IndexError: Index out of bound

[rubinanoor9]

Dear PyEMMA users!

I am following the tutorial " http://www.emma-project.org/latest/legacy-notebooks/applications/pentapeptide_msm/pentapeptide_msm.html" in my own project and following the same strategy as mentioned in it but during analysis I got the error mentioned here: "Traceback (most recent call last): File "", line 1, in File "", line 1, in File "", line 1, in IndexError: index 192 is out of bounds for axis 0 with size 173" by using this "proj_ev_all = [np.hstack([M.eigenvectors_right()[:,i][dtraj] for dtraj in M.discrete_trajectories_full]) for i in range(1, 10)]" command line. I am looking forward to here from you. Thanking You, Regards Rubina

[clonker]

Hi Rubina, it would be good if you could give us a bit more information. How many discrete trajectories do you have? How many Markov states do you use? Another thing to look out for (and I think the source of your issue) is the active_set. You probably don't have a full active set, so you'll have to exclude the frames in dtraj that are not contained before using them with the eigenvectors. Best Moritz

[rubinanoor9]

Thank you so much for your reply. I have applied different number of clusters (50, 75, 100, 250, 300) but I got similar index error but with different index number, With the 100 number of cluster, I got error; IndexError: index 99 is out of bounds for axis 0 with size 98" and when I am using cluster number=50, I got; IndexError: index 49 is out of bounds for axis 0 with size 48" I am attaching an input script. Thanks in advance. Regards Rubina input-script.odt

[clonker]

This is indeed an issue with the discrete trajectories. You have to map them to the active set before applying them to the eigenvector components. so instead of taking M.discrete_trajectories_full, you take

dtrajs_active = M.discrete_trajectories_active
dtrajs_active_reduced = [dtraj[dtraj != -1] for dtraj in dtrajs_active]
proj_ev_all = [np.hstack([M.eigenvectors_right()[:,i][dtraj] for dtraj in dtrajs_active_reduced]) for i in range(1, min(10, M.eigenvectors_right().shape[1]))]

[rubinanoor9]

Thank you so much for the help. I have applied this command line as you mentioned, now it did not get error but when I have tried to plot them by using the following command line: fig, axes = plt.subplots(1, 3, figsize=(16,4)) for i, ax in enumerate(axes): plot_sampled_function(xall, yall, proj_ev_all[i], ax=ax, cbar=False, cmap=plt.cm.Blues) plot_labels(ax)

I got error, which is:

IndexError Traceback (most recent call last) /tmp/ipykernel_29983/3979245165.py in 1 fig, axes = plt.subplots(1, 3, figsize=(16,4)) 2 for i, ax in enumerate(axes): ----> 3 plot_sampled_function(xall, yall, proj_ev_all[i], ax=ax, cbar=False, cmap=plt.cm.Blues) 4 plot_labels(ax)

/tmp/ipykernel_29983/3546841851.py in plot_sampled_function(xall, yall, zall, ax, nbins, nlevels, cmap, cbar, cbar_label) 18 # average over bins 19 for t in range(len(xall)): ---> 20 z[xI[t], yI[t]] += zall[t] 21 N[xI[t], yI[t]] += 1.0 22 with warnings.catch_warnings() as cm:

IndexError: index 164687 is out of bounds for axis 0 with size 164687

I am also new user of python, so not be able to solve this issue. Thanks in advance.

[thempel]

I believe that you are trying to plot the reduced projected eigenvectors (which are reduced to the active set states) as a function of the un-reduced x and y coordinates xall and yall. That means that your x and y-coordinates have more points than the z-coordinate. Therefore, you got to reduce the x-y coordinates in the same way, e.g. by

xall_reduced = xall[np.concatenate(dtrajs_active) != -1]
yall_reduced = yall[np.concatenate(dtrajs_active) != -1]

Then, call the plot function plot_sampled_function by replacing xall -> xall_reduced and yall -> yall_reduced. (Disclaimer: I did not test this code.)

A note of attention: These reduced trajectories should be used for plotting only. The dynamics is perturbed by removing inactive states, so one should not use them for estimating MSMs or the like.

[rubinanoor9]

Thank you so much. Now the problem has been resolved based on your suggestion. Thanks again.

[clonker]

You're welcome! Closing this now.