No more memory available in Mathematica, Fit the parameters of system of differential equation
Posted
by
user1058051
on Stack Overflow
See other posts from Stack Overflow
or by user1058051
Published on 2011-11-29T13:35:39Z
Indexed on
2011/12/01
9:51 UTC
Read the original article
Hit count: 185
I encountered a memory problem in Mathematica, when I tried to process my experimental data. It's a system of two differential equations and I need to find most suitable parameters. Unfortunately I am not a Pro in Mathematica, so the program used a lot of memory, when the parameter epsilon is more than 0.4. When it less than 0.4, the program work properly. The command 'historylength = 0' and attempts to reduce the Accuracy Goal
and WorkingPrecision
didn`t help. I can't use ' clear Cache ', because there isnt a circle. I'm trying to understand what mistakes I made, and how I may limit the memory usage. I have already bought extra-RAM, now its 4GB, and now I haven't free memory-slots in motherboard
Remove["Global`*"];
T=13200;
L = 0.085;
e = 0.41;
v = 0.000557197;
q = 0.1618;
C0 = 0.0256;
R = 0.00075;
data = {{L,600,0.141124587},{L,1200,0.254134509},{L,1800,0.342888644},
{L,2400,0.424476295},{L,3600,0.562844542},{L,4800,0.657111356},
{L,6000,0.75137817},{L,7200,0.815876516},{L,8430,0.879823594},
{L,9000,0.900771775},{L,13200,1}};
model[(De_)?NumberQ, (Kf_)?NumberQ, (Y_)?NumberQ] :=
model[De, Kf, Y] = yeld /.Last[Last[
NDSolve[{
v (Ci^(1,0))[z,t]+(Ci^(0,1))[z,t]==
-((3 (1-e) Kf (Ci[z,t]-C0))/
(R e (1-(R Kf (1-R/r[z,t]))/De))),
(r^(0,1))[z,t]==
(R^2 Kf (Ci[z,t]-C0))/
(q r[z,t]^2 (1-(R Kf (1-R/r[z,t]))/De)),
(yeld^(0,1))[z,t]==
Y*(v e Ci[z,t])/(L q (1-e)),
r[z,0]==R,
Ci[z,0]==0,
Ci[0,t]==0,
yeld[z,0]==0},
{r[z,t],Ci[z,t],yeld},{z,0,L},{t,0,T}]]]
fit = FindFit[data,
{model[De, Kf, Y][z, t],
{Y > 0.97, Y < 1.03, Kf > 10^-6, Kf < 10^-4, De > 10^-13, De < 10^-9}},
{{De,7*10^-13}, {Kf, 10^-5}, {Y, 1}},
{z, t}, Method -> NMinimize]
data = {{600,0.141124587},{1200,0.254134509},{1800,0.342888644},
{2400,0.424476295},{3600,0.562844542},{4800,0.657111356},
{6000,0.75137817},{7200,0.815876516},{8430,0.879823594},
{9000,0.900771775},{13200,1}};
YYY = model[
De /. fit[[1]],
Kf /. fit[[2]],
Y /. fit[[3]]];
Show[Plot[Evaluate[YYY[L,t]],{t,0,T},PlotRange->All],
ListPlot[data,PlotStyle->Directive[PointSize[Medium],Red]]]
the link on the .nb file
© Stack Overflow or respective owner