) * pi) / 2
(tᵢ <= τₑₚ) * 0
DE = (Eₘₐₓ - Eₘᵢₙ) * DEₚ

return DE

```

end

# Model parameter values

Eshift = 0.0  
Eₘᵢₙ = 0.03  
τₑₛ = 0.3  
τₑₚ = 0.45  
Eₘₐₓ = 1.5  
Rmv = 0.06  
τ = 1.0

function NIK!(du, u, p, t)  
pLV, psa, psv, Vlv, Qav, Qmv, Qs = u  
τₑₛ, τₑₚ, Rmv, Zao, Rs, Csa, Csv, Eₘₐₓ, Eₘᵢₙ = p

```
# 1) Left Ventricle
du[1] = (Qmv - Qav) * ShiElastance(t, Eₘᵢₙ, Eₘₐₓ, τ, τₑₛ, τₑₚ, Eshift) + pLV / ShiElastance(t, Eₘᵢₙ, Eₘₐₓ, τ, τₑₛ, τₑₚ, Eshift) * DShiElastance(t, Eₘᵢₙ, Eₘₐₓ, τ, τₑₛ, τₑₚ, Eshift)
# 2) Systemic arteries 
du[2] = (Qav - Qs ) / Csa     
# 3) Venous
du[3] = (Qs - Qmv) / Csv 
# 4) Left Ventricular Volume
du[4] = Qmv - Qav 
# 5) Aortic Valve flow
du[5] = Valve(Zao, (pLV - psa)) - Qav
# 6) Mitral Valve flow
du[6] = Valve(Rmv, (psv - pLV)) - Qmv 
# 7) Systemic flow
du[7] = (du[2] - du[3]) / Rs
nothing 

```

end

## 

M = [1. 0 0 0 0 0 0  
0 1. 0 0 0 0 0  
0 0 1. 0 0 0 0  
0 0 0 1. 0 0 0  
0 0 0 0 0 0 0  
0 0 0 0 0 0 0  
0 0 0 0 0 0 1. ]

Nik\_ODE = ODEFunction(NIK!,mass\_matrix=M)  
u0 = [8.0, 8.0, 8.0, 265.0, 0.0, 0.0, 0.0]  
p = [0.3, 0.45, 0.06, 0.033, 1.11, 1.13, 11.0, 1.5, 0.03]  
tspan = (0, τ \* 30)  
prob = ODEProblem(Nik\_ODE, u0, tspan, p) #ODEProblem(Nik\_ODE, u0, tspan, p)  
@time sol = solve(prob, Rodas5P(autodiff = false), adaptive = false, dt = 0.00225, reltol = 1e-12, abstol = 1e-12)  
using RecursiveArrayTools # for VectorOfArray  
using Optimization, ForwardDiff, OptimizationOptimJL, OptimizationBBO,DiffEqParamEstim  
t = collect(range(6 \* τ, stop = 7 \* τ, length = 200))#collect(start,step,stop) #6 \* τ, 7 \* τ  
randomized = VectorOfArray([(sol(t[i]) + 0.01randn(7)) for i in 1:length(t)])  
data = convert(Array, randomized)  
cost\_function = build\_loss\_objective(prob, Rodas5P(), L2Loss(t, data),  
Optimization.AutoForwardDiff(),  
maxiters = 100000, verbose = false,save\_idxs = [1,4]) #idxs 2 too low ;save\_idxs=[1, 2]  
#the original p parameter is :[0.3, 0.45, 0.06, 0.033, 1.11, 1.13, 11.0, 1.5, 0.03]  
optprob = Optimization.OptimizationProblem(cost\_function,[0.1, 0.4, 0.05, 0.023, 1.31, 1.23, 11.5, 1.6, 0.03]) # little change around the p  
optsol = solve(optprob,BFGS())#Rodas5P(),BFGS()  
print(optsol)

---

<div class="post-metadata">

**Author:** ![ChrisRackauckas](https://sea2.discourse-cdn.com/julialang/user_avatar/discourse.julialang.org/chrisrackauckas/32/77_2.png) [@ChrisRackauckas](https://discourse.julialang.org/u/ChrisRackauckas)\
**Post date:** [February 15, 2024, 10:13am UTC](https://discourse.julialang.org/t/why-do-i-get-the-same-parameter-after-the-optimizationproblem/110243/2 "2024-02-15T10:13:01Z")

</div>

What does it print out? What warnings or errors?

---

<div class="post-metadata">

**Author:** ![whucsu](https://sea2.discourse-cdn.com/julialang/user_avatar/discourse.julialang.org/whucsu/32/207015_2.png) [@whucsu](https://discourse.julialang.org/u/whucsu)\
**Post date:** [February 15, 2024, 10:41am UTC](https://discourse.julialang.org/t/why-do-i-get-the-same-parameter-after-the-optimizationproblem/110243/3 "2024-02-15T10:41:40Z")

</div>

1.238706 seconds (7.12 M allocations: 182.010 MiB, 4.37% gc time, 85.22% compilation time: 98% of which was recompilation)  
retcode: Success  
u: [0.1, 0.4, 0.05, 0.023, 1.31, 1.23, 11.5, 1.6, 0.03]  
Final objective value: Inf

====  
the output of u is the same as the initial “guess” p.  
of course not as my expection. it should be near to the original p parameter :[0.3, 0.45, 0.06, 0.033, 1.11, 1.13, 11.0, 1.5, 0.03]

---

<div class="post-metadata">

**Author:** ![ChrisRackauckas](https://sea2.discourse-cdn.com/julialang/user_avatar/discourse.julialang.org/chrisrackauckas/32/77_2.png) [@ChrisRackauckas](https://discourse.julialang.org/u/ChrisRackauckas)\
**Post date:** [February 15, 2024, 10:43am UTC](https://discourse.julialang.org/t/why-do-i-get-the-same-parameter-after-the-optimizationproblem/110243/4 "2024-02-15T10:43:07Z")

</div>

What if you use ADAM?

---

<div class="post-metadata">

**Author:** ![whucsu](https://sea2.discourse-cdn.com/julialang/user_avatar/discourse.julialang.org/whucsu/32/207015_2.png) [@whucsu](https://discourse.julialang.org/u/whucsu)\
**Post date:** [February 15, 2024, 10:53am UTC](https://discourse.julialang.org/t/why-do-i-get-the-same-parameter-after-the-optimizationproblem/110243/5 "2024-02-15T10:53:08Z")

</div>

I am not so familiary with it.  
do you mean :  
change optprob = Optimization.OptimizationProblem(cost\_function,[0.,…  
to : optprob = Optimisers.Adam ?  
thanks.

---

<div class="post-metadata">

**Author:** ![ChrisRackauckas](https://sea2.discourse-cdn.com/julialang/user_avatar/discourse.julialang.org/chrisrackauckas/32/77_2.png) [@ChrisRackauckas](https://discourse.julialang.org/u/ChrisRackauckas)\
**Post date:** [February 15, 2024, 11:09am UTC](https://discourse.julialang.org/t/why-do-i-get-the-same-parameter-after-the-optimizationproblem/110243/6 "2024-02-15T11:09:09Z")

</div>

