search

gloewing / photometry_FLMM

Code for Functional Mixed Models for Fiber Photometry
7 stars 0 forks source link

Error for function 'as.matrix': element (1, 1) is zero, so the inverse cannot be computed #2

Closed KathaSchmack closed 3 weeks ago

KathaSchmack commented 1 month ago

I have been playing around with the toolbox, which is very easy to use and looks very powerful! Sometimes I get this error:

mod1 = fui(photometry ~ outcome + (1 | sessionId), data = data, parallel = TRUE) [1] "Step 1: Fit Massively Univariate Mixed Models" [1] "Finished fitting univariate models." [1] "Step 2: Smoothing" [1] "Step 3: Inference (Analytic)" [1] "Step 3.1: Preparation" [1] "Step 3.1.1: Method of Moments Covariance Estimator Random Intercept" [1] "Step 3.1.2: Smooth G" [1] "Step 3.2: First step" Error in value[3L] : Error when parallelizing variance calculations: Error in checkForRemoteErrors(val): one node produced an error: error in evaluating the argument 'x' in selecting a method for function 'as.matrix': element (1, 1) is zero, so the inverse cannot be computed

The data does not contain any nans, and the model runs fine for a similar analysis on photometry data recorded at the same time in a parallel channel.

gloewing commented 1 month ago

Does the same error occur if parallel=FALSE?

On Sun, Sep 29, 2024 at 7:35 AM Katharina Schmack @.***> wrote:

I have been playing around with the toolbox, which is very easy to use and looks very powerful! Sometimes I get this error:

mod1 = fui(photometry ~ outcome + (1 | sessionId), data = data, parallel = TRUE) [1] "Step 1: Fit Massively Univariate Mixed Models" [1] "Finished fitting univariate models." [1] "Step 2: Smoothing" [1] "Step 3: Inference (Analytic)" [1] "Step 3.1: Preparation" [1] "Step 3.1.1: Method of Moments Covariance Estimator Random Intercept" [1] "Step 3.1.2: Smooth G" [1] "Step 3.2: First step" Error in value[3L] http://cond : Error when parallelizing variance calculations: Error in checkForRemoteErrors(val): one node produced an error: error in evaluating the argument 'x' in selecting a method for function 'as.matrix': element (1, 1) is zero, so the inverse cannot be computed

The data does not contain any nans, and the model runs fine for a similar analysis on photometry data recorded at the same time in a parallel channel.

— Reply to this email directly, view it on GitHub https://github.com/gloewing/photometry_FLMM/issues/2, or unsubscribe https://github.com/notifications/unsubscribe-auth/AJVLMQOBWJZUYGZHUCM3KW3ZY7QZZAVCNFSM6AAAAABPBUFBM2VHI2DSMVQWIX3LMV43ASLTON2WKOZSGU2TIOBYG42DIMI . You are receiving this because you are subscribed to this thread.Message ID: @.***>

KathaSchmack commented 1 month ago

Yes, same error I am afraid:

Error in h(simpleError(msg, call)) : error in evaluating the argument 'x' in selecting a method for function 'as.matrix': element (1, 1) is zero, so the inverse cannot be computed

gloewing commented 1 month ago

Hmmm, that’s funny. If you send me a dataset that can reproduce the error I can try to troubleshoot it.

On Sun, Sep 29, 2024 at 2:18 PM Katharina Schmack @.***> wrote:

Yes, same error I am afraid:

Error in h(simpleError(msg, call)) : error in evaluating the argument 'x' in selecting a method for function 'as.matrix': element (1, 1) is zero, so the inverse cannot be computed

— Reply to this email directly, view it on GitHub https://github.com/gloewing/photometry_FLMM/issues/2#issuecomment-2381450872, or unsubscribe https://github.com/notifications/unsubscribe-auth/AJVLMQJMW5RXH73RO24W2GLZZA76FAVCNFSM6AAAAABPBUFBM2VHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMZDGOBRGQ2TAOBXGI . You are receiving this because you commented.Message ID: @.***>

gloewing commented 3 weeks ago

As we discussed via email, this issue came from a non invertible matrix for one point on the functional domain (a trial timepoint at the very end of the trial). Removing this trial timepoint from the analysis resolved this issue.