1

Do I have to clear all my definitions whenever I experiment with different parameter values for my model? For example, I have a simple model as follows:

w[t_]:=w[t]=w[t-1]+h
w[0]=0    
h = 5

I want to do:

Table[w[t], {t, 0, 10}]

with varying h. But every time I change h, I have to clear the definitions with 

Clear["Global`*"]

Otherwise, the result for w[t] remains the same. This practice is ok with this simple model. But when the model is pretty big, it really is time-consuming. I was expecting to get a new result directly by changing h. Any helpful comments in this direction?

m_goldberg
  • 107,779
  • 16
  • 103
  • 257
ppp
  • 817
  • 8
  • 15
  • 1
    Could you use w[t_,h_] instead of setting h afterward ? – William Briand Mar 13 '14 at 20:20
  • @WilliamBriand, thanks. I can try. But my actual model is pretty big with so many parameters. So it would be impossible to put all of them into the bracket. – ppp Mar 13 '14 at 20:31
  • @Kuba, thanks. I think that's what I did. You can see h gets value at the end. – ppp Mar 13 '14 at 21:17
  • Enclosing things in Dynamic allow for post hoc update of variables -- though it may make it considerably slower to run. – William Briand Mar 13 '14 at 21:26

1 Answers1

1

You could just clear the DownValues you don't want.

w[t_] := w[t] = w[t - 1] + h
w[0] = 0;

Table[DownValues[w] = Drop[DownValues[w], {2, -2}]; Table[w[t], {t, 0, 3}], {h, 2, 5}]

{{0, 2, 4, 6}, {0, 3, 6, 9}, {0, 4, 8, 12}, {0, 5, 10, 15}}
m_goldberg
  • 107,779
  • 16
  • 103
  • 257
  • Thank you so much, m_goldberg. But it seems quite complicated. In fact, I was moving over from R. With R, you just change h directly and you will have new results for w[t]. I was expecting something similar to this. It seems R is much more convenience than Mathematica for experimenting with a dynamic model with many variables and parameters. Is it true? – ppp Mar 13 '14 at 21:47