```julia
using OptimizationOptimisers
OptimizationOptimisers.Adam()

```

---

<div class="post-metadata">

**Author:** ![whucsu](https://sea2.discourse-cdn.com/julialang/user_avatar/discourse.julialang.org/whucsu/32/207015_2.png) [@whucsu](https://discourse.julialang.org/u/whucsu)\
**Post date:** [February 15, 2024, 11:31am UTC](https://discourse.julialang.org/t/why-do-i-get-the-same-parameter-after-the-optimizationproblem/110243/7 "2024-02-15T11:31:08Z")

</div>

sorry…  
the core code is 🙂  
cost\_function = build\_loss\_objective(prob, Rodas5P(), L2Loss(t, data),  
Optimization.AutoForwardDiff(),  
maxiters = 100000, verbose = false,save\_idxs = [1,4])  
the optimize below got the not expected result  
optprob = Optimization.OptimizationProblem(cost\_function,[0.1, 0.4, 0.05, 0.023, 1.31, 1.23, 11.5, 1.6, 0.03]) # little change around the p  
optsol = solve(optprob,BFGS())#Rodas5P(),BFGS()

I don’t know how to change the above code to Adam() .

---

<div class="post-metadata">

**Author:** ![whucsu](https://sea2.discourse-cdn.com/julialang/user_avatar/discourse.julialang.org/whucsu/32/207015_2.png) [@whucsu](https://discourse.julialang.org/u/whucsu)\
**Post date:** [February 16, 2024, 10:14am UTC](https://discourse.julialang.org/t/why-do-i-get-the-same-parameter-after-the-optimizationproblem/110243/8 "2024-02-16T10:14:07Z")

</div>

would you please give me the exact suggestion ?  
thanks very much.

---

<div class="post-metadata">

**Author:** ![ChrisRackauckas](https://sea2.discourse-cdn.com/julialang/user_avatar/discourse.julialang.org/chrisrackauckas/32/77_2.png) [@ChrisRackauckas](https://discourse.julialang.org/u/ChrisRackauckas)\
**Post date:** [February 16, 2024, 11:21am UTC](https://discourse.julialang.org/t/why-do-i-get-the-same-parameter-after-the-optimizationproblem/110243/9 "2024-02-16T11:21:47Z")

</div>

> [@whucsu](#):
>
> optsol = solve(optprob,BFGS())

```julia
using OptimizationOptimisers
optsol = solve(optprob,OptimizationOptimisers.Adam())

```

---

<div class="post-metadata">

**Author:** ![whucsu](https://sea2.discourse-cdn.com/julialang/user_avatar/discourse.julialang.org/whucsu/32/207015_2.png) [@whucsu](https://discourse.julialang.org/u/whucsu)\
**Post date:** [February 17, 2024, 10:37am UTC](https://discourse.julialang.org/t/why-do-i-get-the-same-parameter-after-the-optimizationproblem/110243/10 "2024-02-17T10:37:46Z")

</div>

> [@ChrisRackauckas](#):
>
> `using OptimizationOptimisers`

the changed code is :  
#the original p parameter is :[0.3, 0.45, 0.06, 0.033, 1.11, 1.13, 11.0, 1.5, 0.03]

optprob = Optimization.OptimizationProblem(cost\_function,[0.1, 0.4, 0.05, 0.023, 1.31, 1.23, 11.5, 1.6, 0.03]) # little change around the p

optsol = solve(optprob,OptimizationOptimisers.Adam())

I got running error:

 ![image](https://global.discourse-cdn.com/julialang/original/3X/3/e/3e11c129d85042db78b3c3e42943e957ce51bd66.png)

thanks .

---

<div class="post-metadata">

**Author:** ![Vaibhavdixit02](https://sea2.discourse-cdn.com/julialang/user_avatar/discourse.julialang.org/vaibhavdixit02/32/2916_2.png) [@Vaibhavdixit02](https://discourse.julialang.org/u/Vaibhavdixit02)\
**Post date:** [February 17, 2024, 11:50am UTC](https://discourse.julialang.org/t/why-do-i-get-the-same-parameter-after-the-optimizationproblem/110243/11 "2024-02-17T11:50:05Z")

</div>

> [@whucsu](#):
>
> solve(optprob,OptimizationOptimisers.Adam())

Change that to

`optsol = solve(optprob,OptimizationOptimisers.Adam(), maxiters = 100)`

---

<div class="post-metadata">

**Author:** ![whucsu](https://sea2.discourse-cdn.com/julialang/user_avatar/discourse.julialang.org/whucsu/32/207015_2.png) [@whucsu](https://discourse.julialang.org/u/whucsu)\
**Post date:** [February 17, 2024, 12:03pm UTC](https://discourse.julialang.org/t/why-do-i-get-the-same-parameter-after-the-optimizationproblem/110243/12 "2024-02-17T12:03:11Z")

</div>

> [@Vaibhavdixit02](#):
>
> maxiters = 100

thanks.  
the error was solved.  
but after this, I still got the same parameter as the initial “guess” parameter.

the core code is as below:  
using RecursiveArrayTools # for VectorOfArray  
using Optimization, ForwardDiff, OptimizationOptimJL, OptimizationBBO,DiffEqParamEstim  
using OptimizationOptimisers  
t = collect(range(6 \* τ, stop = 7 \* τ, length = 200))#collect(start,step,stop) #6 \* τ, 7 \* τ  
randomized = VectorOfArray([(sol(t[i]) + 0.01randn(7)) for i in 1:length(t)])  
data = convert(Array, randomized)  
cost\_function = build\_loss\_objective(prob, Rodas5P(), L2Loss(t, data),  
Optimization.AutoForwardDiff(),  
maxiters = 100000, verbose = false,save\_idxs = [1,4])  
#the original p parameter is :[0.3, 0.45, 0.06, 0.033, 1.11, 1.13, 11.0, 1.5, 0.03]  
optprob = Optimization.OptimizationProblem(cost\_function,[0.1, 0.3, 0.05, 0.023, 1.31, 1.23, 11.5, 1.6, 0.03]) # little change around the p  
optsol = solve(optprob,OptimizationOptimisers.Adam(),maxiters = 100000)  
and the output is still the same as in the optprob

 ![image](https://global.discourse-cdn.com/julialang/original/3X/6/b/6ba114ebb101225ed337eafe8f23581a6ac6bd3e.png)

---

<div class="post-metadata">

**Author:** ![whucsu](https://sea2.discourse-cdn.com/julialang/user_avatar/discourse.julialang.org/whucsu/32/207015_2.png) [@whucsu](https://discourse.julialang.org/u/whucsu)\
**Post date:** [February 18, 2024, 7:53am UTC](https://discourse.julialang.org/t/why-do-i-get-the-same-parameter-after-the-optimizationproblem/110243/13 "2024-02-18T07:53:51Z")

</div>

in the .jl file, after running, the output is like this:  
Warning: Instability detected. Aborting  
and then gave the same parameter—  
Warning: Instability detected. Aborting  
└ @ SciMLBase C:\Users\whu.julia\packages\SciMLBase\8XHkk\src\integrator\_interface.jl:606  
retcode: Success  
u: 9-element Vector{Float64}:  
0.2  
0.3  
0.05  
0.035  
1.31  
1.23  
11.5  
1.6  
0.03

maybe the not expected parameter comes from the Instability?

Here I gave the whole code. ask for help…  
best wishes.

(new user still can not upload attachment)  
using DifferentialEquations, Plots, LinearAlgebra, Distributions, OffsetArrays, Random, LaTeXStrings

function Valve(R, deltaP)  
q = 0.0  
if (-deltaP) \< 0.0  
q = deltaP/R  
else  
q = 0.0  
end  
return q

end

function ShiElastance(t, Eₘᵢₙ, Eₘₐₓ, τ, τₑₛ, τₑₚ, Eshift)  
τₑₛ = τₑₛ_τ  
τₑₚ = τₑₚ_τ  
#τ = 4/3(τₑₛ+τₑₚ)  
tᵢ = rem(t + (1 - Eshift) \* τ, τ)

```
Eₚ = (tᵢ <= τₑₛ) * (1 - cos(tᵢ / τₑₛ * pi)) / 2 +
     (tᵢ > τₑₛ) * (tᵢ <= τₑₚ) * (1 + cos((tᵢ - τₑₛ) / (τₑₚ - τₑₛ) * pi)) / 2 +
     (tᵢ <= τₑₚ) * 0

E = Eₘᵢₙ + (Eₘₐₓ - Eₘᵢₙ) * Eₚ

return E

```

end

function DShiElastance(t, Eₘᵢₙ, Eₘₐₓ, τ, τₑₛ, τₑₚ, Eshift)  
τₑₛ = τₑₛ_τ  
τₑₚ = τₑₚ_τ  
#τ = 4/3(τₑₛ+τₑₚ)  
tᵢ = rem(t + (1 - Eshift) \* τ, τ)

```
DEₚ = (tᵢ <= τₑₛ) * pi / τₑₛ * sin(tᵢ / τₑₛ * pi) / 2 +
      (tᵢ > τₑₛ) * (tᵢ <= τₑₚ) * pi / (τₑₚ - τₑₛ) * sin((τₑₛ - tᵢ) / (τₑₚ - τₑₛ) * pi) / 2
(tᵢ <= τₑₚ) * 0
DE = (Eₘₐₓ - Eₘᵢₙ) * DEₚ

return DE

```

end

# Model parameter values

Eshift = 0.0  
Eₘᵢₙ = 0.03  
τₑₛ = 0.3  
τₑₚ = 0.45  
Eₘₐₓ = 1.5  
Rmv = 0.06  
τ = 1.0

function NIK!(du, u, p, t)  
pLV, psa, psv, Vlv, Qav, Qmv, Qs = u  
τₑₛ, τₑₚ, Rmv, Zao, Rs, Csa, Csv, Eₘₐₓ, Eₘᵢₙ = p

```
# 1) Left Ventricle
du[1] = (Qmv - Qav) * ShiElastance(t, Eₘᵢₙ, Eₘₐₓ, τ, τₑₛ, τₑₚ, Eshift) + pLV / ShiElastance(t, Eₘᵢₙ, Eₘₐₓ, τ, τₑₛ, τₑₚ, Eshift) * DShiElastance(t, Eₘᵢₙ, Eₘₐₓ, τ, τₑₛ, τₑₚ, Eshift)
# 2) Systemic arteries 
du[2] = (Qav - Qs ) / Csa     
# 3) Venous
du[3] = (Qs - Qmv) / Csv 
# 4) Left Ventricular Volume
du[4] = Qmv - Qav 
# 5) Aortic Valve flow
du[5] = Valve(Zao, (pLV - psa)) - Qav
# 6) Mitral Valve flow
du[6] = Valve(Rmv, (psv - pLV)) - Qmv 
# 7) Systemic flow
du[7] = (du[2] - du[3]) / Rs
nothing 

```

end

## 

M = [1. 0 0 0 0 0 0  
0 1. 0 0 0 0 0  
0 0 1. 0 0 0 0  
0 0 0 1. 0 0 0  
0 0 0 0 0 0 0  
0 0 0 0 0 0 0  
0 0 0 0 0 0 1. ]

Nik\_ODE = ODEFunction(NIK!,mass\_matrix=M)

u0 = [8.0, 8.0, 8.0, 265.0, 0.0, 0.0, 0.0]  
p = [0.3, 0.45, 0.06, 0.033, 1.11, 1.13, 11.0, 1.5, 0.03]  
tspan = (0, τ \* 30)  
prob = ODEProblem(Nik\_ODE, u0, tspan, p) #ODEProblem(Nik\_ODE, u0, tspan, p)  
@time sol = solve(prob, Rodas5P(autodiff = false), adaptive = false, dt = 0.00225, reltol = 1e-12, abstol = 1e-12)

using RecursiveArrayTools # for VectorOfArray  
using Optimization, ForwardDiff, OptimizationOptimJL, OptimizationBBO,DiffEqParamEstim  
using OptimizationOptimisers  
t = collect(range(6 \* τ, stop = 7 \* τ, length = 200))#collect(start,step,stop) #6 \* τ, 7 \* τ  
randomized = VectorOfArray([(sol(t[i]) + 0.01randn(7)) for i in 1:length(t)])  
data = convert(Array, randomized)  
cost\_function = build\_loss\_objective(prob, Rodas5P(), L2Loss(t, data),  
Optimization.AutoForwardDiff(),tspan=(6 \* τ, 7 \* τ),save\_idxs = [1,4]) #idxs 2 too low ;save\_idxs=[1, 2]  
#the original p parameter is :[0.3, 0.45, 0.06, 0.033, 1.11, 1.13, 11.0, 1.5, 0.03]  
optprob = Optimization.OptimizationProblem(cost\_function,[0.2, 0.3, 0.05, 0.035, 1.31, 1.23, 11.5, 1.6, 0.03]) # little change around the p  
optsol = solve(optprob,Optim.BFGS())  
#optsol = solve(optprob,OptimizationOptimisers.Adam(),maxiters = 100000)#OptimizationOptimisers.Adam(),BFGS()

---

<div class="post-metadata">

**Author:** ![ChrisRackauckas](https://sea2.discourse-cdn.com/julialang/user_avatar/discourse.julialang.org/chrisrackauckas/32/77_2.png) [@ChrisRackauckas](https://discourse.julialang.org/u/ChrisRackauckas)\
**Post date:** [February 18, 2024, 1:49pm UTC](https://discourse.julialang.org/t/why-do-i-get-the-same-parameter-after-the-optimizationproblem/110243/14 "2024-02-18T13:49:39Z")

</div>

> [@whucsu](#):
>
> maybe the not expected parameter comes from the Instability?

Yes, if your differential equation is unstable at the chosen starting parameters and requires aborting then it will not be able to update the parameters. That’s the reason for the warning.

---

<div class="post-metadata">

**Author:** ![ChrisRackauckas](https://sea2.discourse-cdn.com/julialang/user_avatar/discourse.julialang.org/chrisrackauckas/32/77_2.png) [@ChrisRackauckas](https://discourse.julialang.org/u/ChrisRackauckas)\
**Post date:** [February 22, 2024, 4:26am UTC](https://discourse.julialang.org/t/why-do-i-get-the-same-parameter-after-the-optimizationproblem/110243/16 "2024-02-22T04:26:48Z")

</div>

What is the solve like at the initial parameters?

---

<div class="post-metadata">

**Author:** ![whucsu](https://sea2.discourse-cdn.com/julialang/user_avatar/discourse.julialang.org/whucsu/32/207015_2.png) [@whucsu](https://discourse.julialang.org/u/whucsu)\
**Post date:** [February 23, 2024, 5:45am UTC](https://discourse.julialang.org/t/why-do-i-get-the-same-parameter-after-the-optimizationproblem/110243/17 "2024-02-23T05:45:13Z")

</div>

it seemed that the initial parameters are good.  
the output is like this:

retcode: Success  
Interpolation: specialized 4rd order “free” stiffness-aware interpolation  
t: 13335-element Vector{Float64}:

---

<div class="post-metadata">

**Author:** ![whucsu](https://sea2.discourse-cdn.com/julialang/user_avatar/discourse.julialang.org/whucsu/32/207015_2.png) [@whucsu](https://discourse.julialang.org/u/whucsu)\
**Post date:** [February 24, 2024, 6:27am UTC](https://discourse.julialang.org/t/why-do-i-get-the-same-parameter-after-the-optimizationproblem/110243/18 "2024-02-24T06:27:02Z")

</div>

I checked the initial parameters of p=[0.25, 0.4, 0.05, 0.023, 1.31, 1.23, 11.5, 1.6, 0.03] again, the return code is retcode: Success.  
is there any method that could print out the result when the system is instability while updating the parameters ?  
but it just seemed that the parameters have never been updated .

thanks.

---

<div class="post-metadata">

**Author:** ![whucsu](https://sea2.discourse-cdn.com/julialang/user_avatar/discourse.julialang.org/whucsu/32/207015_2.png) [@whucsu](https://discourse.julialang.org/u/whucsu)\
**Post date:** [February 28, 2024, 3:54am UTC](https://discourse.julialang.org/t/why-do-i-get-the-same-parameter-after-the-optimizationproblem/110243/19 "2024-02-28T03:54:26Z")

</div>

is there any other method that could be used to check the stability ?  
debug ? check the level of output ?  
thanks. best wishes.  
whu

---

<div class="post-metadata">

**Author:** ![ChrisRackauckas](https://sea2.discourse-cdn.com/julialang/user_avatar/discourse.julialang.org/chrisrackauckas/32/77_2.png) [@ChrisRackauckas](https://discourse.julialang.org/u/ChrisRackauckas)\
**Post date:** [March 5, 2024, 9:02am UTC](https://discourse.julialang.org/t/why-do-i-get-the-same-parameter-after-the-optimizationproblem/110243/20 "2024-03-05T09:02:48Z")

</div>

I think @Vaibhavdixit02 was looking at this?

---

<div class="post-metadata">

**Author:** ![Vaibhavdixit02](https://sea2.discourse-cdn.com/julialang/user_avatar/discourse.julialang.org/vaibhavdixit02/32/2916_2.png) [@Vaibhavdixit02](https://discourse.julialang.org/u/Vaibhavdixit02)\
**Post date:** [March 5, 2024, 4:32pm UTC](https://discourse.julialang.org/t/why-do-i-get-the-same-parameter-after-the-optimizationproblem/110243/21 "2024-03-05T16:32:28Z")

</div>

@whucsu there are two things off in your script that make it not work. The way to diagnose it was run the `cost_function(p)` which gives `Inf` in your current script.  
I am pasting the lines that were changed to make it work

```julia
data = convert(Array, randomized)[[1,4], :]
cost_function = build_loss_objective(prob, Rodas5P(autodiff = false), L2Loss(t, data),
                                    Optimization.AutoForwardDiff(),save_idxs = [1,4])

```

Basically, since you want to use `save_idxs` you should only pass those states’ data. Also the `Rodas5P` didn’t have the `autodiff = false` in the `cost_function` definition.

[Next page](https://discourse.julialang.org/t/why-do-i-get-the-same-parameter-after-the-optimizationproblem/110243.md?page=2)